The Future of Mobile Wireless Technology

Document technical information

Format pdf
Size 1.3 MB
First found May 22, 2018

Document content analysis

Category Also themed
not defined
no text concepts found


Ernest Hemingway
Ernest Hemingway

wikipedia, lookup

Dizzy Gillespie
Dizzy Gillespie

wikipedia, lookup




© SHG 2001
The Role of Pervasive
Low-Cost Networks and Devices
A Sag Harbor Group White Paper
February 2002
Copy #: _______
©SHG 2001. Confidential Draft. Not for distribution or reproduction without express consent from SagHarbor Group. The
opinions expressed here are those of Sag Harbor Group.
© SHG 2002
© SHG 2002
-1 -
© SHG 2002
Confidentiality and Other Preliminaries
The information contained in this memorandum with respect to wireless messaging technology is being made
available on a confidential, copyrighted basis by Sag Harbor Group. This memorandum should be read in
conjunction with any additional information that has been included in the Appendix. The information
contained herein has been obtained from sources that SHG believes to be reliable. However, SHG makes no
representations or warranties expressed or implied as to its accuracy or completeness.All views expressed are
those of Sag Harbor Group.
631-725-7994 (FAX)
[email protected]
© SHG 2002
© SHG 2002
“The killer application for wireless data… mobile messaging.”
-- Mohsen Banan, “The WAP Trap,” May 2000
“It is simplicity that is difficulty to make.”
-- Bertoldt Brecht
© SHG 2002
© SHG 2002
Table of Contents
Executive Summary .................................................................................................................6
Introduction – So What Became of the “Mobile Wireless Revolution”? .................................9
1. Why Yet Another Wireless White Paper? ........................................................................................................................... 10
2. Great Expectations.............................................................................................................................................................. 10
3. Return to Earth................................................................................................................................................................... 12
DoCoMo’s Lessons ................................................................................................................ 17
1. Introduction - Japan’s Exceptionalism ................................................................................................................................. 18
2. Japan’s Mobile Internet Takeoff ......................................................................................................................................... 18
3. Key Market Conditions for i-mode’s Success......................................................................................................................... 19
4. The Rise of SMS Messaging............................................................................................................................................... 22
5. NTT DoCoMo’s Key Strategic Choices............................................................................................................................... 28
6. Beyond i-mode ! 3G’s Inevitability? .................................................................................................................................. 32
7. Implications for 3G’s Future Elsewhere ............................................................................................................................... 33
8. What Do Customers Really Want?..................................................................................................................................... 38
9. Key Implications, DoCoMo’s Experiment .......................................................................................................................... 40
10. Conclusion – DoCoMo’s Lessons ...................................................................................................................................... 42
ReFLEX™’s Technology – Origins, Key Attributes, and Direction ..................................... 43
1. Introduction – Key Low-Speed Mobile Data Networks ....................................................................................................... 44
2. ReFLEX’s Origins ........................................................................................................................................................ 46
3. The First Launch – Mistakes and Roadblocks ................................................................................................................... 47
4. ReFLEX’s Original Technical Attributes ...................................................................................................................... 49
5. The Importance of Version 2.7, WCTP, and Other Developments .................................................................................. 52
6. Summary – The Technical Foundations of ReFLEX’s Revival ...................................................................................... 57
ReFLEX’s Competitive Advantages ...................................................................................... 58
1. Introduction -- ReFLEX’s Key Competitors .................................................................................................................... 59
2. Mobitex™’s Origins, Decline and Revival ......................................................................................................................... 60
3. Mobitex™’s Strengths and Weaknesses ............................................................................................................................. 62
4. Summary - Mobitex’ Vs. ReFLEX™.......................................................................................................................... 65
5. Applications “Fit” – Valuing Technical Attributes ............................................................................................................ 68
6. The Other Data-Only Network Alternatives ..................................................................................................................... 69
7. 2.5G – A Threat to Everyone Else?................................................................................................................................... 70
8. Summary – ReFLEX’s Competitive Advantages ............................................................................................................... 78
9. Conclusion ........................................................................................................................................................................... 80
Appendix A: Key Technical Features, ReFLEX
 Version 2.7............................................. 81
1. Introduction ...................................................................................................................Error! Bookmark not defined.
2. Background Scanning ...................................................................................................Error! Bookmark not defined.
3. Auto-Collapse – “Chat Mode” ......................................................................................Error! Bookmark not defined.
© SHG 2002
© SHG 2002
4. Broadcasting Maximum Inbound Message Lengths .......................................................Error! Bookmark not defined.
5. Unscheduled Inbound Messaging....................................................................................Error! Bookmark not defined.
VII. Appendix B: Detailed Comparisons, Key Data Networks..................................................... 86
1. Network Comparison Charts .............................................................................................................................................. 87
2. Devices Comparison, ReFLEX™, Mobitex™ and DataTAC™, 2001........................................................................ 93
3. Network Cost Comparisons .............................................................................................................................................. 94
4. Applications Fit – Technology Requirements per Application Type (Illustrative) ................................................................. 95
VIII. Appendix C: Glossary of Technical Terms ........................................................................... 96
© SHG 2002
© SHG 2002
© SHG 2002
Executive Summary
© SHG 2002
Executive Summary
This study examines the future of mobile data messaging (MDM) in general, and the
comparative advantages of the ReFLEX™, Mobitex™, 2.5G cellular, and 3G cellular families
of two-way MDM technologies in particular.
This is a good time to pause and catch our breaths in the worldwide mobile Internet
revolution, and pay close attention to some key messages that this so-called “revolution”
has been trying to send us. Among the most important are the following:
1. What customers – especially enterprise customers -- really want from mobile data is
easy to describe: (1) reliable (2) low cost (3) easy-to-use (4) secure (5) pervasive (6)
interoperable MDM.
2. What it is not at all clear that they want is what the global cellular industry has recently
been desperately trying to sell them – costly upgrades to complex new handsets, the
ability to watch video and surf at lighting speed on cell phones and PDAs, and the poor
service quality and costly network expansions that have too often been associated with
realizing this vision.
3. MDM has the chameleon-like property of being disguised by local market conditions. In
Japan it takes the form of “i-mode;” in Europe, of person-to-person SMS; in the US, of
two-way wireless data-only networks like ReFLEX, Mobitex, DataTAC™ and CDPD.
Brought to light and aggregated, however the global evidence clearly shows that if there
is one killer application for mobile wireless data, it is messaging – email, chat, and
information broadcasting – not Web surfing, shopping, or mobile multimedia. (Chapter
4. Unfortunately, if the global cellular industry has its way, what the world might get
would be a very expensive two-step upgrade of all existing cellular networks and
handsets – first to so-called “2.5G” in the next 2-3 years, and then to even higher-fixed
cost 3G networks. Fortunately, under the strain of current economic conditions, this
“vision” is now receiving much more critical scrutiny, especially in the US.
5. While 2.5G networks are supposed to offer major improvements for wireless data users
over existing circuit-switched data services – including “always on” capability, higher
data rates, and lower usage costs – on closer inspection, most of these advantages turn
out to be highly questionable, especially for enterprise MDM applications. (Chapter V.)
3G networks, if they are ever built at all, may turn out to be even more dubious, the
“HDTV” of wireless networking. They are a costly, ill-conceived kluge of the mobile
Videophone, the PocketPC, pay-per-view, and MP3, with little to offer to enterprise
customers who just want affordable, reliable, and ubiquitous MDM applications now.
(Chapter III.)
7. In general, our analysis of low-speed networks, i-mode, SMS, and 2.5G and 3G
networks leads us to be deeply skeptical about what has really become the central value
© SHG 2002
© SHG 2002
proposition behind the $350 billion+ 2.5 G and 3G cellular network and handset
upgrades now going on around the world. For almost all mission-critical enterprise
MDM applications that we can think of, these upgrades will provide virtually no
discernable improvements in application performance. Indeed, to the extent that
enterprises are seduced to adopt the data solutions offered by the cellular voice industry,
actual MDM application performance is likely to suffer, even while the total costs of
application ownership soar.
8. Relative to new “2.5 G” network technologies like GPRS and CDMA2000, for the
foreseeable future, low speed networks like Mobitex™ and ReFLEX is also likely to
offer several key advantages for “mission-critical” MDM applications, including
1. Much more reliable messaging, based on superior coverage and in-building
2. Interoperable networks, including support for user-focused, device- and
network-agnostic data services;
3. Support for much lower-cost, more flexible devices;
4. Much better support for specific applications, including information
broadcasting, store-and-forward messaging, and low-cost chat;
Tremendous nation-wide network capacity, supporting very competitive
service costs for messaging, and much lower total costs of ownership,
especially for enterprise applications. (Chapter V.)
Overall, therefore, we are optimistic about the future of low-speed MDM networks
in general – assuming that leading service operators can solve their pressing
business problems. Assuming that the industry can restructure, this network’s reliable,
low-cost services should be around for a very long time to come. As these service providers
proceed to roll out new devices and other capabilities over the next few months, we believe
that such networks deserve a close, comparative look from enterprise customers, solutions
providers, and investors.
© SHG 2002
© SHG 2002
© SHG 2002
Introduction – So What Became of
the “Mobile Wireless Revolution”?
© SHG 2002
1. Why Yet Another Wireless White Paper?
he wireless industry is notorious for burying its hapless customers in a blizzard of
conflicting claims and shifting predictions about the capabilities of rival networks,
devices, application development platforms, and protocols. Just when it seems that
one storm has lifted, another blinding flurry arrives and the trail disappears again.
In the midst of all this confusion, it is important to keep an eye on fundamentals. This white
paper starts from the little-regarded fact that, at least in the US market, two-way wireless data
networks that are based on the ReFLEX™ and Mobitex™ platforms now account for far more
wireless data messaging and subscribers than all other data networks combined.
Furthermore, as discussed below, both these low-speed networks are now on the brink of
new network upgrades. Their US service providers are also about to offer a new generation
of messaging devices with outstanding designs and competitive price/performance. Finally,
on the applications development front, these networks are finding it easier and easier to
deliver more powerful, highly-scalable wireless solutions, especially to enterprise customers..
All told, even as 2.5G and 3G cellular technologies begin to roll out this year, we will also be
seeing dramatic improvements in low-speed platforms, conslidating their positions as the
low-cost, reliable, easy-to-design and-deploy, pervasive platforms of choice for enterprise
mobile data messaging (MDM) applications. The rest of this white paper is devoted to
telling this story, which we regard as one of the (unintentionally) best-kept secrets in the
wireless industry today.
2. Great Expectations
At the outset, it will be useful to examine what became of the so-called “mobile wireless
revolution” that was so widely predicted just a short while ago, to see what can be learned
from the way things actually turned out.
The global wireless industry started this decade with extraordinary expectations. As late as
Fall 2000, a wide variety of industry observers were still trumpeting the notion that mobile
wireless technology would soon bring the benefits of Internet access and mobile messaging
to hundreds of millions of users around the globe. The expectation was that this might
easily dwarf the PC Revolution of the 1980s and even the Internet Revolution of the late
These great expectations were partly based on the “tulip craze” mentality that prevailed in
global capital markets in the late 1990s. But they were also based on assumptions about the
wireless industry that have since turned out to be wildly optimistic. For example:
Greatly Improved Cellular Data Networks and Devices. Industry observers were
bullish about the pace of technical progress in wireless networks and devices. They expected
© SHG 2002
- 10
© SHG 2002
that the so-called 2.5 and 3G cellular data technologies would arrive quickly, delivering
significant advances in bandwidth, capacity, and reliability. Conversely, traditional lowspeed, “data-only” network technologies like ReFLEX™, Datatac™, CDPD, and
Mobitex™ were viewed as mature technologies that were headed for history’s dust bin.
Investments in Spectrum and Network Capacity. Many observers also expected that
wireless network operators would easily raise the hundreds of billions of capital required to
purchase spectrum for these new networks and build them out.1
WAP Forum Members, 1999
Improved Wireless Software.
Industry analysts also contemplated
the widespread adoption of new
mobile wireless standards like WAP
and Bluetooth,
new wireless
mobile--friendly operating systems
like Sun Microsystem’s J2ME,
Qualcomm’s BREW, the Palm OS,
Microsoft’s WinCE and Stinger,
and Symbian’s EPOC. All these
capabilities would be supported by
an increasingly global community of wireless solutions providers.
Innovative Wireless Data Applications / Solutions. It was also expected that all these
new wireless hardware, software, and application capabilities would quickly lead to a rich
assortment of applications and solutions, in both the consumer and enterprise segments of
the market. The most
exciting applications fall
Chart 1. Cellular Voice Base by Region (MMs of Users)
into two categories – (1)
corporate e:mail, corporate
communications (including
MM Subs 400
Africa/M E
applications like sales force
Other Asia
automation, field service
E Europe
W Europe
S Amerca
applications), and locationN America
based “m:commerce;” and
management, and meter
Other Asia
E Europe
W Europe
S Amerca
N America
S ource: Global Mobil S ubscriber DB (2000)
© SHG 2002
- 11
© SHG 2002
Pervasive Adoption. Finally, the expectation was that all these new solutions and
services, combined with lower device costs and faster networks, would finally lead to a
global takeoff for wireless data, a natural sequel to the rapid growth of voice-based cellular
services. 2 (See Charts 1 and 2. ).
Hundreds of millions of new subscribers would supposedly sign up for mobile wireless
Internet connections as early as 2002. (For just one example of many such forecasts, see
Chart 3.) 3 This would dwarf the number of PCs used to provide Internet connectivity,
bring applications like email and browsing to vast new audiences, and change the balance of
forces in the cellular, PC software, hardware, and network equipment.
3. Return to Earth
rom the standpoint of all
these great expectations,
the last year has been a
sobering experience. For example:
C h a rt 2 . US C e ll Ph o n e s v s O ne -P a g in g S u b s c rib e rs ,
1 9 8 4 -2 0 0 1
Missing in Action – New
Networks and Devices That
Work. In Europe, Japan and the
US, the deployment of new 2.5G
and 3G networks have been
seriously delayed. In the US, 3G is
even farther behind, because
adequate spectrum for it has not yet been identified, much less licensed. There have also
been serious delays on the 2.5G and 3G device side. Meanwhile, several mobile data
startups, like Metricom’s Richochet,
tried to deploy new high-speed
Chart 3. Mobile Internet Users by Region,
networks, and simply ran out of
money before they found enough
C e ll P h o n
M M S ubs
P a g e rs
ME/A frica
Latin Amer
MMs of Subscribers
Network Capacity Slowdown.
Except for a handful of special cases
like Korea, Finland and Japan, the
pace of investment in new high100
speed networks by cellular service
substantially. This is partly because
European cellular operators, among
the earliest adopters of 3G technology, got involved in spectrum bidding wars that cost
more than $125 billion. And they will still have to spend another $100 billion in order to
Actual 82 mm vs. 115 mm forecast
Other Asia
N. A mer.
Source: Yankee Group (September 2000)
© SHG 2002
- 12
© SHG 2002
build out these new networks, and the additional costs required for the development of
applications and content. Especially in the US, this capacity expansion problem has been
further aggravated by the spectrum issues noted above, and by the fact that network vendors
have not been able to agree on a standard upgrade path to 3G networks.5
Conflicting Protocols, Disappointing Software. Meanwhile, the hoped-for
convergence of wireless application developers around a common set of standards,
Two examples – Mobile Wireless Irrational Exuberance, Circa Early 2000
middleware, and operating systems is also missing. The WAP Forum, in particular, turned
out to be a fiasco,6 a thinly-veiled attempt by its some of its founders to generate royalties
and content tolls. WAP’s Release 1.0 yielded applications that were painfully slow, hard to
develop or use, expensive, and “seldom on.”7 Not surprisingly, as one study of US
corporate users of WAP phones recently reported, 85-90 percent of them quickly abandoned
the phones’ Internet and data messaging capabilities entirely.8 The Forum’s “walled garden”
approach to content hosting -- with carriers charging content providers hefty fees in order
to get access to their subscribers – also discouraged application development.9 For the
subscriber, the result is a kind of applications ghost town, where the few applications that
actually work would probably work better by way of voice calls !
Overall, as summarized in Table 1, the WAP Forum’s approach could scarcely have been
more un-Web-like.
© SHG 2002
- 13
© SHG 2002
Table 1. Web vs. WAP Strategies to Internet Service Development
Web Approach
WAP Approach
" Low marginal costs of use – often flat rates,
" Per minute pricing -- $.20 per minute or
declining capacity costs
" Standard, interoperable open systems (3WC
" Closed/ proprietary protocol elements
" Relatively easy to use
" Hard to use (screens, keyboards)
" Easy to access (dial up everywhere;
" Hard to access (carrier coverage limitations;
interoperable networks)
non-interoperable networks)
" Easy to develop for (low cost/ standard
" Hard to develop (need to learn WML)
application languages – HTML, XML, etc.)
" Open access -- content providers
" Closed “garden,” with access tolls and
content rent
Source: SHG interviews and analysis
© SHG 2001
Meanwhile, the more robust micro-operating
systems that are designed to provide local
processing power on mobile devices, like
Sun’s J2ME, have been slow to enter the
market, partly just because they have been
waiting for new devices and networks. And
there has been little agreement on
“middleware” standards, either. At least count
there were more than 30 rival wireless
middleware vendors.10
Chart 4. Forecast Changes, Key Global Wireless Metrics (MM, 2004)
Cahners - WW
Mobile Internet
Yankee Group -- WW IDC - WWMobile
Mobile Internet
Cahners - Bluetooth
Fall 2000 forecasts Spring 2001 forecasts
Prosaic Solutions, Sluggish Adoption.
Given these constraints on networks, devices, and software, it is not really surprising that in
most major markets other than Japan, the actual adoption of wireless Internet access has
been modest -- less than 3
million users in the US and 5.9
Chart 5. Revised 2001 Mobile Handset Sales Forecasts
million in Europe by the end of
(MM Global Units)
2000, and only 25 million by the
end of 2001. This compares
with more than 29 million users
very active Web phone users –
almost all non-WAP -- in Japan
by yearend 2000. Furthermore,
most of these so-called “users”
rarely use their Web phones to
access the Internet or do
Source: Industry forecasts, SHG analysis
© SHG 2002
- 14
© SHG 2002
email.11 Outside Japan, the volume of “m-commerce,”12 location-based services,13 and
mobile advertising services is tiny, and the number of truly exciting wireless data applications
deployed so far by enterprise customers is also trivial.
Accordingly, as shown in Charts 4 and 5, the latest estimates by leading industry
forecasters for key indicators like the adoption of mobile Internet services, Bluetoothenabled devices, and mobile handset sales are well below the projections that were made
only just last Fall. Of course wireless industry analysts have also been known to make
incredible underestimates of industry.14 But the uniformity of these recent overestimates is
Not surprisingly, given this underperformance, market valuations for the global cellular
industry as a whole have suffered sharp declines this year. (Charts 6, 7, and 8). The gloomy
trends were especially hurtful to “pure play” mobile wireless data companies and solutions
providers,15 as well as private-equity valuations in the wireless arena.16
C h a rt 6 . Up tu rn s a n d D o wn tu rn s , G lo b a l W ire le s s M a rk e t
In d ic e s b y K e y R e g io n , 1 9 9 9 -2 0 0 1 (J a n . 1 9 9 9 = 1 0 0 )
Sp rin g 9 9
Sp rin g 0 0
Sp rin g 0 1
Eu ro p e
Ja p a n
Overall, it would be easy to conclude – as many analysts and investors already have – that
the outlook for wireless data is just plain bleak. However, we will argue here that this, too, is
an overreaction, and that all these clouds do have some silver linings, if we look closely
Chart 8. Stock Price Changes, Leading Cellular Wireless
Companies, 2000-1
Chart 7. Stock Price Changes, Leading “Pure Play” W ireless
Data Companies, Sept. 2000-June 2001
S&P 500
S&P 500
Sprint PCS
Source:, SHG analysis
© SHG 2002
Source:, SHG analysis
- 15
© SHG 2002
To sharpen our vision, we’ll begin with what might seem to be a complete detour – a visit
to Japan, a market that has (1) more Web phones than anywhere else, (2) almost no lowspeed, two-way, data-only networks of any kind; and (3) investment in 3G networks already
under way.
If the case for low-speed data networks can stand up to this combination of circumstances,
it can probably survive anywhere.
© SHG 2002
- 16
© SHG 2002
© SHG 2002
DoCoMo’s Lessons
- 17
© SHG 2002
1. Introduction - Japan’s Exceptionalism
t is important for us to understand Japan’s recent experience with mobile data
messaging in the last few years, because it has been by far the most important exception
to the negative patterns described above. Surprisingly, despite Japan’s overall economic
malaise, it has in fact been the only real success story that champions of cellular data can
point to, when they argue for the inevitable triumph of 2.5 and 3G technology over lowerspeed networks. This chapter takes a closer look at what has really been going on with
mobile wireless in Japan, and argues that, beneath the surface, it actually supports the case
for the long-term viability of lower-speed MDM networks. 17
2. Japan’s Mobile Internet Takeoff
Japan’s success to date with the mobile Internet has indeed been dramatic. From a standing
start in February 1999, by July 2001 there were more than 40.3 million Japanese subscribers
– 31 percent of the country’s entire adult population – who were using Web-enabled cell
phones to send messages and access the Internet on the fly.
This growth was spearheaded by NTT DoCoMo, which now commands about 62 percent
of Japan’s mobile Internet market. By yearend 2001 DoCoMo will have at least 32 million
mobile Internet subscribers, and Japan as a whole will have more than 56 million, 18 twice the
number in North America and Europe combined, and two-thirds of the world’s total. (See
Charts 9 and 10.)
This year DoCoMo19 – more formally, the “NTT Mobile Communications Network,”
which is two-thirds owned by NTT -- will generate more than $2.8 billion of revenue from
its “i-mode” mobile Internet service. This makes it NTT’s most profitable business unit,
accounting for more than 100 percent of NTT’s net profit. 20 In less than three years DoComo
has transformed itself into the world’s largest and most innovative mobile data service
provider. 21 This is quite an achievement for a state-owned company that started out in 1959
as a maritime radio services provider, moved on to one-way paging in the 1960s and bulky
executive car phones in the 1970s, and as recent as 1995, officially expressed doubts that
more than 10 million cell phones would ever be sold in Japan. It is also quite an
achievement for a country whose capacity for growth and innovation have recently been
C ha rt 9 . G ro wth o f M o b ile In tern e t Su b s crib e rs in J ap a n ,
1 9 99 -2 0 01
M M Subscribers
© SHG 2002
O ct-99
NTT DoK oM o (I-M ode)
K DDI (E z W eb)
O ct-00
J-S k y
- 18
© SHG 2002
widely questioned.
C h a rt 1 0 . W e b P h o n e Us ers : US , E u ro p e, J a p a n , 20 0 0-1
(M M Yea re n d )
20 .9
5.9 4
2.8 4
55 .98
29 .24
20 00
20 01e
Sourc e: Morgan Stanley (June 2001); SHG analy s is
E urope
3. Key Market Conditions for i-mode’s Success
mong the key factors responsible for Japan’s exceptional adoption of Web phone
services are the following:
Cell Phone Penetration. To begin with, in the last decade Japan quickly attained high
cell phone penetration -- 77 percent of households, compared with less than half of US
households.22 In May 2000 the number of mobile phones in Japan actually surpassed the
number of wire line phones. 23 By 2004 cell phone penetration is expected to reach 95
This high cell phone penetration rate has been driven by several special market conditions,
including the relatively high cost of wire-line telephone services in dense urban areas,24
NTT’s continued monopoly over wired local access, the use of metered billing for wired
phone calls, the prevalence of public transportation and long commuting times, 25 the early
adoption of “calling-party-pays” rules for pricing, and most important, the existence of
high-quality, densely-sited national cellular networks that offer a very reliable alternative to
wired service. Since most new cell phones that have shipped since 2000 have been Internetenabled, this automatically provided a strong foundation for the mobile Internet.
Ironically, in the early 1980s Japan had been known for its low cell phone penetration. The
very same NTT that is now the hero of our story had systematically overpriced analog
phones, even requiring customers to lease them! In December 1988, Japan’s Ministry of
Posts finally understood that NTT’s monopoly over wireline services created a conflict of
interest with services, and ended its mobile wireless monopoly. The cell phone market really
took off after April 1994, when pricing and services were more fully deregulated.
© SHG 2002
- 19
© SHG 2002
PC, Internet, and Broadband Penetration. Another key factor behind Japan’s recent
mobile Internet surge is its relative backwardness in PC and wired Internet use. Partly just
because the Japanese language has a difficult time with Qwerty keyboards, and double-byte
compatibility came late to US operating systems, PC penetration in Japan has always lagged
US levels. As of yearend 2000, just 38 percent of Japan’s 46 million households had PCs,26
compared with 63 percent of US households.27 US Internet access from wired home PCs is,
accordingly, also sharply higher. 28 Only 19 percent of Japanese households have wired PC
connections, 29 compared with 57 percent of US households.30 Americans also use their PC
Internet connections much more intensively than Japanese or Europeans.31 This also reflects
the fact that dial-up connections in Japan are relatively expensive, compared with the
unmetered dial access available in the US.32 The share of US workers with wired connections
at the office is also higher.33
Finally, compared to the US, broadband Internet access is also relatively scarce in Japan. As
of 2001, just 5 percent of Japan’s Internet users have broadband access34 at home or at
work, compared with 31 percent of US Internet users.35 Indeed, at home, less than 4 percent
of Japanese households now have broadband connections, compared with 14.1 percent in
the US and 3.3 percent in Europe.36 While NTT made a commitment in 1994 to provide
fiber to all Japanese homes by 2010, for at least the next five years this broadband gap is
expected to persist. 37
Given these special conditions, it is not surprising that there are now more than 40 million
Web phones in Japan, twice the number of wired PC connections, and that this number is
expected to double to 75-80 million in the next three years.38 These services already generate
more than $3 billion a year of revenue for Japan’s three leading mobile operators.
Of course we should remember that Internet devices are not the same as Internet use. One
recent analysis of Internet use in Japan found that PCs accounted for more than 93 percent
of all Internet use, while mobile phones accounted for just 3 percent.39 This is a key point for
our analysis, because it turns out that most of what i-modes subscribers are actually doing
with their Web phones is not “browsing” or higher-speed Internet applications, but lowspeed MDM.
Two-Way Data Networks and Wireless Handhelds. Two other factors that help to
explain the rise of Web phones in Japan are the shortage of two-way data-only networks
and the relatively small size of the domestic PDA market.
A. Two-Way Data Networks in Japan
Unlike the US, Japan never built nation-wide data-only networks like ReFLEX.™
Mobitex,™ DataTAC,.™ or CDPD. This was not for want of one-way paging. One-way
paging was first introduced in 1968, and by 1997 there were more than 10 million
subscribers. Nearly sixty percent of them were served by a state-of-the–art national
FLEX™ network that was owned by NTT DoCoMo.40 But one-way paging in Japan stalled
at roughly 8 percent penetration in the mid-1990s,41 half the US rate and well below the 2530 percent levels achieved in other Asian markets like Korea and Hong Kong. This was
partly due to stiff competition from the “personal handyphone,” a popular low-cost mobile
© SHG 2002
- 20
© SHG 2002
phone that first appeared in 1995. Meanwhile, for its cellular voice services, NTT decided
in the mid 1990s to go with its own proprietary standard, a variant of Japan’s native
“Personal Digital Cellular” (PDC) technology. When faced with the question of what to do
about two-way data services, and whether or not to upgrade its FLEX™ network to
ReFLEX™, DoCoMo later decided to go with i-mode, a digital packet-switched service that
ran over PDC at 9.6 kbps. It launched i-mode in February 1999.
As for other two-way data networks, CDPD was not an option, because CDPD was an
upgrade to US-developed AMPS analog cellular networks. One other local DataTAC 5000
network, at 450 MHz, was built in Tokyo in 1997, but the company performed poorly and
was acquired by DoCoMo in 1998. No Mobitex™ network was ever built, partly because
of frequency issues.42 In mid-2000, Glenayre finally sold a ReFLEX™ network to a paging
company in Tokyo for telemetry applications.43
B. PDAs/ Handhelds
Kluge - the price of phone-centric designs for MDM
Given the paucity of low-speed dataonly
networks and low PC
penetration,44 it is not surprising the
market is also small. As of 2001, there
are less than 2 million PDAs in Japan,
none of which are wireless.45 This
compares with about 7 million PDAs
in the US, including a million that are
Recent market surveys indicate that
Japanese users might actually prefer
PDAs over Web phones or even PCs
for purposes of Internet messaging.47
But cellular service providers like NTT
DoCoMo – and handset vendors like
Nokia, Motorola, Ericsson, and Samsung -- have a built-in bias toward devices that combine
voice and data. Responding to the increasing demand for PDA form factors in Japan,
however, Toshiba, NEC, and Sharp recently announced that they will produce new PDAs
with slots for wireless modems.48
© SHG 2002
- 21
© SHG 2002
4. The Rise of SMS Messaging
Another basic condition for i-mode’s takeoff in Japan is that it offered a substitute for the
short-message service (SMS) messaging that has proved so successful in Europe and some
other Asian markets. SMS, yet another low-speed two-way MDM technology, operates in the
signaling/ control channel of circuit-switched cellular networks, and was originally intended
to provide voicemail notification to cell phone users. In the last decade, especially in
C h a rt 1 1 . G lo b a l G ro wth o f S M S M e s s a g in g , 1 9 9 9 -2 00 1
Million Messages per Month
E urope
Rest of World
Europe and Asia, mobile operators have used it to provide a very simple, cheap way to send
short (< 160 characters) text messages.
The results have been phenomenal, at least outside the US. As of July 2001, more than 20
billion SMS message per month were being sent by the world’s 553 million GSM phone
subscribers. Monthly SMS use now averages more than 35 messages per user in Europe,
and up to 240 per user in some Asian markets ! 49 (See Chart 11. ) While SMS’s imminent
demise had been predicted for years, a variety of new SMS technologies and services on the
horizon are likely to permit this growth to continue for some time.50
SMS messaging has been a real boon to cellular operators. Depending on the market, they
receive an average of $.06 per SMS message. In terms of data throughput, this is more than
$5.80 per MB, several hundred times the price per unit of data for a minute of voice traffic ! 51
SMS now accounts for 7-10 percent of operator revenues in Europe,52 and several report
even better results.53 While new SMS services like chat boards, subscriptions to “push” news
services, and handset personalization are growing, the key point is that 98 percent of this
traffic is just plain two-way person-to-person messaging.54
So here we have yet another success story for simple two-way mobile data messaging, and
another striking contrast to the “WAP flop.” It occurred despite SMS’s numerous technical
shortcomings. These include high latency, unreliability,55 limited in-building penetration (by
© SHG 2002
- 22
© SHG 2002
definition, no better than cellular voice service), short message lengths, no email
attachments, no graphics, an inflexible applications development platform, limited content,
and the cell phone’s awkward numerical keypad. In fact, if we were trying to design a
technically-limited messaging platform, it would be hard to design one more limited than
Despite these technical limitations, SMS has exceeded the cellular industry’s wildest dreams,
in terms of sheer numbers of users, traffic volume, and profitability. Again, we believe that
this is for two simple reasons: (1) SMS (outside the US) followed the Web model for service
development, and (2) it gives users what they really want (..all together now..!):
(1) Reliable (2) Low cost (3) Easy-to-use (4) Secure (5) Pervasive (6)Interoperable
Mobile Data Messaging.
A. Why SMS in Europe?
But why did SMS take off in Europe, and not Japan or the US? As in Japan, Europeans
quickly became heavy users of cell phones, partly because landlines were relatively expensive
and population density was high, permitting the efficient build-out of high-quality cellular
Furthermore, beginning in the early 1980s, Europe’s Conference of European Postal and
Telecommunications Administration, and its successor, the European Telecommunications
Standards Institute, promulgated the GSM standard for 2G digital cellular service,
mandating it for all Western European service providers.56 It also provided that customers
would be charged per message sent, not per minute of use, and that “calling party pays”
would also be implemented.57
Contrary to the conventional wisdom, ETSI implemented this in Europe, not out of some
bureaucratic desire to impose uniformity or do economic planning, but because Europe’s
prior experience with analog cellular systems had been sheer chaos. Unlike the US, where the
AMPS standard was adopted by all cellular operators, Europe had ended up with nine rival
analog systems, none of which could talk to each other! 58
In adopting the GSM standard, ETSI not only encouraged cell phone use to take off in
Europe. It also managed to stumble on the Web model. This provided the industry with a
simple, interoperable messaging standard and business model that permitted text messages
to be sent or received cheaply by cellular subscribers across all GSM networks.59 The
adoption of this low cost messaging model, in turn, helped drive the growth of person-toperson SMS messaging in Europe through the roof.
The ironic thing, of course, is that SMS was introduced by many of the very same
regulators, network equipment vendors, and service providers who later played a leadership
role in the WAP Forum. But unlike WAP, SMS was introduced almost as a boring
afterthought, and inadvertently endowed with many of the same features that made the
© SHG 2002
- 23
© SHG 2002
Web so successful – low marginal costs, interoperability, pervasiveness, and even
comparative ease of use (at least to the European teenagers one sees hammering away at cell
phone keypads with pens and other blunt instruments.)
B. Why Only Limited Two-Way Data in Europe?
At the same time, two-way mobile data networks never got much traction in Europe. The
Europeans developed their own standards for digital one-way paging,60 in opposition to
Motorola’s FLEX™ standard,61 which became the de factor digital one-way standard
elsewhere by the mid-1990s. Since there was no FLEX™ base in Europe, ReFLEX™ was
not an option. The “calling party pays” rule also severely hurt the paging industry.62 CDPD
was not an option, because it was precluded by the GSM standard. While Mobitex™ public
networks were built in the UK, the Netherlands, Belgium, Sweden, and Finland in the early
1990s, this was long before Internet email and wireless PDAs, and the networks didn’t
interoperate. Mobitex™ in Europe has recently experienced a surge, following in the
footsteps of the US, but it is still relatively small.63 One DataTAC™ 6000 network was
deployed by DeTeMobil64 in Germany, but it was not focused on the mobile Internet or
enterprise messaging.
So SMS provided Europe with an i-mode equivalent for simple two-way messaging. Its
success there has also been echoed in many other countries where the GSM standard
prevails. (See Chart 12.).
C h art 12. C ellu lar S u b scrib ers b y R eg io n an d N etw o rk
T yp e, 2001
MM S ubscribers
An a lo g
Source: GSM Aassociation, CDMA Worldwide, SHG analysis
C. Why No SMS or Two-Way Data in Japan?
Meanwhile, back in Japan, for a variety of reasons neither SMS nor two-way data-only
networks ever took off. First, the conditions that led to the success of SMS in Europe were
absent. When it was choosing a 2G network in the mid-1990s, NT&T DoCoMo preferred
to use its own local PDC technology rather than GSM. And NTT’s two competitors, KDDI
© SHG 2002
- 24
© SHG 2002
and Japan Telecom, went with a combination of PDC and CDMAOne networks. While all
these networks support SMS messaging, unlike GSM, they are not interoperable.
Later, when i-mode was introduced in 1999, DoCoMo decided to play a role similar to that
of ETSI. It permitted users to send and receive e:mails from any other Web phone users,
whether or not they were i-mode subscribers. Combined with DoCoMo’s 60 percent share
of the cell phone market, this insured interoperability.
D. Why Limited SMS and Cellular Data in the US?
As for the US, neither SMS nor i-mode-like cellular messaging ever became dominant,
because there was not a dominant national cellular operator, nor a government regulator like
ETSI willing to impose a uniform cellular standard and sensible rules like “calling party
pays.” In contrast to Europe and Japan, in the late 1980s the FCC – with vocal support
from equipment vendors like Lucent, Motorola, and Qualcomm – decided to leave digital
cellular network standards up to the so-called “free market.”
This laissez-faire approach to wireless networks was of course a striking contrast to the
rigorous – wildly successful, in retrospect -- standards that the Internet’s founding fathers,
including academic institutions and the US Government, adopted for wired networks at the
very same time. In a sense, the FCC’s approach was rather like leaving the choice of
railroad gauges, highway dimensions, Internet routers, or central office switch design up to
local communities. The result was an alphabet soup of conflicting, proprietary cellular
networks that – relative to Europe and Japan -- generally provide limited coverage and
poor service, to this day. These include the original AMPs system, TDMA/IS-136 (AT&T
Wireless, Cingular Wireless), CDMA (Verizon, Quest, Sprint PCS), GSM (Voicestream), and
iDEN (Nextel). Each of these networks has very different performance characteristics and
upgrade paths.
Later on we will examine the profound implications that the resulting potpourri will have for
the adoption of 3G mobile wireless in the US – for better or worse, in the words of the
Charlie Parker song, “Gonna Be a Long Time Coming…” But for purposes of
understanding the fate of SMS and cellular data messaging in the US, the key result was that
the laissez faire approach fundamentally undermined interoperability, which was so important
to the growth of data messaging in both Europe and Japan.
In the US, unlike Europe, SMS messages simply can’t be sent from subscribers on one
cellular network to those on another. On the other hand, in the US the wired Internet is also
more readily available. The result is that even more though SMS messages in Europe and Imode messages in Japan have higher unit prices than SMS messages in the US,65 US cell
phone users only send an average of less than 1 message a month, compared with 35-40 in
Europe and Japan. (See Table 3 below.) If US cell phone users want to send text messages
to each other, they are forced to choose between non-interoperable SMS, cumbersome
WAP, and a hard place.
For our purpose another key result of the US approach to telecommunication regulation was
that it left the door wide open to two-way mobile data-only networks. On the one hand, as
© SHG 2002
- 25
© SHG 2002
we’ve just seen, SMS messaging was crippled. On the other hand, for years cellular service
simply wasn’t very good, nationwide service was impossible because of disparate standards
and sheer geographic area, and the difficulties created for paging by “calling party pays ”
were absent.
So today the US plays host – in addition to four different digital cellular network standards – to
all the low-speed data-only networks known to man. These include three nationwide
ReFLEX™ networks, a nationwide Mobitex™ network, two semi-national and five regional
CDPD networks, and two nationwide DataTAC™ networks. (See Table 2.)
Table 2. Two Way Data Networks by Market, 2001
ReFLEX™ 1 (Tokyo) – telemetry
3 national
1 nationwide
5 national
2 (1 national)
1 local (Germany)
5 regional, 2 semi0
2 providers
Other SMS
3 interoperable
8 non interoperable
Internet messaging i-mode/ interoperable
WAP (limited)
WAP (limited)
Source: SHG analysis
© SHG 2001
Of course since SMS messaging and, increasingly, Web phone capability, are built into all
digital cell phones whether
they are used or not,66 if we
Chart 13. US Outook for Messaging Devices, 2000a-2005e
simply look at the number
of subscribers in the US
market, we might get the
impression that data-only
overwhelmed by cellular
data alternatives. (See Chart
13.) However, in terms of
Web Phones
actual two way messaging,
this is grossly misleading.
As one recent comparison
of US versus European
SMS messaging showed,
Source: Morgan Stanley (May 2001), SHS analysis. For purposes of this table, “two way”
the intensity with which cell
Includes both two-way pagers and all other non-cellular voice handheld two-way devices.
phones are actually used for
messaging is sharply lower in the US.67
MM subs
The good news for low-speed data-only network providers is that even in terms of
subscriber headcounts, the latest forecasts agree that there is a robust future for non-voice
networks. For example, if we group two-way pagers with wireless handhelds, one recent
estimate is that there will be at least 20 million two-way data-only subscribers in the US by
© SHG 2002
- 26
© SHG 2002
2004. 68 Furthermore, we will argue here that if network operators fully exploit the
opportunities presented by new platforms like ReFLEX 2.7, new devices, application
platforms, channel partners, and interoperability, this number could actually be much higher.
Summary – Market Conditions for Success: DoCoMO, SMS, and Two-Way Data
Table 3 below summarizes the most important market structure factors behind DoCoMo’s
success, contrasting the US, Europe, and Japan.
Table 3. Key Market Structure Conditions for DoCoMo’s Success
Cell Phone Penetration
“Calling Party Pays”
National Cellular
Network Coverage
Wireline Costs
Industry structure
Wired Internet Access
PC Penetration
Wireline Cost
2-Way Data Networks
PDA Penetration
One-Way Paging
Two-Way Data
Networks/ Service
SMS/ Internet Messaging
Net. Interoperability?
Good (9.3 cell
sites/ 100 sq miles)
High/ Metered
calls, shared
business phones
3 digital standards
(pdc, cdma, gsm)
DoCoMo) - >60%
+ 2 competitors
Excellent (20-40 cell
sites/100 sq miles)
High/ metered rates
3 digital standards
(cdma, tdma,
10 national
carriers +
One digital standard
(GSM), all countries
2-5 carriers per country (
See Above
Very Low
Very Limited
See Above
See Above
Mod/ Growing
Very Low
Very Low
Not Applicable
Few ( 1 ReFLEX™
operator for
telemetry (Tokyo)
7+ nationwide
(ReFLEX™ - 3;
Mobitex™ -1;
Limited Mobitex use for
5 Mobitex™ local
country networks; SMS
preferred for messaging
High (DoCoMo
i-mode; CSC SMS
on other networks
control channels
SMS- CSCD control
channel (GSM standard)
Source: SHG interviews and analysis
© SHG 2002
Poor (2.3 cell
sites/ 100 sq. m.)
Low/flat rates
© SHG 2001
- 27
© SHG 2002
5. NTT DoCoMo’s Key Strategic Choices
In addition to all these influences on market
structure, DoCoMo’s success was also aided
by a series of adroit strategic choices. Among
the most important were the following:
A Blooming Garden With Low Walls.
To begin with, NTT DoCoMo’s relatively
open approach to content delivery and
pricing encouraged the proliferation of Web phone content. Rather than set up a “walled
garden” that was off limits to non-subscribers or content suppliers that didn’t rent space,
DoCoMo adopted a tiered content production and distribution model. This permitted
anyone to develop and sell content for i-mode subscribers without fees, while giving some
marketing preferences to official content partners. DoCoMo also started out by convincing
leading companies in key industries – for example, Sumitomo in banking – to develop
content. And it also deployed a team of 60 developers to work with outside content
providers, at no cost to them.69
The result is that there are now more than 750 “official” I-mode content sites. In exchange
for 9 percent of their revenues, DoCoMo pre-positions their urls70 on new handsets and
allows them to sell “push” content to i-mode subscribers, who pay for the services on their
phone bills. In addition, because of the open nature of the development platform (see
below), there are also now about 45,000 unofficial sites, offering some 1200 other
applications, including online TV schedules, sports scores, personal banking, gambling,
horoscopes, anonymous “pen pal” dating services, wireless stock trading (now about 15
percent of all Japanese retail stock trading), “Hello Kitty” comics, and the ability to track
daily basal temperatures for women who want to get pregnant. This compares with WAP’s
2000 sites in the entire world.71 Subscribers to “push” content – about 80 percent of all
users72 -- pay an extra $2.50 per site per month. According to DoCoMo, these site
subscription revenues now average more than $40 million per month.73
While initially DoCoMo followed the AOL “high wall” model and didn’t permit its content
providers to be accessed by competitors’ subscribers, by 2001 this policy had changed –
KDDI’s 8 million mobile Web users and J-Telecom’s 7.4 million can now access i-mode
Web content, and vice versa.
The overall result is a medium that offers a compelling combination of mobile messaging,
entertainment, and business services, a strong fit to Japan’s “hima tsubusi”74 urban
commuting culture.
Great Marketing and Strategic Pricing. The explosive growth of i-mode’s service was
also partly due to NTT DoCoMo’s basic product marketing strategy. This was to drive
market penetration and traffic with a combination of (1) cool new handsets (which
DoCoMo designed itself, subcontracted to device manufacturers, and sold through retail
channels under its own logo), (2) the open content platform model described above, which
was easy to develop for and generous to outside developers, and (3) an aggressive pricing
© SHG 2002
- 28
© SHG 2002
strategy. DoCoMo started off by making its basic devices dirt cheap – the lowest cost
handset is only 1 yen. i-mode’s service pricing model is to charge only for packets actually
sent, rather than airtime, at $2.50 per month plus $.02 per kilobyte for the basic service,
plus $2.50 per month per “push” site subscription. Compared with data services in the US,
this is relatively expensive per marginal kilobyte, but on a per message basis it looks cheap –
just $.01 for a short message and 4 cents for a longer email.75 Compared to the high cost of
wired Internet access in Japan, this is very competitive. The average DoCoMo subscriber
spends only about $20 per month, less than half the cost of wired Internet service.76
A Reliable National Network. The fact that NTT DoCoMo’s digital packet-switched
network provides national coverage, good penetration, and is “always on” has also helped
to stimulate both demand and supply.
“Good Enough” Open Technology. Another key reason why DoCoMo was able to
quickly garner so much third-party content was its adroit choice of a development platform
for wireless applications. Rather than use WAP, it opted for cHTML, a protocol that was
rooted in the open HTML 2.0 standard. The protocol specification had been developed by
Access, a Tokyo software company. With NTT’s encouragement, in 1998 Access offered it
to the 3WC, the Internet’s global standards body, to make the language/ protocol an open
standard. This “open” approach has had several key advantages.
o First, even though cHTML is not XML, it is very good at what it does. From day one in
February 1999 cHTML was able to access any website written in HTML, supporting
features like colored screens with up to 256 colors, animated gifs, MIDI ring-tone
downloads, and multi-user gaming, even at 9.6 kbps. Meanwhile, the WAP Forum, which
started two years earlier, lost time trying to introduce a whole new set of proprietary
networking protocols as well as a new web-page language, WML. Until WAP 2.0 ships later
this year, it will have nothing comparable.
o Second, cHTML was easier to learn than WML, requiring HTML developers only to
learn a few new external tags. Unlike cHTML, WML isn’t backwards-compatible with
HTML, and there are very few WML development tools available.77
o Third, i-mode devices talked directly to standard Web servers. WAP devices, on the
other hand, speak only to WAP gateways, not directly to the Internet. As one developer has
noted, WAP is not really Internet access – “It is access to some data that may also be on the
Internet.” These WAP gateways are expensive carrier-run servers that translate Web content
and control user access to selected Web sites.78 In practice, this means that even though imode’s packet-switched version of the PDC network only runs at 9.6 kbps, while its WAPbased competitors run at 14.4 kbps on CDMAOne or circuit-switched PDC networks, imode often performs faster and more securely.79
o Fourth, while NTT’s cHTML is now an open standard, WAP really isn’t, at least not
yet. Among other things, this exposes service providers and customers to risks of patent
infringement or licensing charges. While the WAP Forum has always claimed to be
developing “open” software, the fact is that there could already be some patented “booby
traps” contained within its specification.80 Already, two Forum members have been litigating
© SHG 2002
- 29
© SHG 2002
over rival patent claims, and pressing enterprise customers and carriers for royalties. 81 In
December 2000 the 3WC, i-mode supporters, and the WAP Forum agreed to unite around a
common standard based on xHTML-Basic, which would help to resolve these issues, but the
agreement has not yet been implemented. DoCoMo has actually been eager to see the longdelayed WAP 2.0 completed and deployed, because it has to rely on it for i-mode services
that it launches in Europe or the US.82
Beyond the Browser – Support for Java™-Based Applications. Building on these
foundations, as an extension to cHTML-based services, in January 2001 DoCoMo launched
its “i-appli” (Internet application) version of i-mode. This supports a new series of Web
phones that are enabled with Java 2 Platform Micro Edition (J2ME™), the mobile version
of Sun Microsystem’s cross-platform application development language.83 The main benefit
of running Java on mobile devices is that, unlike browser-based services like WAP or imode, Java permits applications to be downloaded and run locally. This eliminates the need
to be connected to a Web site in order to run games or other applications. Java also supports
better graphics, sound, and agent-based applications, where information – like weather, stock
quotes, corporate sales data --- can be automatically updated, depending on event-driven
triggers. Java also provides better end-to-end security for mobile applications, because it
supports SSL encryption and provides byte code verification.
Of course J2ME™ is just one of several competing micro-operating systems that have been
designed to run on mobile devices.84 And DoCoMo’s experiment with Java has hardly been
glitch-free.85 However, J2ME’s DoCoMo launch has already given it a head-start over rival
mobile OS candidates like EPOC and Qualcomm’s BREW™, which have yet to appear on
devices in any quantity. In less than three months, DoCoMo has already acquired 4 million
i-appli users, who are now downloading applications from at least 38 Java-enabled websites.
By the end of this year this figure is expected to approach 7 million J2ME™ users in Japan
Furthermore, while i-mode was originally positioned as a consumer service, the security and
robustness offered by J2ME™ is also encouraging Japanese enterprises to deploy new
wireless solutions, including corporate e:mail and database access, field sales automation, and
inventory management. Sun Microsystem’s Japanese office reports that many Japanese
corporations are now experimenting with J2ME™ front ends on DoCoMo handsets, to
provide secure access to corporate databases and email.87 Not to be outdone, in June 2001
KDDI also started its “ezplus” Java-based service, using handsets from Hitachi and Casio,
and providing downloadable applications from 32 new content sites.88 Meanwhile, hedging
its bets a little on mobile operating systems, NTT DoCoMo has also formed an alliance with
Microsoft to market new mobile data services to business customers.89
The Real Key ! Mobile Data Messaging. Together, all these cHTML and Javaenabled platform advantages have helped give NTT a commanding lead over its domestic
WAP-based competitors. They have also helped to provide DoCoMo with credibility as a
technology partner, in its efforts to expand its global relationships with players like AOL,90
Microsoft, AT&T Wireless, KPN, Telecom Italia, and Telefonica.91
© SHG 2002
- 30
© SHG 2002
However, if we examine closely where most of DoCoMo’s actual i-mode success has come
from, in terms of traffic, revenue, profit, and customer satisfaction, at least two-thirds of it has little
to do with all the prolific Web content, much less DoCoMo’s choice of Web phones as a
delivery vehicle.
Rather, we would argue that DoCoMo’s distinctive value proposition has really been its
ability to provide reliable, easy to use. secure, “fast enough” two-way mobile data messaging to millions
of users, at lower costs than the available alternatives, with excellent geographic coverage.
For example:
o One recent study of i-mode users showed that 42 percent were using it mainly for e:mail,
37 percent for voicemail, and just 21 percent for receiving Web content.92 Another recent
analysis reported that 78 percent of time spent on i-mode is accounted for by voice (40
percent) or email (38 percent) communication, and just 22 percent by surfing.93
o Other studies have concluded that “email accounts for nearly half of i-mode traffic, ”
that the average i-mode users sends more than 100 messages a month, and that “e:mail is the
killer app on i-mode.” 94
o When i-mode subscribers are asked why they subscribed in the first place, 82 percent
say their key reason was to get email, which is much less expensive and more convenient
than using a wired ISP. Only 28 percent subscribed to use i-mode for browsing.95
o Finally, in July 2001 it was reported that the latest data on i-mode’s “ARPUs” had
DoCoMo a little worried, because they indicated that users might actually be substituting
e:mail for mobile voice calls, reducing ARPUs.96 If this bears out, it would have serious
implications for DoCoMo and other 3G supporters, which have always assumed that data
traffic would supplement voice revenue, not undermine it.97 Indeed, 3G business cases
usually have to assume steep increase in ARPUs over the next decade, to pay for the heavy
initial investments required in 3G networks, while making up for declining voice revenue.98
The success of DoCoMo’s e:mail platform is even more striking, when we recall that all its
messaging traffic is being generated on a 9.6 kbps (maximum) network from devices with
tiny screens, numeric keypads, and relatively short battery lives that were initially designed
for voice, 99 that i-mode permits no e:mail attachments, and that it is limited to less than
500 single-byte characters or 250 double-byte characters.
In our view, this is the key lesson of the DoCoMo experience. Effectively, NTT DoCoMo
has used its market dominance wisely, creating a defacto “open” national standard for
messaging that provides low-cost, pervasive, interoperable services among all wireless and
wired users. The resulting “messaging commons,” combined with strong marketing and
some nifty content, has permitted Japan’s MDM market to soar.
© SHG 2002
- 31
© SHG 2002
6. Beyond i-mode ! 3G’s Inevitability?
Even while building these solid low-speed foundations, DoCoMo has also been moving
aggressively to launch 3G services, which it promises will eventually provide up to 384 kbps
(actually 384 kbps for downlink speed, and 64kbps for the uplink) of shared bandwidth.100 In
late May 2001 it piloted its “Freedom of Mobile Access” (FOMA) W-CDMA service in
Tokyo, offering video phones that deliver 64 kbps of shared bandwidth in both directions,
with working prototypes of new services like mobile video calls and MP3 downloads. The
prototype service and handsets have so far received mixed reviews, but of course it is still
very early.101
NTT DoCoMo’s initial pricing for this service is less than one-sixth that of ordinary i-mode
per incremental kilobyte. Whether or not this is a compelling value proposition depends on the
application. For applications like e:mail the implied prices per unit of value – per message,
video call, download, or still image -- are compelling, but for multimedia applications they
are still pretty rich. For example, as shown in Table 4 below, incremental e:mail on the imode network costs about $.11 cents, but only $.02 cents on FOMA. Even on FOMA,
however, one minute of compressed video costs about $19, and a 30-second video call
Clearly DoCoMo’s 3G profit model needs work. Who do they imagine will pay such prices?
And who will bear the fixed costs of $760 handsets and the 3X increase in the number of
base stations that may be required? 102 Recall the other key factors in i-mode’s initial success
– easy to use, easy to develop content for, profit sharing on content, and simple messaging.
How easy will it be to produce and sell multimedia content for FOMA? Furthermore, what
kind of multimedia do people really want on their mobile phones, anyway? What are the
compelling new services that will raise ARPUs sufficiently over the next decade to replace
declining voice revenues and cover 3G’s incremental costs?
© SHG 2002
- 32
© SHG 2002
Table 4. NTT DoCoMo’s Initial 3G FOMA Pricing vs. I-mode by Application
Plain Old E:
Video Phone
Still Picture
.025 sec
30 sec
1 minute
.5 sec
File size
3MB (?)
6 MB
50 k
Price/ packet: 300 yen/ n +
300 yen/ n +
300 yen/ n +
300 yen / n +
- i-mode:
.3 yen
.3 yen
.3 yen
.3 yen
05 yen
05 yen
05 yen
05 yen
(n = ∑ packets)
(n = ∑ packets)
(n = ∑ packets)
(n = ∑ packets)
Cost per
Marginal MB
- I Mode:
i-mode cost
$.11 per
Not feasible
$.95 per still
FOMA cost
$.02 per
$9.48 (?)
$.16 per still
Source: NTT DoCoMo (2001); SHG interviews and analysis
© SHG 2001
Despite all these fundamental questions about the 3G business model, and the fact that
DoCoMo delayed FOMA’s commercial launch from April 2001 until October 2001 and
nationwide service until 2002, NTT is still predicting that there will be at least 150,000
FOMA customers by yearend 2001, and millions more in 2002.103 When it launches the
service this fall, DoCoMo will become the world’s first 3G service provider -- other than
BT/Manx Telecom on the Isle of Man – to deliver on the ITU’s 15-year-old 3G vision. Its
competitors are watching closely. KDDI, for example, has already announced that it will
upgrade to a CDMA2000 network next year, and J-Telecom has also said that it will launch a
3G network sometime in 2002.104
7. Implications for 3G’s Future Elsewhere
Earlier we saw that much of DoCoMo’s success has really been based on low-speed mobile
data messaging – and more fundamentally, on giving customers what they really want with,
with simple, but reliable technology. Does DoCoMo’s early adoption of 3G undermine this
analysis? If mobile messaging and content at 9.6 kbps have proved so successful, why is
DoCoMo moving so fast with these new services?
DoCoMo’s Exceptionalism. To begin with, there are just as many special background
conditions at work on DoCoMo’s 3G strategy as there were on its i-mode strategy.
© SHG 2002
- 33
© SHG 2002
o First, precisely because it has been so successful with i-mode, DoCoMo is facing an
acute shortage of network capacity on its 2G PDC network. It has to expand in the
next 6-9 months, and it is exploring both 2.5G and 3G alternatives to do so. 105
o Second, from DoCoMo’s standpoint, 3G’s economics are artificially attractive for one
simple reason. Unlike cellular operators in Europe, the US, and elsewhere, DoCoMo
and the two other Japanese cellular operators got all their 3G spectrum – two-by-20
MHz bands per carrier -- for free.
By comparison, Germany’s 14-day marathon auction in early 2000 raised $46 billion for
six two-by-10 MHz band 3G licenses. The UK’s April 2000 auction raised $35 billion for
just five licenses. As discussed below, in the US the 3G spectrum auction has been put
off until September 2002 at the earliest, and if it does occur it will probably not match
these inflated prices. But the costs of clearing the necessary US spectrum of other
incumbents could be huge, and will mostly be born by bidders. (See below.) All told,
once again, Japan may be the exception that proves the rule.
o Third, even with these 3G spectrum subsidies, DoCoMo appears to be hedging its bets.
Even while testing 3G, it is also upgrading many of its 2G PDC base stations to a local
2.5 G equivalent, doubling the number of simultaneous sessions they can handle.106
Perhaps in light of its recent experience with video phones, DoCoMo may also be
backing off positioning its 3G services as radically-different. According to a senior
marketing executive, i-mode services on 3G will look “exactly the same” as on imode.107
Comparative Advantages – Wired and Fixed Wireless Broadband Alternative. As
we have seen, DoCoMo’s services strategy has been heavily influenced by the laws of
comparative advantage. This is also the case with 3G. It turns that the case for 3G in the US
is much more problematic than for 3G in Japan, just because of local market conditions.
o Un-Wired Tokyo vs. Wired New York. In the case of broadband, as noted earlier,
Japan’s deployment of wired Internet broadband service to the home and office is well
behind that in the US, and the associated markets for multimedia accessories like PC
cameras and speakers are also relatively small. 108 Early wired broadband adopters in the
US market are also already concentrated in major cities like New York, Los Angeles, and
San Francisco, otherwise the preferred candidates for 3G wireless. Since 3G networks
like W-CDMA require up to three times as many cell sites and base stations per unit of
area as 2G, this could be a severe barrier to 3G in the US.109
o Urban Density and High Cost Build-Outs. The US population density is also much
lower than that in Japan or Europe. So is the share of the population living in urban
areas.110 The average cell site density for 2G cellular operators in the US is just 2.3 sites
per 100 square mile, compared with 9.4 in Japan, 20 in the UK, and 30 in Germany.111
It has also taken more than twenty years for 91 percent of the US population to have
competitive choices among at least three cellular voice operators112 -- and as we’ll
examine in Chapter V, the cellular network technology with the largest US footprint only
covers about 43 percent of the US population Even apart from any other issues, this
makes the task of providing national competitive 3G coverage – or for that matter, 2.5
© SHG 2002
- 34
© SHG 2002
G coverage – in the US a daunting one. Nor does it encourage us to believe that most
residential or corporate buyers will face serious competition among rival 3G or 2.5 G
cellular service providers in their local markets any time soon. This fact alone
considerably strengthens the long-term prospects for low-speed data-only networks in
the US.
o Fixed Wireless’s Prospects in the US. While fixed wireless ISPs have not made a
huge dent in the US market so far,113 its prospects might actually be about improve. This
is because of a combination of unsatisfied demand for broadband, dissatisfaction with
wired alternatives (especially DSL), significant recent improvements in fixed wireless
technology,114 and the fact that several leading IXCs are now considering the deployment
of nationwide fixed wireless access to the residential and small business markets.115
This is important to us for several reasons. First, 3G networks require a great deal of
spectrum.116 For example, W-CDMA, the preferred 3G upgrade for GSM networks,
requires up to 5MHz for carrier channel spacing, compared with just 200 KHz for 2G
GSM service, 30 KHz for plain old TDMA/AMPS, and 1.25 MHz for CDMA2000, the
“2.5G” upgrade offered by Qualcomm. (See Table 5 below. )
In Japan, NTT’s dominant role in both the market and the government helped it push
regulators, as well as network vendors and handset manufacturers, quickly down the 3G
path, avoiding Europe’s expensive spectrum auctions. In the US, the fact that 3G needs
so much spectrum is somewhat relieved by the fact that the FCC – unlike European
regulators – has not mandated that 3G networks only be built with new 3G spectrum.117
However, the FCC has not yet licensed the additional 160 MHZ needed to provide
national 3G services, and there are many obstacles to doing so.118
First, the FCC has recently experienced a costly reversal of its attempt to re-license
spectrum in the 1900 MHz band that carriers like Verizon were hoping to use for 3G.119
Second, there are serious potential conflicts with fixed wireless service providers like
Sprint and MCI/Worldcom.120 Third, unlike modern Japan, the US has an influential
military and the most politically-influential TV broadcasting networks in the world.121 3G
also faces a serious spectrum conflict with these two powerful incumbents.122
Finally, to the extent that fixed wireless does succeed in the US, it might also significantly
erode the demand for 3G. For example, 3G’s maximum bandwidth in the stationary
mode, 2 Mbps, is far exceeded by that already delivered by 802.11b technology today,
and there are schemes afoot to deploy nationwide networks of “Wi-Fi” hotspots that
might well could limit the need for high-speed mobile devices. 123
© SHG 2002
- 35
© SHG 2002
Table 5. Key Parameters, Leading Proposed 2.5G and 3G Network Systems
“2.5G” Systems
“3G” Systems
3X Ev
US telecoms
SK Telcom
Sprint PCS
200kHz –
144Kbps –
384 Kbps
3.75 MHz
Data rates
1.25 MHz
Up to 115
153.6 Kbps
$.104 -$.415
$.059 - $.089
Nextel, Bell
South, AT&T
461 Kbps
5 Hz
Car: 144
2 Mbps
5 MHz +n*.2 MHz
Car: 144
Walking: 384
Indoors: 2
$..017- $.069
Source: FCC(2001; NTIA (2001); GSM Association;; SHG interviews and analysis
© SHG 2002
© SHG 2001
- 36
© SHG 2002
" Other 3G Supply-Side Problems -- Spectrum, Networks, Handsets, and
Technology. There are also many other serious supply-side obstacles for 3G to overcome,
especially in the US.
o 3G has recently encountered several technical snafus, notably the problem of designing
handsets that are “multimode” – capable of running on 2G as well as 3G -- while
maintaining adequate battery life. The longer it takes to deploy 3G, of course, the more
important multi-mode capability will be.
o Another possible vicious cycle is the shortage of multimedia content and applications
developers. There are exceptions, like Sony, Nintendo, and Sega, but Japanese and
European content applications and multimedia content often don’t translate very well to
US audiences. So this will be an issue even if 3G networks and services perform well
o These technical snafus have already been responsible for significant delays in 3G
services. As noted, in April 2001 NTT DoCoMo slipped its commercial release of the
FOMA service significantly, and indicated that it would not deliver nationwide service in
Japan until sometime in 2002. Even then, its bandwidth will probably only be 64kbps.
In July 2001, Vodafone, the UK cellular operator, delayed its 3G service to 2003,
blaming a shortage of multimode handsets.
o Even apart from the spectrum issues discussed earlier, because the US is such a
potpourri of different digital networks, with many more carriers and vaster areas, it will
take much longer to establish nationwide 3G services. Even sympathetic observers have
recently picked the year 2007 as the earliest that this could happen.126
Other Recent Experiments -- High-Speed Mobile Wireless. It is still not clear just
how great the demand really is for high-speed mobile wireless data in the US, or even Japan
and Europe. There have been several other recent attempts to launch high-speed wireless
data services in the US with technologies other than cellular data, at speeds comparable to
those of 3G’s first release.127 Several of these efforts – notably Metricom’s Ricochet -- have
failed.128 In general, the failures occurred, not because of weak technology, but because
service providers tried to pursue expensive network build-outs without a clear enough value
proposition to attract sufficient customers.129
There is also a host of development work going on with high-speed fixed wireless, as well as
RF and optical local area networks. In the US, the focus has been on using 802.11b or
Bluetooth technologies in the unlicensed 2.4GHz or 5.6-5.8GHz frequency ranges. While
cell phones might may be enabled with these technologies, the objective has been to provide
“last 100 meters” access at speeds up to 11 Mb/s or more to PC laptops and PDAs.
Companies like Texas Instruments and Spectrix are also developing wireless optical LANs
with much higher data rates and more security, at speeds up to 100 Mb/s or more, and
other companies are pursuing even higher-speed “ultra-wide bandwidth” solutions.130 Service
providers are also entering the fray – for example, one well-known national retailer has
begun to deploy wireless LANs in its stores, allowing patrons to surf while they sip.
© SHG 2002
- 37
© SHG 2002
Unfortunately for the 3G industry, if a distributed network of local WLANs, connected to
the Internet, ever took off, it might provide one more reason why high-speed mobile
networks won’t ever prosper, at least in the US. It
" A Brief History Lesson – AT&T’s
would be easy to network all these local POPs
Videophone Debacle. AT&T’s Bell Labs
had been pushing the concept of sending
high speed fiber or fixed wireless
video over ordinary phone lines ever since it
connections, confining high-speed wireless to
successfully delivered a one-way image of
what it arguably does best anyway – the last few
President Herbert Hoover over a phone line
to New York in 1930. By 1964, after years
of work, AT&T was ready to demonstrate a
prototype “PicturePhone” system at the
New York World’s Fair.
The system
worked, but it cost about $500,000, and the
price of a 3 minute video call was $16.
Furthermore, as one report on the
PicturePhone at the time concluded, “Most
people did not like the PicturePhone. They
were not comfortable with the idea of being
seen during a phone conversation. However,
the system’s designers at Bell Labs were
convinced that the PicturePhone was viable
and could find a market.”
Despite this feedback, over the next two
decades Bell Labs continued development of
the phone, and in January 1992 it launched a
new version that was designed for the home,
the “Videophone.” Robert Kavner, the
AT&T Group Executive in charge of the
commercial launch, stated at the time, “This
is the way people want to communicate. The
time is right. The price is right. The
technology is right.”
Evidently, however,
Mr. Kavner was
wrong. The phone, which was priced for
$1000 and cost $1500, only sold about 5000
units. The project cost the company more
than $ 1 billion. The phone was pulled off
the market in 1994. The key reason?
AT&T’s “You Will” mentality. As AT&T’s
corporate historian later concluded, “It is
still not entirely clear that people really want
to be seen when on the telephone. This was
not a question that was really studied before
the introduction.” (Emphasis added.)
8. What Do Customers Really Want?
We refer to this issue generically as the
“Videophone problem,” or the “You Will”
problem, because we believe that at least in the US,
the 3G industry may be about to repeat, on a
much grander scale, the same kind of hubris-driven
disaster that AT&T experienced with its failed
PicturePhone/Videophone experiments in the
1960s and the1990s. (See the sidebar.) 132
In general, even in Europe and Japan, there is still
little public awareness about what 3G even is.133
Nor have we been able to find, in the enormous
3G technical literature, a single piece of market
research that looks closely at what potential
business and residential customers really want to
do with high-speed mobile wireless. In fact, so far
as we can tell, this SHG white paper is the first
critical look at this question.
This may be for a very good reason. For if
customers were really asked about 3G services and
prices, we believe they might well express serious
doubts about their willingness to pay higher and
higher ARPUs for high-speed mobile services that
have such unclear value, especially in “high-wired”
markets like the US.
Mobile Videophones…for Consumers? In
the US market, many customers are having a hard
enough time just making ordinary voice calls over
their cell phones safely while driving, much less while trying to download videos or receive
voice calls.
While DSL and cable broadband services have problems, they are improving, as are fixed
wireless broadband options. And while PCs are certainly highly imperfect devices, even
multimedia fanatics must wonder whether mobile phones with tiny screens, speakers,
© SHG 2002
- 38
© SHG 2002
memories, on-board processing, and earpieces will ever be able to compete with the
bandwidth, storage, processing power, software, integration with peripherals ( printers,
PDAs, MP3 players, etc.), input devices (keyboards, mice, microphones), power supplies
(“infinite” battery life, at least for desktops), high-resolution monitors, powerful speakers,
network integration, and overall ease of use that are routinely offered by wired or wireless
AT&T’s PicturePhone (1964)
Of course, every so often, we might all
enjoy the idea of being able to shop
while walking down the street or chewing
gum, send visual greetings to friends or
enemies while on the move, download a
brand new MP3 or short film while
rushing to catch a plane, or check out the
map of a new neighborhood while
driving through it.
But how much are we really willing to
pay for such frivolities, and how
frequently will we use them? Will it
really add up to the incremental $164 billion of new multimedia service revenues per year
that the UMTS Forum, a leading 3G advocate, has recently claimed that 3G networks will
produce by 2010?134
Mobile Videophones…. for Enterprises? We also suspect that if anyone bothered to
ask potential enterprise customers about it, they would quickly discover that such customers
have even greater doubts about the value of 3G applications, especially in the US. Most large
companies already have high-speed WANs, videoconference centers, and Web collaboration
tools, and are moving toward even much faster wired access and backbone technologies like
10GigE. 135
Furthermore, while several startups are pursuing two-way video business applications for 3G
cell phones,136 the business value of being able to stream video to mobile phones is just not
clear. Apart from, say, providing live TV coverage from the runway on Hainan Island, there
is simply not that much time-critical video that has to be collected and distributed on the
AT&T’s Videophone (1992)
© SHG 2002
For remote presentations, wire-line
distribution is more reliable. Nor are
there vast libraries of business training
videos that just can’t wait until we
return to the office. The very thought
of employees running around the
corporate office with live mobile video
terminals in their pockets is positively
scary, from a security standpoint.
Finally, as other observers have noted
recently, from a service provider’s
- 39
© SHG 2002
viewpoint, the notion of dedicating network capacity that could easily handle more than 700
voice calls to just one video download seems like a high opportunity cost,137 unless we’re
talking about midnight surfers.
Of course running large enterprises often requires the collaboration of large teams. But
unless we are talking about the US Army invading Iraq or Panama, they are not likely to be
highly mobile. Ordinarily, if project teams need remote collaboration, it is easiest to do so
over the Web, using conferencing platforms like Interwise, WebEX, or Groove, desktop
videoconferencing, or room-based Picturetel systems, rather than mobile videophones.
Indeed, desktop video conferencing and Web collaboration both started to take off in the
late 1990s, driven by broadband, falling equipment costs, and better standards. 138
However, after two decades, the entire US
and Web conferencing
The Kyocera VP 210 “Visual Phone” -‘”World’s first color videophone” (2001)
equipment and services market is still less than $34 billion a year. As we saw earlier, there are also
moves afoot to focus high-speed broadband
wireless on the last few yards.139 It seems unlikely
that the market for real-time video conferencing
will accelerate dramatically just because we can
now make movies while we drive.
9. Key Implications, DoCoMo’s Experiment
So what have we learned from this brief detour to
Tokyo, Europe and 3G Fantasyland? How is it
relevant to the future of low-speed data networks?
In any important new market where there is rapid
technical change and substantial uncertainty about
the future, we find that customer decisions about
which technologies to adopt are often based, not
on cold clear technical details, but on popular
perceptions. These include the latest industry buzz
about “the next big thing,” conventional wisdom
about precisely how many units will be sold
in China five years from now, and the latest
brave pronouncements from industry giants
about their wonderful upcoming releases,
only a few months away. The graveyard is
littered with the dusty bones of companies
that may or may not have had vastly superior
technologies, but undoubtedly did not realize
soon enough just how much influence such
subjective perceptions, the clear articulation
of value propositions, and raw marketing
© SHG 2002
- 40
© SHG 2002
have on corporate investments in
This is not really surprising. All truly
important technologies are never quite
finished, so it is hard to characterize
them individually, much less to
compare them. The incompleteness
means that they are really belief systems,
commitment from their sponsors and
developers. It is no accident that
engineers from rival companies can
often never agree about anything. This
is not just because technology is hard
to pin down and compare, but also
because at some level its protagonists
have to act on conviction rather than
pure reason.
Given this subjectivity, we believe that the best place to start in all technology assessments is
with a deep understanding of the historical development of particular markets and customer
requirements. The last thing the wireless world needs right now is yet another round of
mindless forecasts, yet another technically-oriented “white paper” that starts and finishes
with a one-sided comparison of Ts and Cs. We will present ample technology comparisons
in Chapters IV and V below. But we hope that now they will be grounded on a more
fundamental understanding of what customers really want from mobile wireless.
The last two years have taught us some expensive lessons about what customers really want.
The Japanese and European experiences, in particular, reveal several things.
" Simple low-speed mobile messaging accounts for a huge share of non-voice Japan’s so-called
Web phone traffic. Simple low-speed mobile messaging, not mobile Web browsing,
entertainment, or information distribution, appear to be the killer mobile application.
Japan turned to Web phones in droves mainly because other wired and wireless
alternatives for data messaging were not available – just as Europeans turned to SMS
messaging, and Americans turned to data-only networks.
Given a choice, many users in all regions might well prefer to do most of their messaging
and Web services on non-phone devices. We may never know – the history of
telecommunications is that voice services preempted data services very early on, and we
continue to pay the price for the biases that result. At the very least, from the standpoint of
what has become our mantra -- reliable, low-cost, easy-to-use, secure, pervasive,
interoperable mobile messaging -- it is clear that Web phones, much less the mobile
VideoPhones of the future, leave a great deal to be desired.
© SHG 2002
- 41
© SHG 2002
" The Japanese experience also shows that, given an open-standards platform and a
business model that rewards content creation, low-speed networks can be “good enough”
for most messaging, and even for many Web browsing and information retrieval
applications. In other words, the open Web model, not the WAP model, should be our
guidepost, as we endeavor to make mobile wireless services more popular.140
On the other hand, the global wireless industry is now embarked on a costly worldwide
push toward 3G technology, and our hero, NT&T DoCoMo, for reasons of its own, is
leading the way. This is unfortunate, because as we’ve seen, Japan’s own experience raises
serious questions about the phone-centric, bandwidth-hungry, capital- and spectrumintensive, technically risky, proprietary, and vendor-driven strategy that the industry is now
pursuing. It is this “You Will” mentality, spawned in boardrooms and backrooms without
a single customer in the room, that accounts for many of our recent setbacks with wireless
This story is hardly unique. We find similar supply-side hubris in the history of AT&T’s
Videophone, RCA’s Videodisc, France’s Minitel, Sony’s Betamax, the Concorde, nuclear
power, and many other examples. But in most of those cases the impacts were local,
limited to a few companies or at most a single country. Here, we are dealing with the
future of global communications. It would seem to be in order that we pay more attention to
what customers really want, before defaulting to the industry’s blind faith in “newer, faster”
10. Conclusion – DoCoMo’s Lessons
Despite the global wireless industry’s diminished expectations, DoCoMo’s experience
shows us that there is still much to be excited about. The “revolution” that has been so
greatly over-predicted may eventually arrive. But if customers were actually given a voice,
we suspect that it might have quite a different character from the “high-bandwidth/ cellphone based/ mobile browsing” model that many have been pursuing.
In our view, the way forward now is to focus on mobile data messaging, especially for
enterprise customers. From this perspective it turns out that, at least in the US, conventional
low-speed networks like ReFLEX™ and Mobitex™ have much more vitality than is widely
believed. Like most important technologies, these networks are not nearly as “mature” as
they seem. As we’ll see below, ReFLEX, in particular, will soon be revitalized with new
network capabilities, devices, applications, and middleware support that will substantially
enhance its performance and competitiveness.
© SHG 2002
- 42
© SHG 2002
© SHG 2002
ReFLEX™’s Technology – Origins,
Key Attributes, and Direction
- 43
© SHG 2002
1. Introduction – Key Low-Speed Mobile Data Networks
he argument so far has been an extended way of encouraging customers and
application developers to take their minds off vendor daydreams about the highspeed networks of the future, and focus on the incredible value that can be delivered
with today’s low-speed MDM networks, right here and now.
As we’ve argued, beneath the surface, the world’s most successful wireless data services to
date in Japan and Europe have succeeded precisely because they delivered reliable,
affordable MDM data speeds that may look “slow” by comparison to broadband, but are
more than enough to get the job done, especially for enterprise applications. In addition to
the “mobile Videophone” analogy, there is also the case of the “Lamborghini SUV.” Italian
race car makers can easily produce cars that can hit 220 mph in 6 seconds, but this is not
very helpful to those of us who have to battle rush hour traffic and unforgiving cops on the
way to work every morning. We just want to get there reliably, securely and at reasonable
So at this point we will direct the reader’s attention to a narrower question. What are the
Chart 14. Mobile WAN Wireless Networks -- Family Tree
Higher-Speed 2-Way
Mobile Data
(32 Kbps -2 Mbps)
3G Digital Packet Data
(CDMA 2000 3X, W-CDMA,
Analog Cellular
Dial-Up Data 1G
Walkie Talkie
and Mobile
2.5G Digital Packet
(GPRS, CDMA2000)
2G Digital CSD
Key Low-Speed
2-Way Mobile Data
(4Kbps - 32 Kbps)
Digital Paging
Other Proprietary
Data-Only Networks
Reflex™ 25/50
Reflex™ 2.7
Aeris™, Cellemetry™
© SHG 2001
most important low-speed MDM technologies to consider, and how do these network
technologies compare?
As summarized in Chart 14, the two-way mobile wireless “family tree” can be divided into
three branches. First, there are technologies like Cellular Digital Packet Data (“CDPD”),141
digital circuit-switched data (“DCSD”), analog control channel (sponsored by telemetry
players like Aeris and Cellemetry), and 2.5- and 3G digital packet-switched cellular data that
are rooted in cellular voice technology. Second, there are technologies like Motorola’s
© SHG 2002
- 44
© SHG 2002
ReFLEX, Nexus™,GWcom’s Planet™,142 and AT&T Wireless/ Ericsson’s pACT™143 that
had their roots in one-way paging
networks. Third, there are a wide
Chart 15. US Wireless Data Subscribers by Network
Technology and Service Provider, Q1 2001
variety of other proprietary two-way
networks that have diverse roots, like
Ericsson’s Mobitex™, Motorola’s
and Siemen’s
™, Nexus™,
For our purposes here, the key
competitors to focus on are those
2000 Metrocall
networks that have either already
(000 Subscribers)
achieved substantial subscriber bases
in the US person-to-person MDM
market,144 or are 2.5 G technologies like CDMA2000 or GPRS that may become important
competition in the next 2-3 years. As noted in Chart 15, in 2001, from this angle the key
US data-only networks are ReFLEX, Mobitex™, CDPD, and DataTAC™, with Metricom’s
hapless Ricochet™ bringing up the rear.
US Subs
All told, these data-only networks now account for nearly 2.7 million subscribers in the US,
growing at more than 15
percent per quarter. (See Chart
Chart 16. Growth of Key Two-Way Mobile Date Service
16.) It is especially interesting to
Providers , US Market, 2000-01
note the leading role played by
providers in this story. During
the first half of 2001 their
subscriber base has increased by
forty-five percent, and they now
account for over sixty percent of
all two-way date network
subscribers in the US. Until now,
Mobitex™) and Motient (on
DataTAC™) have received more attention. This is partly just because they began operations
several years before ReFLEX, and supported popular devices like Palm™ VIIs and the
RIM Blackberry™. It is also because ReFLEX’s carriers were associated with one-way
paging and its financial difficulties, were viewed as competing among themselves on
different networks (e.g., Skytel’s ReFLEX 50 network vs. Arch/Weblink’s ReFLEX™ 25),
and perhaps also because they paid less attention to marketing. Whatever the reasons, one
key theme of this chapter is that this is all about to change – in particular, the ReFLEX
providers are all now uniting behind a common platform, Version 2.7, with many
If we compare these figures on two-way subscribers with the nation’s 117 million cellular
voice subscribers,146 or even the 5.8 million cellular subscribers who now have circuitswitched “WAP” services, at first glance the two-way data numbers look modest. (See Chart
17.) However, as noted in Chapter III, the actual use of WAP phones for data messaging
© SHG 2002
- 45
© SHG 2002
is very low, while data network subscribers use their devices an average of more than twice a
each day. 147
Chart 17. Growth of Web Phones and Two-Way Data, US
Market, 2000-2001
MM Subs
2Way Data
Furthermore, for the foreseeable
future, data networks like
ReFLEX™ will also be quite
competitive with the services
offered by cellular operators, in
network coverage, reliability,
devices, and cost.
Sprint PCS
ATT PocketNet
Finally, there are now more than
37 million one-way paging
Mar-00 June 00 Sep-00 Dec-00 Mar01 Jun-01
subscribers in the US, two-thirds
of whom are the customers of
Arch, Weblink, and Skytel, ReFLEX™’s three nationwide service providers.148 Many of them
could easily become two-way data customers, assuming that there is adequate network
capacity. And as we’ll see below, one of Version 2.7’s most important features is its capacity.
Combined with new low-cost, more powerful devices, applications and content, and other
new capabilities, we believe this will provide a strong technical foundation for ReFLEX’s
continued growth.149
2. ReFLEX’s Origins
It is helpful to review ReFLEX™’s origins and evolution, in order to understand its current
architecture and comparative advantages, and how some of its key early barriers to growth
are now being overcome.
In June 1993 Motorola took the paging world by storm with the announcement of its new
FLEX™ binary protocol for one-way digital paging. Compared with its two other key rivals
at the time, POCSAG150 and ERMES, 151 FLEX™ was much more efficient, with higher data
rates, a more flexible range of data rates152 --- hence the name -- better error correction and
reliability, and greater network capacity. For the global paging industry at the time, which
was growing by leaps and bounds and was running out of capacity, this was a godsend.153
Skytel launched the first commercial FLEX™ service in the US in March 1995.154 By 1999 it
had been adopted by 229 carriers in 47 countries, and became the official paging standard
in China, India, Japan, Korea, and Russia. By then, the peak of the one-way paging market,
18 of the top 20 US operators were using FLEX™, accounting for more than half of the
country’s 48 million one-way subscribers.
FLEX™’s success provided solid foundations for ReFLEX™, which Motorola released in
September 1994. 155 To the FLEX™ industry, ReFLEX™ was positioned as a way to further
expand network capacity, allowing devices to register their locations and avoid the need to
broadcast to all geographic regions at once.156 ReFLEX was also the world’s first two-way
paging platform,157 designed to take advantage of the FCC’s 1994 decision to auction 2
MHz of spectrum in the 900-941 MHz band for what it called “narrowband personal
communication (N-PCS) services.” 158 Most of these licenses went to the corporate
© SHG 2002
- 46
© SHG 2002
precursors of the same players that are still ReFLEX™’s leading service providers today.159
The resulting uniform nationwide coverage continues to provide an important advantage
over cellular data alternatives.
3. The First Launch – Mistakes and Roadblocks
Unfortunately, from a commercial standpoint, ReFLEX™ got off to a very bumpy start.
Ironically, it seems that Motorola neglected the lessons it should have learned from FLEX™
about the value of defacto standards, prolific low-cost devices, and clear value propositions.
One reason why ReFLEX™ is now on the verge of a comeback is that several of these early
mistakes are now being overcome.
Protocol Subdivisions. On of Motorola’s first missteps with ReFLEX was that in
March 1995 it effectively bifurcated the protocol, granting MTEL, Skytel’s parent, an
exclusive license to co-develop a higher-speed version, “ReFLEX50. 160 Skytel launched a
nationwide two-way messaging service based on ReFLEX50 in September 1995.161 But the
R50 network proved to be costly to expand, requiring many more receivers per unit of area,
and it was not interoperable with ReFLEX25, the lower-speed version later adopted by all
other US service providers. In contrast to the open-systems platform approach later take by
DoCoMO, this bifurcated, semi-proprietary platform approach limited the development of
common devices, network gear, applications, and services for ReFLEX™.
The “VoiceNow” Distraction. Motorola was also confused about ReFLEX™’s true
value. In 1996 it entered into other development agreements, this time with PageNet, the
largest US paging operator, and ConXus, to use InFLEXion™ and part of their PCS
spectrum to launch a service called “voice paging.” PageNet’s “VoiceNow” and ConXus’
“Pocketalk” services, based on a sort of gold-plated version of the InFLEXion™ network,
allowed subscribers to receive short compressed voice messages on special Motorola Tenor
pagers, which PageNet relabeled “portable answering machines.” It required special
network gear and devices, 163 was a heavy consumer of capacity,164 and provided spotty
coverage. Most important, its value proposition was completely unclear, especially as cell
phones with voicemail proliferated. The problem was that, unlike two-way text messaging
devices or even cell phones, voice paging didn’t provide an easy way to respond to messages,
unless users carried both voice pagers and cell phones – which already had voicemail ! In
the grand tradition of the Videophone, no one bothered to test voice paging on real
customers before its launch, and it soon proved to be an expensive flop.165 Announced with
great fanfare in 1997-98,166 by mid-1999 voice paging was virtually dead, and ConXus
declared bankruptcy in May 1999.167
Capacity Expansion Issues. Another key obstacle to ReFLEX™’s takeoff was the
fact that its first-generation network architecture was inflexible – some said that it should
have been called “InFLEX.” While it was relatively inexpensive to upgrade an existing
FLEX™ network to ReFLEX over a broad geographic area, adding capacity in specific
areas was difficult – capacity either had to be added everywhere in the network at once, or
nowhere. Over time, progress was made toward establishing sub-zones that could be
expanded independently, but it was still difficult to add capacity precisely where local
bottlenecks developed.
© SHG 2002
- 47
© SHG 2002
Proprietary Platforms and Devices. Motorola’s original profit model for ReFLEX™
may also have been too “closed.” The strategy appears to have been an extension of its
FLEX™ strategy, where it sold millions of its own proprietary pagers and hundreds of
networks. But this didn’t do justice to three distinctive facts about two-way messaging. First,
there were many more portable devices that could take advantage of it. Second, it played a
crucial role in a much wider variety of applications. Third, the rise of the Internet made an
open-systems approach toward third-party developers and vendors much more viable.
Motorola was justifiably proud of ReFLEX™ and the fact that it introduced the world’s first
two-way paging device, the Pagewriter 2000, in April 1996. 168 However, for a variety of
reasons, it appear to have been somewhat slow to license ReFLEX technology to other
companies that were really leading technology vendors. It was also slow to take advantage
of the Internet’s takeoff and the boom in the demand for personal digital assistants
(“PDAs) – in particular, the fact that by the late 1990s, millions of Palm™ PDAs were
being sold in the US, offering desktop connectivity and a base for future wireless data
Motorola saw to it that FLEX™ and ReFLEX™ were both included in the WAP Forum’s
protocol specifications in 1998, provided an SMTP gateways for ReFLEX, and tried to
launch an email client “VClient” for the Pagewriter2000 in June 1998 that offered access to
standard corporate email systems like Lotus Notes™ and Microsoft Exchange™. But it had
a hard time matching the demand for Palm and – later – RIM’s much more popular devices.
Starting out with at least a two-year lead over competing two-way device and network
vendors in the MDM market, it managed to squander this lead, in large measure because it
had a hard time transcending its “proprietary” model for hardware, software, and
In the late 1990s Motorola started to become more open with ReFLEX™, perhaps just
because it decided to focus on cellular networks. In September-November 1998 it
announced the development of a first generation ReFLEX™ chipset, in conjunction with
Texas Instruments and several other companies, to encourage the development of thirdpartly ReFLEX applications and devices.170 In April 1999 it licensed Glenayre to produce
ReFLEX™ infrastructure gear and to develop ReFLEX™’s protocol.171 More recently, as
we’ll see below, it has more aggressively pursued licensing the ReFLEX™ protocol to several
other device and modem manufacturers. All these moves were a defacto admission that, in
this environment, continuing to pursue a “hard-over” proprietary strategy was not an option.
One-Way Paging’s Hard Times. As the US cell phone market took off in the late
1990s, one-way paging stopped growing around the world, and then began to contract. In
the US it fell from a peak of 48 million in 1999 subscribers to 37 million in 2001, even as the
number of cell phone users soared from 74 million to 117 million.172 This sudden
contraction undermined one of ReFLEX™’s key value propositions to system operators,
network capacity. In the US this slowed the migration to ReFLEX™ dramatically. Two of
the largest one-way paging operators, PageNet and PageMart, eventually joined Skytel with
nationwide ReFLEX™ networks in 1999. Tri-State Radio, Metrocall, and Verizon also signed
© SHG 2002
- 48
© SHG 2002
on as resellers for the three US ReFLEX networks. But only a few other ReFLEX™
networks were built, mainly in Canada and Mexico.
The one-way paging contraction has meant hard times and corporate reorganizations for
most paging operators in the last few years. By 2001, Skytel had been acquired by
MCI/Worldcom, PageNet ‘s network and licenses by Arch Wireless, and PageMart became
Weblink Wireless and filed for bankruptcy protection. When the dust settled, these three
ReFLEX operators ended up with networks that now serve more than 1.6 million twoway subscribers, more than any other US MDM network.173 Despite this, all three have
continued to struggle with profitability, as they cross the chasm from one-way to two-way
services. Meanwhile, ReFLEX™’s original technology partners, Motorola and Glenayre, have
both basically decided to exit the business of providing ReFLEX™ network equipment and
devices.174 So the three leading service providers are faced with having to support
ReFLEX’s continued technical development, including new devices and applications.
Ironically, we believe that this might actually turn out to be a good thing. The need to
provide customers with devices and applications that actually deliver real value, to focus on
their true competitors, and to rely on more responsive suppliers for software and hardware,
may be just what the ReFLEX™ alliance needs to survive. After all, it is not unprecedented
for service providers to take a lead role in designing their services, devices, network
technologies, and applications – that is what AT&T did for decades with telephony, and it is
also precisely what DoCoMo did with i-mode in Japan. In fact, one can make a case that, at
least for communications, the separation of specialized “engineering/network” companies
from service providers has been detrimental. But that is a larger issue. For our purposes,
while the jury is by no means in, we believe that important progress is already being made
toward the goal of removing all the key obstacles to ReFLEX’s revitalization described
With this background, let’s now look closer at ReFLEX™’s current network architecture and
the role that Version 2.7 and other pending technology improvements will play in this
4. ReFLEX’s Original Technical Attributes
As noted, ReFLEX was originally designed as an upgrade to FLEX™’s one-way paging
infrastructure, extending its core advantages -- high network capacity, low systems
capacity cost, wide-area coverage reliable message delivery, long battery life, and easy
upgrades. These roots continue to determine many of ReFLEX’ most important
Great Battery Life. ReFLEX adopted FLEX ’s frame structure and synchronous
digital messaging protocol, which were already very efficient. This means that ReFLEX
devices offer unparalleled battery life – one month or more on a single AA battery.
Signal Fade and Error Correction. ReFLEX uses FLEX™’s protection against signal
fading, which can withstand up to 10 ms of signal fade at all speeds and still accurately
© SHG 2002
- 49
© SHG 2002
decode information. ReFLEX’s error correction mechanisms -- checksum validations and
positive end-of-message control – are also very robust. This is especially important for
reliable WAN messaging and information distribution applications.
Rapid Two-Way Messaging and Delivery Confirmation. From one angle, what
ReFLEX’s designers basically did was to take a successful one-way paging network and add
a separate return channel on paired licensed frequencies. This provided a very efficient
system for delivering short asynchronous two-way messages, including immediate message
delivery confirmation – a feature that to this day is still not provided by ordinary wired
Internet email.175
Frequency Re-Use and Store-and-Forward Messaging. The return channel also
permitted frequency re-use and an increase in overall network capacity, by using the return
path to identify where the user devices were located, then channeling messages only to that
area.176 If devices were out of range, ReFLEX also automatically stores messages until the
devices return to coverage.
Integration With Other Data Networks. Partly because of its roots, ReFLEX’s
protocol is able to operate concurrently with most other existing paging protocols around
the world. A ReFLEX system can also run in time-share mode, permitting service providers
to run more than one protocol on the same network.
Architecture – Pros and Cons of Alternative Designs.
A. Cellular Networks. Most twoChart 18. Basic Data Network Architectures
way data networks (e.g., Mobitex™,
A. One-Way Paging
Ricochet™, and 2.5 G) are designed
as cellular systems, with each base
station dedicated
to serving
geographic areas (“cells.”)
theory this makes better use of
spectrum, because it is only used
where devices are actually located.
But this doesn’t come for free. It
requires the network to keep track
of device locations, providing
capacity as needed in particular cells. Cellular architectures are also less expensive to modify,
once enough base stations have been installed to cover a certain geographic service area. If
more capacity is needed in a given area, new base stations can be added locally. Noncellular networks, including one-way paging networks, require capacity to be added
everywhere at once. But once again, this advantage doesn’t come for free – cellular systems
generally also have significantly higher fixed costs of coverage, and a much higher
infrastructure cost per customer served at full capacity.177
© SHG 2001
© SHG 2002
- 50
© SHG 2002
B. ReFLEX’s Network.
In conventional one-way paging systems, the network acts as one big cell, multi-casting all
messages throughout the system. Outbound messages are collected at a network operations
center, relayed by satellites or wireline to paging transmitters across the country, and
multicast by all transmitters at once to all devices. Unlike a cellular system, the traditional
paging network has no idea where any given device is. All messages are sent over every
transmitter to every endpoint device.178 (See Charts 18 and 19.)
Device Simplicity. The essential non-cellular design has advantages that ReFLEX has
inherited. It simplifies device design, lowers device costs, and reduces messaging overhead -since the network doesn’t have to manage complex handoffs among cells, there are fewer
administrative transmissions and
receptions, and longer battery life
Chart 19. Basic Data Network Architectures
B. Two-Way Paging (ReFLEX™ example)
compared with cellular networks.
Capacity Cost. While the
incremental cost of expanding a
Forward Channel
896-902 MHz
paging network has historically
been higher on the margin, as
noted above, the initial cost of
covering a given area is much
lower. As we’ll see below,
ReFLEX Version 2.7 directly
Return Channel
929-942 MHz
address this incremental capacity
cost issues, provides a one-time
3-5X capacity increase at virtually zero incremental cost, and brings it on a par with existing
cellular low-speed networks.
© SHG 2001
In-Building Penetration/ Reliability. The fact that ReFLEX signals are broadcast to
devices from multiple transmitters at once – properties known as “simulcast” on
transmission and “macro-diversity” on reception – also means far better in-building
penetration and reliability, as compared with cellular systems.
o “Simulcast on transmit” means that individual ReFLEX devices receive transmissions
from more than one transmitter at once. This increases the probability of receiving messages
correctly, raising the effective link budgets for transmissions significantly.
o “Macrodiversity on receive” means that multiple receivers “hear” any messages sent
from a device, boosting the probability of error-free reception. This increases the effective
link budgets on receive. 179
Greater reliability also means improved network efficiency, because busy-hour delays are
minimized, cutting the need for retransmissions. The reduction in retransmissions has a
significant effect on network delays, especially in the busy hour. In addition, ReFLEX™’s
© SHG 2002
- 51
© SHG 2002
network design also permits flexible transmission rates. The network can reduce
transmission rates on receive where coverage capacity needs are low, so coverage can be
provided by fewer base stations. Depending on a carrier’s capacity requirements, the
protocol supports inbound data rates at 800, 1600, or 6400 bits per second, allowing
increased coverage at lower data rates in low-traffic areas, while retaining higher speed
transmissions in high-demand locations.
ReFLEX™’s original design also included many other advantageous features.
o High Power. ReFLEX uses GPS timing signals to synchronize transmission and avoid
interference from adjacent transmitters. Transmitters can use greater power than in
cellular systems – up to twenty times more powerful. The result is a lower signal-tonoise ratios, better coverage, more reliable communications, and better in-building
o Multiple carriers. With ReFLEX’s use of linear transmitters, multiple carriers can be
carried on a single transmitter, reducing the cost of incremental capacity additions. These
additional carriers, which can be dedicated to data payloads (e.g., without network control
overhead), can be added simply by reprogramming the transmitters. Smart antennas can
also be used to improve reception or capacity where it is too expensive to add receivers.
o Non-symmetrical Capacity Increases. ReFLEX permits transmit and receive paths to be
sized separately, and receivers to be added independently. In a cellular system, inbound and
outbound capacity are usually only adjustable in fixed proportions.
5. The Importance of Version 2.7, WCTP, and Other Developments
All of the attributes just described are already available from ReFLEX’s current protocol,
Version 2.6. However, Version 2.7, the first upgrade of ReFLEX’s protocol in three years,
will be deployed by all leading ReFLEX service providers in the next six months.
Furthermore, a consortium of ReFLEX service providers and technology vendors is also
working hard on a new interface for wireless networks, the Wireless Communications
Transfer Protocol (WCTP), an XML-based applications interface for ReFLEX networks.
This will be delivered in the same time frame as Version 2.7.
Finally, several device manufacturers are also about to deliver a variety of new low-cost
designs for ReFLEX-based PDAs, “cradles” and other devices that will work with
Version 2.7 networks.
While there is still plenty of work to be done on all these developments, we believe that
there is enough industry momentum behind them to justify “plausible speculation” about
the advantages they will bring. It turns out that these benefits will be very substantial. The
following are the most important ones for potential customers and development partners:
© SHG 2002
- 52
© SHG 2002
(1) Increased Capacity. A significant one-time increase in network capacity, at almost no
cost, for all existing ReFLEX™ networks.
(2) More Flexible, Lower-Cost Capacity. Increased flexibility in capacity expansions,
and lower incremental capacity costs.
(3) Much Lower Latency. A dramatic reduction in perceived latency over the network,
enabling ReFLEX to support “near-real-time” applications like instant messaging, financial
transactions and wireless POS.
(4) Interoperability and Roaming – “One Big Network.” The ability of new Version
2.7 compliant devices, applications, and subscribers to interoperate and roam across all
ReFLEX networks, assuming that carriers implement the appropriate roaming agreements
and interconnections.
Chart 20. Key ReFLEX™ v. 2.7 Technical Features
and Benefits
AutoCollapse –
“Chat Mode”
Roaming/ InterOperability
Max. Inbound
© SHG 2001
(5) An Open-Systems Application Platform. Much more powerful, easier-to-develop
applications, based on an open-systems based platform that already has strong support in
the global Internet and software development community.
(6) New Low-Cost/ “Open” Devices. As noted, there will be a much broader selection
of new devices for Version 2.7 networks. These will not only deliver lower costs, but also
support much more powerful applications enablers, like J2ME.
© SHG 2002
- 53
© SHG 2002
This rest of this section takes a closer look at these benefits delivered by Version 2.7 and
WCTP. For more technical readers, the specific features responsible for these benefits are
summarized in Chart 20, and described at greater length in Appendix A.
Increased Network Capacity.
One immediate consequence of Version 2.7 is that there will be a sharp one-time increase in
the capacity of all existing ReFLEX networks. This will be realized at virtually zero
incremental cost, because it only requires a software upgrade. Our estimate is that Version
2.7, if fully deployed, will permit all existing ReFLEX networks to realize at least a onetime quadrupling of network capacity – even apart from additional capacity benefits that
may come from interoperability and roaming.181
Depending on how ReFLEX service providers choose to use this capacity windfall, they
may be able to steal a march on non-ReFLEX™ competitors – for example, by introducing
more aggressive flat rate pricing models. As we will see in Chapter V, this improvement
alone should permit ReFLEX™ to compete very effectively with the costs of rival low-speed
networks like Mobitex™ and DataTAC™.
Lower Incremental Capacity Costs/ More Flexible Capacity. By enabling capacity
and coverage to be added in much smaller units, Version 2.7 also reduces the marginal cost of
capacity. ReFLEX™ has always permitted some degree of cellularization, in the sense that
networks could be divided into geographic sub-zones. But within each sub-zone -- usually
metropolitan areas or larger – the network had to retreat to the paging model, broadcasting
information to all transmitters and allocating return capacity from a non-prioritized,
homogeneous pool.
The key thing about Version 2.7 from the standpoint of incremental capacity costs is that it
allows operators to take this sub-zoning approach much farther, by sharply reducing the
minimum size of sub-zones without losing efficiency or boosting latency. The new design is
capable of what we will call “selective cellularization.” This means that operators will now
be able to combine the best of both the paging and cellular worlds, optimizing for either
coverage or capacity, depending on market conditions. As explained more fully in
Appendix A, the specific 2.7 features that permit this are background scanning and a new
capability of broadcasting maximum inbound message lengths.
Where wider coverage is needed, as in less populated areas, it will now be possible to tune
ReFLEX to function more like a pure paging network, with higher power and
synchronized base stations. Where greater capacity is needed, as in urban areas with intensive
network traffic, ReFLEX can be tuned to act like a cellular network, with lower-powered
transmitters, more extensive frequency reuse, and “micro-cells,” individual enterprise
campuses. For operators this means an incredible degree of operating flexibility. For
customers, it means that ReFLEX can deliver even more reliable messaging at low cost,
even across service areas that vary greatly in congestion.
© SHG 2002
- 54
© SHG 2002
M M s o f IM Use rs
Lower Latency/ Quasi-Real-Time Applications. As noted in Chart 21, instant data
messaging is one of the fastest-growing communications applications in the US, yet another
example of the success of
simple messaging technology.
The use of IM by corporate
Chart 21. Growth of US IM Messaging
enterprises is projected to
grow even faster, driven by
its low cost, ease of use, , and
C h art 21. G ro w th o f U S In stan t M essag in g ,
the proliferation of Web
conferencing.182 So far almost
all this IM activity is among
wired PCs -- AOL is now the
leading IM provider, with
more than 33 million IM or
ICQ183 users. But there is also
providing wireless data users
with IM capability.
mobile data devices are also increasingly being used for other applications that demand
quasi-real-time communication, including wireless point-of-sale, multi-player games and
stock trading.
C o rp U s e rs
O th e r U s e rs
S o urce : IDC (2001), S HG (2001)
© SHG 2001
Unfortunately, all such “quasi-real-time” applications face the issue of latency, the speed
with which wireless devices set up and complete communication with other devices on the
network. This is not a bandwidth issue, but a question of network signaling and control.
ReFLEX’s latency has historically been relatively high, on the order of 30-60 seconds or
more, because of its approach to conserving battery life and scheduling inbound
messaging.184 As described in more detail in Appendix A, however, several new features in
Version 2.7 – including “auto-collapse,” chat mode, broadcasting maximum inbound
message length, and unscheduled inbound messaging – promise to reduce ReFLEX’s
latency by more than 75 percent, from 30-60 seconds down to 5-15, or even less under
certain circumstances.185 Assuming it also follows through with V.2.7-compliant endpoint
devices, this should permit ReFLEX to compete very effectively for real-time applications.
Interoperability and Roaming – “One Big Network”
When Version 2.7 is fully deployed, from the user’s perspective there will no longer be any
difference between ReFLEX 25 and ReFLEX 50. Applications and devices that are V.
2.7-compatible will work seamlessly on all networks. Assuming that ReFLEX operators can
reach agreements, it should also be easy to roam transparently across all these networks,
sharing capacity and coverage. In addition to facilitating interoperable services, roaming, and
shared economies in applications development, this should also result in more efficient use
of the combined network capacity.
Given that, as we’ve seen, ReFLEX™ operators already have the largest collection of twoway MDM customers in the US, and that Version 2.7 will provide capacity for at least
© SHG 2002
- 55
© SHG 2002
three-four times this customer base, this should also help to attract even more device
manufacturers, application developers, and solution providers.
Finally! -- An “Open-Systems” Application Platform
In addition to V. 2.7, there are also several other important supply-side developments that
should help to strengthen ReFLEX’s competitive prospects. One involves an effort to
make ReFLEX’s application development environment easier to use and more accessible.
Following in the footsteps of the approach taken by leading wireless players like RIM and
Palm, several ReFLEX operators formed a consortium, under the auspices of the Personal
Communications Industry Association, to develop an “open systems” gateway for
ReFLEX networks.186
This “Wireless Communications Transfer Protocol”(WCTP) 187 provides an additional
gateway (in addition to an SMTP188 gateway) between ReFLEX™ networks and the outside
world, using standard Internet protocols like XML and HTTP. Applications can treat
ReFLEX devices as ordinary Internet nodes, with WCTP handling all the gory
requirements of interfacing with the network.
This approach offers several key advantages:
o WCTP permits developers to create applications without having to know all the
intricacies of ReFLEX networks. Any developer with a basic understanding of XML
should be able to create applications quickly and cheaply.
o Once WCTP-compliant applications are written, they should work on all ReFLEX
networks. WCTP can also provide an interface to other wireless data networks, including
SMS. This provides more ways for ReFLEX applications to reaching other MDM users.
Since applications are written in standard code, they will also be portable to nonReFLEX networks.189
o WCTP is a more efficient user of network capacity than naked TCP/IP, which was
designed primarily for wired networks.190
Overall, WCTP should open up ReFLEX to a much wider community of solutions
providers and application developers. In a sense, therefore, the addition of all this new
“development capacity” is just as important as the additional network capacity that V. 2.7
Application-Specific Latency. Another helpful development for mobile wireless
applications is that the new ReFLEX V. 2.7 protocol allows for an adjustable mix of
multiple access protocols on the return channel, and application-specific latency. For
example, capacity can be reserved for contention multiple access for minimum latency or
maximum capacity. Even for a given device, some applications can use the lower-latency
allocation while others use the higher-capacity allocation. This feature is unique to
© SHG 2002
- 56
© SHG 2002
New Devices – With Support for J2ME ! As noted earlier, one key constraint on
ReFLEX’s success has been the shortage of really exciting, customizable MDM devices.
This obstacle, too, is about to give way. Without revealing any trade secrets, a whole new
family of V. 2.7-compliant mobile devices are now in the pipeline, for delivery in the next 36 months. Some are aimed mainly at simple, low-cost two-way messaging, which was
initially targeted by Motorola’s T900 – for this segment, ReFLEX provides a unique
advantage, because of its ability to handle very low-cost endpoint devices. Others include
cradles for leading brand-name PDAs and new integrated wireless PDAs that will offer
“industry standard” operating systems and cross-platform operating systems like Sun
Microsystems’ “J2ME™, the important enabler for wireless messaging in the enterprise
segment that we examined in
Chapter III. 191
Motorola’s T900
6. Summary – The Technical
Foundations of ReFLEX’s
All told, the innovations just
described should be able to
overcome all the key obstacles to
encountered during its first launch in the mid-1990s, including inflexible capacity, noninteroperable networks, “closed” approaches to hardware and software, a shortage of
devices, and perhaps most important, a lack of clarity about ReFLEX’s real value
The only cloud that still lingers is the continued plight of one-way paging operators.
However, as noted earlier, we believe that this could actually be an opportunity for today’s
ReFLEX-based service providers. If they move quickly, and apply some of their newfound
capacity, pricing flexibility, improved devices, and capabilities like “instant messaging,” they
may be able to convert a significant share of one-way paging customers to two-way services.
At the same time, we’ve seen that ReFLEX has many compelling performance
advantages even before we get to V. 2.7 and the other innovations. These include attributes
like low cost, reliability, battery life, coverage, and building penetration that are inherent in
ReFLEX’s architecture.
Combined with the exciting developments now in the works, plus some creative marketing,
we believe that there could be an opportunity to actually relaunch ReFLEX, as well as
Mobitex™, as leading technologies for the US enterprise MDM market.
Of course the wonderful thing about capitalism is that no matter how good or bad you are
in absolute terms, it does not matter very much at the teller’s window unless you can also
beat the competition. Chapter V. below examines how the alternative low-speed networks
stack up against to their most important competitors, now and in the future.
© SHG 2002
- 57
© SHG 2002
© SHG 2002
The Comparative Advantages of
Low- and Medium Speed Mobile
Wireless Networks
- 58
© SHG 2002
1. Competitors
ReFLEX™’s technology in some
Chart 22. Wireless Network Proliferation
detail, we can proceed to examine
how low-speed networks like
Reflex™ and Mobitex™ stack up
against each other and their most
2.5 G Networks
important 2.5G and 3G rivals. Of
course in the last twenty years
there has been a
3G Networks
proliferation of wireless data
networks (see Chart 22), driven
by supply-side factors like the
availability of more powerful
microprocessors, new spectrum
allocations, and increasingly clever
engineering, and, on the demand
side, by a seemingly insatiable
demand for mobile communications.192 This proliferation complicates our mission – how
can we possibly evaluate all the alternatives?
Big Leos
Little Leos
CDMA 2000
Cellular Data
(wCDMA flavors)
© SHG 2001
We’ve already dealt with the
hapless world of circuit-switched
Chart 23. Growth of Leading US Two-Way Data Networks, 1994WAP 1.0 and the elusive 3G
“vision” in Chapter III. Those
analyses will serve as bookends
for this chapter. Here, we’ll focus
on those competitors that have
either already achieved the
greatest market success, or are
about to be introduced by major
US cellular operators. As shown
in Chart 23, the key data-only
competitors include Cingular’s
Mobitex™ network,
DataTAC™ network, and CDPD,
which is offered by AT&T
Wireless and several other cellular operators. We’ll concentrate here on Mobitex, which
nearly two-thirds as many subscribers as ReFLEX, relatively high ARPUs, strong technical
foundation and global industry support.
The following section takes a detailed look at Mobitex™, focusing on the factors behind its
revival and its comparative advantages. For DataTAC™ and CDPD, we’ve provided
thumbnail sketches, and they are also included in the detailed tables on network
performance and costs in Appendix B. We’ll also look at the pros and cons of the two
© SHG 2002
- 59
© SHG 2002
leading so-called 2.5G candidates, GPRS and CDMA 2000.193 Finally, we’ll draw some
conclusions about ReFLEX’s longer-term competitive prospects.
2. Mobitex™’s Origins, Decline and Revival
Like ReFLEX, Mobitex also went through a long period of stagnation after its first
launch, and then a sharp revival. Understanding the conditions that made this possible will
also help us understand ReFLEX™, which has also experienced a long maturation period.
Mobitex’s history also illustrates another central theme of this paper – being able to
deliver reliable, low-cost, user-friendly, and pervasive, even if somewhat “slow” -- mobile
data services can be a sufficient condition for competitive success.
© SHG 2001
Mobitex™ was originally developed by Eritel, an Ericsson subsidiary, for Sweden's National
Communications Authority in the early 1980s. Announced in 1984, the first commercial
version was launched in Sweden in 1986, used by Telia, the local phone company, to
manage field service calls. In the late
1980s Mobitex™ was brought to
Chart 24A. Cingular’s Mobitex™ Network
Canada by Rogers Cantel in 1988 and
Coverage, 2001
to the US by a New York startup,
RAM Broadcasting Corp., in 1989.
Over the next decade, 27 more
Mobitex™ networks were built in 19
other countries, mainly in northern
Europe and a handful of Asian
countries. 195
In the US, Mobitex™’s expansion
was accelerated by BellSouth’s 1992
decision to acquire 49 percent of
RAM Mobile Data,196 and its
October 1997 decision to buy
Chart 24B. Arch’s ReFLEX™ Network
control of RAM. At that time,
Coverage, 2001
Mobitex appeared to offer the
lowest latency and a clear path
toward a cellular data future. Over
the next three years, BSWD invested
more than $300 million to build a
nationwide hierarchical network that
grew from 840 base stations and 40
regional switches in 1994 to 1200
base stations in 1996, 1900 in 2000,
and more than 2500 in 2001.197 All
told, according to Cingular Wireless
(Bell South Wireless Data’s new
parent),198 the network now covers about 93 percent of the US “urban business” population,
including 200 million people in 492 metropolitan areas. 199 Of course the US population is
© SHG 2002
- 60
© SHG 2002
now about 276 million, so from Mobitex™’s total population coverage ratio is about 72
percent. (See Chart 24A.) ReFLEX™ networks, in contrast, now cover more than 95 percent
of the US population. (See Chart 24B.)
Despite all this investment, for its first decade Mobitex™ suffered from an acute shortage of
customers. As of 1994, RAM only had about 12,000 subscribers, 5000 fewer than
Motorola’s Ardis’/ DataTAC™ network, and as of 1998, just 125,000, for a network that had
been designed to handle over a million.200
This slow takeoff was due to several factors:
o For most of the decade two-way messaging on Mobitex™ was a high-cost, lowperformance novelty. Before Internet messaging and the PDA booms of the late 1990s,
there was a dearth of low-cost messaging devices and services. Users had to make do with
expensive, hefty external modems or PCMIA modems on bulky, slot- and batteryconstrained laptops.201 Nifty devices like RIM’s Interactive Pager and the Palm VII did not
appear until late 1998.202
o These devices and their service plans were also pretty pricey, until the price cuts of the
last two years.203
o Since the Mobitex™ network didn’t support native IP, even when it arrived, Palm VII’s
“Web clipping” service received terrible WAP-like reviews.204
o The Mobitex™ network didn’t interoperate with the other data-only networks.
Furthermore, before the Internet took off, mobile access was much more cumbersome.
Connecting one’s corporate email system to the Mobitex™ gateway required a dedicated
X.25 connection, which few businesses could afford.
o Nor were there many convenient development platforms around to facilitate application
development. For mission-critical applications, customers also found Mobitex’s coverage
and in-building penetration lacking. For example, when UPS and Fedex considered using
Mobitex™ (or DataTAC™) for heavy-duty package tracking in 1993-94, they decided to
build custom solutions instead.205
What apparently did not hurt Mobitex™ was its relatively low data rate. Indeed, the current
(second) release of its software, which dates from 1992, claims a maximum throughput of
just 8 kbps. And even this is overstated -- the effective maximum shared data rate is actually
only about 4.5kbps,206 and in practice most users average just 1.2 to 2 kbps on Mobitex! 207
Despite this network sloth, it turns out that this is perfectly fine for the great bulk of two-way
and email messaging -- especially the 99 percent that is less than 10kb per message and lacks
file attachments.
What users were aware of was latency, the amount of time it takes for a response to be
received from the network, once a message is sent. And here Mobitex established a strong
track record, with latency of just 5-15 seconds under most conditions. As we will see below,
ReFLEX’s new capabilities in this area mean that it will now be able to overcome one of
Mobitex’s most important historical advantages.
© SHG 2002
- 61
© SHG 2002
Accordingly, as “cool,” easy-to-use devices, more competitive service pricing, better
applications platforms, and just plain better marketing, especially by RIM, became available for
Mobitex™-based services after 1998, its subscriber base began to take off, more than
tripling from 200,000 in 1999 to 570,000 in 2000 and 690,000 by mid-2001.208 Furthermore,
many of these were low-churn, high-ARPU business and professional customers. Overall,
Mobitexis clearly ReFLEX’s most important data-only network competitor, with very
high levels of performance and customer satisfaction.
3. Mobitex™’s Strengths and Weaknesses
Mobitex™ is described by its supporters as an “open,” global network protocol that is
available royalty-free to all members of the Mobitex™ Operators Association. But the
protocol’s copy write is owned exclusively by Ericsson, which carefully controls its
distribution and modification. As noted, the most recent version, V.2, dates from 1992. V. 3
which is supposed to offer better building penetration and adaptive rate coding that boosts
data rates up to a 16-32 kbps, is long overdue. As we saw above, average throughput is
usually well below such maximums. Furthermore, given Ericsson’s recent economic troubles
and its decision – like Motorola -- to focus on its cellular businesses, Mobitex™ operators
are not holding their breathe for this third release.
A. Mobitex’s Network Architecture
From a technical standpoint, Mobitex™ is a nationwide, trunked packet-data mobile radio
network, with a hierarchical cellular architecture. This is a fancy way of saying that the
network has base stations, regional and national message switches, and a national control
center that are all connected by a private wireline data backbone. (See Chart 25.)
The basic network was
designed for licensed public
Chart 25. Mobitex™’s Hierarchical Packet Radio
and private operators in two
Network Architecture
basic flavors.209 In the Americas
it operates in the 896-901 Mhz
National Control
Center (NOC)
(handset) and 935-940 Mhz
(base station) bands, just above
cellular services, where RAM
Mobile Data acquired about
200 frequencies in the 1990s,
each capable of 8 kbps, with
12.5 KHz channel spacing, and
about 500-750 KHz of
spectrum in all, depending on
the region. (Note: ReFLEX
has 2 MHz). In Europe and Asia the network operates at 415-430 Mhz, with a poorer
selection of devices and applications.210 This global split has undermined the potential global
economies that Mobitex™ might have obtained from deploying common interfaces and
devices. It seems to reflect regional differences in regulatory regimes rather than Ericsson’s
LDC Switch
LDC Switch
LDC Switch
Local MX
Loca l MX
X.25, SNP,
IP, etc.
Base Station
Base Station
Dial up
Base Station
Base Station
© SHG 2001
© SHG 2002
- 62
© SHG 2002
Another key interesting feature of Mobitex’s global picture is there tends to be only one
nationwide public operator per country. This means that national roaming is automatic. It
also means that competition has had to come from non-Mobitex™ networks – an unpleasant
surprise for some customers. In Europe, where there are Mobitex networks in 11
countries, international roaming has been established, but there are no international roaming
agreements among the Canadian, US, or South American networks.
B. Key Mobitex Attributes
From the standpoint of network performance and economics, Mobitex™ has the following
key features:
Packet Switching. Like DataTAC™, ReFLEX™, the 2.5G networks, and the Internet
itself, Mobitex™ is a digital packet-switched network. This means that it is “connectionless, - unlike, say, a phone call, there is no end-to-end session. Instead, messages are aggregated
into short bursts of digital data -- “packets” – with their own identifiers. The packets can be
sent out over the network in any order. Each packet is routed to the common destination,
where they are reassembled in order. In Mobitex™’s case each packet (“MPAK”) holds up
to 512 bytes of data, a half-page of text.
Compared with networks like 2G WAP or circuit-switched CDPD,211 this packetized
approach has several advantages, especially for short messages, the bread and butter of
mobile messaging.
o It shares spectrum more efficiently among multiple users, allowing perhaps 10-50 times
more subscribers per channel.
o Because of continuous connectivity, it permits low latency – less than 5-15 seconds per
roundtrip message. As noted earlier, this is very useful in applications like wireless POS,
OLTP,212 instant messaging, and interactive gaming that require quasi-real-time messaging.
o Packetization also enables message “push,” where senders or host computers can
initiate messages. This allows devices to be, at least in theory,213 “always on, ” without tying
up any network capacity.
" Hierarchical Structure.
Another distinctive feature is Mobitex’s hierarchical
structure. This means that messages are routed only to the lowest nodes common to both
senders and receivers. This means that messages are automatically localized, avoiding the
need for redundant wide-area distribution and wasted spectrum. Sophisticated “shortest
path” routing algorithms are also used when messages have to cross multiple switches.
Specific error correction mechanisms like link-level data checking and forward error
correction are also employed at each level of the hierarchy to improve reliability. Meanwhile,
billing information and administrative data gets passed to the system’s highest level,
Cingular’s Network Control Center in Woodbridge, New Jersey.
Cellular Architecture. Each Mobitex™ base station serves just one local radio cell up to
30 kilometers in diameter, providing “last mile” connections to all devices in the cell. As
© SHG 2002
- 63
© SHG 2002
noted, Mobitex™ channels are 12.5 KHz wide, and each cell supports 10 to 30 channels.214
As subscribers move across cells, their frequency-agile Mobitex radio modems, which
operate on Cingular’s several hundred allotted frequencies, stay connected to the network,
switching to the best channels and base stations available. Like ReFLEX™, this automatic
registration permits transparent roaming, as well as store-and-forward messaging, when
users lose coverage or turn off their devices.
" No Macrodiversity/ Simulcasting. Unlike ReFLEX™, Mobitex users are never
communicating with more than one base station at a time. This explains why ReFLEX™ has
much better in-building penetration.™ Mobitex’s distributed architecture is also not wellsuited to applications that require simultaneous broadcasting to all users. Like other cellular
systems, Mobitex is unable to delivery a single message simultaneously to multiple
destinations – an important feature for applications like advertising, information services, or
group chat.
" Very Low Data Rate. Compared with, say, DataTAC™, whose channel bandwidth is 25
KHz and has a maximum throughput of 19.2 kbps,215 Mobitex™’s narrower 12.5 KHz
channel is partly responsible for its 8kbps maximum data rate.
High Initial Coverage Costs. From a network economics standpoint, Mobitex™’s
cellular structure, plus the fact that all its network software and hardware still come from
just one supplier, account for the fact that it has the highest initial coverage cost of any dataonly network.216 As we’ll see, this is at least 1.5-2x the costs of DataTAC™, CDPD, or
Lower Incremental Capacity Costs. On the other hand, Mobitex™’s modular design
also features relatively low incremental capacity costs, because it is able to add base stations
and subdivide channels selectively wherever traffic demands it. As we noted in Chapter IV,
until V. 2.7, ReFLEX suffered from “indivisibility” -- network capacity had to be added
(almost) everywhere to be useful anywhere. (With V. 2.7, ReFLEX actually pulls ahead of
Mobitex in this category. See below.)
Protocol Support. Mobitex™ really shows its age and heritage when we examine its
support for transport protocols. Born in the pre-Internet heyday of ISO and CCITT
standards and IBM-dominated networking, it supports a multitude of aging protocols like
IBM’s SNA,217 X.25, MTP/1, and the X.24 CCITT standard for public packet-switched data
networks. On the other hand, what it doesn’t support very well is native TCP/IP, which has
long since come to dominate the WAN world. For users, this means that gateways like
Palm.Net and the RIM Blackberry server are required to support ordinary POP3 or IMAP
Internet messaging and, especially, secure access to corporate email.218 This considerably
adds to expense and hassle. The only good news is that most of the other data-only
networks are in the same boat – only CDPD provides native IP support.
Security – “Buyer Beware.” Mobitex™’s key sponsors, Ericsson and Cingular, have
long been outspoken in their assertions that it is a “relatively secure” mobile network,219 a
claim that is often repeated in the press. 220 Indeed, unencrypted Mobitex™ applications are
widely used by police forces, emergency services, and even wireless POS services, especially
in Europe and Canada, However, while there are some aspects of Mobitex™’s algorithms
© SHG 2002
- 64
© SHG 2002
that make snooping challenging, 221 detailed protocols for scanning and decoding
unencrypted Mobitex™ traffic have in fact been freely available on Usenet groups since at
least 1997, and the equipment required is easily within reach of any “RadioShack amateur.”
The fundamental fact is for ALL wireless data WANs and LANs, is that for the serious
hacker, unencrypted traffic provides virtually no security. If security is a real concern, the
only answer is end-to-end encryption at application layer – as is increasingly recognized by
most Mobitex and ReFLEX service providers, and recently emphasized by successful
attacks on wireless networks. 223 Both Palm and RIM have recognized this in designing their
gateway services that run on Cingular’s network. Unfortunately, this requires mobile devices
with more powerful CPUs and real operating systems. This helps to explain the increasingly
powerful processors on the latest Palm and RIM devices, and the growing interest in microoperating systems like J2ME, BREW, and Symbian. 224
" Device, Application Platforms, Middleware Tools, and Industry Support. Strictly
speaking, devices, applications platforms, and middleware are not network properties, but
Cingular’s Mobitex™ network certainly owes a great deal of its success to the fact that
technology vendors like Palm, RIM, Handspring, and Aether are developing devices,
middleware tools, and applications for it.225 This has provided Cingular with a great deal of
joint sales/marketing as well as technical support. The continuing strong support of
Ericsson, Mobitex’s original network vendor, has also helped, as has the fact that Cingular
has very wealthy RBOC (“Regional Bell Operating Company”) parents. These factors alone
go a long way to explain why Mobitex™ is ReFLEX’s strongest data-only competitor, and
the other ReFLEX competitors are far behind.
4. Summary - Mobitex Vs. ReFLEX™
Charts 26 and 27 (A-C) in Appendix B provides more details on technical and economic
comparisons among all these networks. As shown there, when ReFLEX™ is compared toeto-toe with Mobitex™ on performance, economics, and industry support, it compares
extraordinarily well – especially for a network technology that almost no one outside the
paging industry has ever heard of.
In particular:
" Technical Advantages. Even before Version 2.7, ReFLEX™ clearly outperforms Mobitex™
on many critical technical attributes, including
1. Date Rate. ReFLEX™ averages at least twice the actual throughput of the Mobitex
network. This also translates into greater capacity for each coverage zone.
2. Wide-area Coverage. ReFLEX’s national coverage is nearly a third greater than
Mobitex’s.226 This is a decisive advantage for ReFLEX™ in applications that require reliable
reachability, especially where access is needed outside core urban areas. (See the coverage
maps in Appendix B, the coverage estimates provided in Chart 26, and Chart 32 below.)
© SHG 2002
- 65
© SHG 2002
3. In-building Penetration. This is also a decisive advantage for ReFLEX, especially
where urgent/ emergency messaging is required.
4. Broadcasting. This ReFLEX advantage has recently been put to work in an
application for Gemstar, TV Guide’s owner. Gemstar will be using Arch’s ReFLEX
network to broadcast TV programming information to thousands of new TV sets that are to
be manufactured with built-in ReFLEX™ modems.227 The application also leverages
ReFLEX’s two-way capability by permitting viewers to order pay-per-view and
5. Reliability/ Reachability. This is a combination of battery life, coverage, in-building
penetration, and the availability of portable devices. When these attributes are considered
jointly, ReFLEX™ has an even more dominant advantage over Mobitex™ than when they are
considered separately, in terms of the percentage of successfully completed messages within
a given latency range.
6. Low-Cost Devices/ Long Battery Life. As discussed, ReFLEX has much longerbattery-life devices at price points that Mobitex™ can’t come close to.
7. Latency. Prior to V. 2.7, as noted, Mobitex™ had a clear advantage over ReFLEX™ on
latency. But V. 2.7 appears to have eliminated this differential, to the point where both
systems now achieve 5-15 second latencies under normal conditions.
8. Security.
Unless devices support end-to-end encryption, both ReFLEX and
Mobitex are vulnerable to security problems from determined hackers. This may or may
not be a serious concern, depending on the application.
" Economic Advantages. The economic advantages of a given network depend on a
combination of (1) device cost, availability, and functionality, (2) network capital and
operating costs, and (3) the cost and quality of application development. From this angle,
Mobitex™ has an early lead in devices, based on its successful relationships with Palm and
RIM. It also has the edge in proven enterprise applications, wireless applications developers,
and middleware support.
However, if ReFLEX™’s supporters can move quickly enough to take advantage of V. 2.7
and WCTP, these could easily become fleeting advantages. As summarized in Chart 28,
ReFLEX™ networks now have several very important economic cards to play.
o A
Capacity. First, as discussed in
Chapter IV, V. 2.7 will deliver a
huge one-time increase in
network capacity,
at zero
marginal cost. In the short run
some ReFLEX operators will
also be able to take advantage of
© SHG 2002
Chart 28. Version 2.7’s Impact on ReFLEX™’s Cost Curves
V. 2.7 Key Impacts -> Tripling of Capacity
> Lower Marginal Costs
Curve Before V. 2.7
ReFLEX’s Higher
Marginal Cost, Before
V. 2.7
Cost Curve
of Coverage
Curve After V. 2.7
- 66
ReFLEX’s Lower Cost of Coverage
© SHG 2001
Traffic capacity
© SHG 2002
surplus network equipment, left over from other ReFLEX operators like ConXus, and the
merger of Arch, PageNet, and MobileCom. Finally, ReFLEX™ operators in the US now own
more spectrum – 2 MHz! – than all other data-only operators combined. In particular, an
Arch/Weblink combination might deliver unsurpassed national coverage and capacity. All
this near-term capacity should permit ReFLEX™ operators to grow their user base
significantly, and perhaps experiment with pricing models that go after Mobitex™ ‘s relatively
high fixed costs.
o Lower Long-Term Marginal Capacity Costs. As noted in Chapter IV, V. 2.7 will
also sharply reduce Mobitex™’s advantage with respect to incremental capacity costs, to the
point where ReFLEX™ and Mobitex’s incremental capacity costs will become almost
identical. (See Appendix B, Table 2, for an estimate of comparative incremental costs per
MB delivered, and Chart 28 for the economic consequences. )
o Harvesting Opportunities. Once again, as noted in Chapter IV, ReFLEX™ operators
not only have 1.6 million two-way customers, two-and-a-half as many as Mobitex,™ and
already growing at a faster rate. The ReFLEX operators also have more than 30 million
one-way customers who might be converted to two-way. Given the excess capacity in
ReFLEX networks, this becomes a question of devices, applications, support costs, and
sheer marketing – the ability to explain to one-way customers why it makes more sense for
them to trade in their one-way pagers for dependable, low cost MDM devices rather than
switch to cell phones or other devices that are less reliable and cost more to own.
o Channel Power. Assuming that this harvesting proceeds briskly, ReFLEX™ might
soon be sitting on a two-way customer base at least four times its current size. This should
catch the eye of even more leading two-way device manufacturers, middleware providers,
and perhaps a network vendor or two. This should help to overcome ReFLEX™’s current
disadvantage in industry support.
o ReFLEX’s Device Advantages/ Device Proliferation. When V. 2.7 is fully
installed, its new capacity should also help light a fire under ReFLEX’ advantages with
respect to devices. For example, since ReFLEX transmitters are relatively low power – just
0.25-1.0W, compared with Mobitex’s and DataTAC™’s 2W transmitters -- they generate
less heat and interference, permitting less expensive components to be used. As noted,
ReFLEX devices also have superior battery life because an advanced sleep cycle is designed
into the protocol.
This means that new ReFLEX devices should have much lower unit costs, even before we
take into account the scale effects of a large subscriber base. As noted in Chapter IV, and
detailed in Appendix B, several device OEMs like Korea’s Standard Telecom (using the
“Nixxo” brand) and Fine Telecom (“the Telica” brand), and Belgium’s Advantra228 have
already understood this potential, and licensed the ReFLEX protocol from Motorola. They
are working on V 2.7-compliant MDM devices, several of which have price points below
$100 retail. Meanwhile, at the high end, as noted in Chapter IV, there are also other new
PDA-like devices in the pipeline that support the Palm OS and J2ME. These will enable
ReFLEX™ to compete more effectively in the enterprise applications market.
© SHG 2002
- 67
© SHG 2002
Overall, this marks a striking reversal of Motorola’s earlier policy of keeping ReFLEX
device development closed and proprietary, and will help catalyze the V.2.7 launch.
Industry Support. Our last item for comparison has to do with industry support, the
degree to which customers can count on a supply chain of solutions developers, software
vendors, network equipment vendors, device OEMs, and service providers to help them
deploy and support first-rate useful mobile solutions. As summarized in Chart 26
(Appendix B), Mobitex™ has been leading in this category, with support from vendors like
RIM and Palm, solution providers like Aether, and Ericsson, its original parent. It also has
a leading national service provider, Cingular Wireless, that generates more than $1.2 billion a
quarter of EBITDA from its cellular/PCS voice services, and has two wealthy RBOCs as
In contrast, two of the three ReFLEX operators are facing near-term financial
reorganizations, and the strategic intent of the third, MCI/Worldcom’s Skytel, is not clear.
Quite frankly, no matter how technically advanced ReFLEX V.2.7 may be, customers and
channel partners are likely to examine the whole “supply chain” that supports a network,
not just its technical merits.
In short, we believe that ReFLEX™’’s new capabilities, including V. 2.7, provides the
technical foundations for a strong rebound. But the most important challenges that
ReFLEX faces are not technical. Indeed, one key lesson from Mobitex™ is that a very
mature technology was able to revive itself largely because it combined solid technical
foundations with strong partnering, marketing, and financial skills. Fortunately, it appears
that a growing number of device manufacturers, applications developers, and middleware
companies are beginning to take an interest in ReFLEX’s potential.
5. Applications “Fit” – Valuing Technical Attributes
Chart 29. Complex value-added solutions” markets vs. “Basic
service provider” markets
End Users/ Customers
> Content
(n = 150)
(n = 7)
End Users/Customers
Distributors/ Retailers (??)
Solutions Providers
Distributors/ Retailers
Handset mfg
(n = 3)
A. Cellular Voice Industry
ISPs Managed Services
Content Provider/ Portal/Aggregators
Software DevelopersCustom Developer/ Integrators
©SHG 2001
B. Wireless Solutions Industry
applications that are demanded by a particular market.
© SHG 2002
Of course the value of technical
attributes like latency, coverage,
and throughput depends on
applications, and this varies a
great deal. See, for example,
Appendix B’s Table 3, which
summarizes the role of different
attributes in various wireless
data applications. For our
purposes this means that there is
no “absolute” answer to the
question, “Is ReFLEX™ “better”
than Mobitex?” The answer
depends on the mix of
- 68
© SHG 2002
In general, for applications like field sales, vehicle location, or wireless distribution of an
electronic TV guide, ReFLEX™’s solid advantages in coverage, building penetration, and
simulcasting might take the lead. ReFLEX™ has a very strong mix of attributes and cost that
– assuming stronger channel partners and devices – should be able to satisfy a wide variety
of MDM application requirements. If markets are diverse enough, however, there may
actually be plenty of room for both ReFLEX™ and Mobitex™, with service providers shifting
their focus from “selling airtime” to “providing solutions.”
Indeed, as described in Chart 29, there is a clear trend toward the emergence of a “wireless
solutions” industry. This reflects the increasing number of complex wireless and wired
networking alternatives that customers now face. It also reflects the fact that designing
solutions increasingly requires the integration of many different disciplines, from network
security and systems integration to network design and the knowledge of specific CRM or
SFA applications.
This trend is especially important to service providers for data-only networks like ReFLEX™
and Mobitex™ to understand. This is not only because the market demands that they think
about solutions in a more network-agnostic way, but also because – as we’ll see below –
they may soon be forced by “common enemies” to regard themselves more as allies than as
6. The Other Data-Only Network Alternatives
Chart 30A. Other Key Two-Way Wireless Data-Only Networks
Cellular Digital
Packet Data (CDPD)
© SHG 2001
•20-year old 8 kbps two way technology
•Packet-switched cellular network structure
•Ericsson - lead network vendor
• Devices by RIM, Palm, Novatel, etc.
•US nationwide coverage - Cingular (Bell South)
•690, 000 US users, 2001
• Networks in 21 countries
•Packet-switched or circuit-switched data over analog AMPs, TDMA and
CDMA cellular networks -- up to 19.2 kbps max. but 4.8kbps ave.
•146,000 US users, 2001
•US, Mexico, and Venezuela technology
•Leading US providers - AT&T Wireless, Verizon
•Not all metro areas covered
•Data-only mobile wireless network -- up to 19.2 kbps
•272,000 US users, 2001
•Origins in IBM/Motorola Ardis network for field service
•Motient is only leading service provider in US (Datatac 4000 tech)
• Outside US, in Asia and Europe at 400 Mhz, 9.6 kbps max
• Devices - RIM, Korea, Motorola
Earlier we argued that surviving
a face-off with Mobitex™ was a
ReFLEX™ to dominate its other
data-only competitors, as briefly
described in Chart 30 (A and
B). As shown in Appendix B,
this does indeed turn out to be
the case, although each network
excels at some features,
reflecting their peculiar histories.
In general, if ReFLEX™ is
able to overcome the industry
support issues noted above with
respect to Mobitex™, it will also
easily handle these less formidable competitors.
The other data-only networks also
provide more evidence for what we
will call the “copper cable”/”DC-3”
hypothesis. This is the notion that
even mature low-speed date
networking technologies can find a
© SHG 2002
Chart 30B - Other Data-Only Networks
Circuit-switched data
"Dial-up access over cellular networks (analog, CDMA, GSM); (Sprint sends
"9.6 kbps (analog) - 14.4kbps usual max (but Sprint PCS +Blue Nite, or
Broadcloud booster service !56kbps)
"5.2 million US users, 2001 ---> but low intensity use
"Providers - all US cellular operators
"Pricey ($30 connector + $150 modem +$7/mo surcharge +airtime)
"Devices - PC cables for cell phones; special CDMA devices for telemetry
- 69
© SHG 2002
new lease on life if their supporters are willing to focus on real customer needs. In the case
of copper cable, the 100-year old dialup phone system provides a low-cost, reliable, global,
easy-to-use network that anyone can access with just a phone or modem. It is not about to
disappear just become someone has discovered how to send gigabits of data per second over
optical fiber or Terabeam™ lasers. Nor did the DC-3, first produced by McDonnell
Douglas for American Airlines in June 1936, disappear just because of jet planes. Of 10,629
DC-3s ever produced, several hundred are still in operation.230 Meanwhile, McDonnell
Douglas has gone the way of all flesh. In the case of ReFLEX, Mobitex™, and DataTAC™,
we suspect that they may also
The DC-3’s first commercial flight, June 1936
outlive some of the engineeringcentric,
organizations that created them.
7. 2.5G – A Threat to Everyone
Of course the notion that US dataonly networks only need to worry
about each other begs the question
companies have in store with
respect to higher speed data
Chart 32. “Higher-Speed” Cellular Wireless Data Alternatives
networks, beyond their dismal
experiments with WAP and
circuit-switched data.
outlined in Chart 31, the two
most interesting candidates are
the so-called “2.5 G cousins,”
GPRS and CDMA2000, both
of which are digital packetswitched upgrades to existing
cellular networks. There is
already a huge technical
literature on these prospective
networks, and we won’t repeat
that here. For the interested
reader, Chart 28 in Appendix B and Appendix Table 2 summarize their key technical
attributes and potential relative costs. Given the fact that these networks are only just now
being deployed, our assessments are necessarily somewhat tentative. But as usual, some
things can be said.
CDMA 2000
© SHG 2001
Qualcomm’s upgrade to CDMA networks…packetized voice and data, initially
144kbps max !384 kbps eventually (…for what?..)
Already commercial in Korea
Industry support - Lucent, Nortel, Qualcomm (patents), Ericsson, Samsung,
Hyundai, NEC, Lucky Goldstar, Motorola, Nokia (terminals)
Initial upgrade also doubles voice capacity ->econ., but costlier than GPRS
US Rollout Q4 2001(?) – Sprint PCS, Verizon
Initially - coverage in selected metro centers only
Limited device availability
Carrier focus on mass-market, horizontal applications
IP packet data upgrade to GSM, TDMA networks
Doesn’t packetize voice…uses part of voice network for packet data
55 operators, 20 countries so far
US – Voicestream nationwide by 12/20 (for $50 mm)..then ATTW, Cingular
Limited devices so far --$500, battery life
“Always-on” is a theory, given low coverage
Throughput – 115kbps max/ shared, 15-20kbps actual
Voice channel cannibalization problem for data business…..high opp. costs,
most apps
Supported by major network equip vendors (Motorola, Ericsson, Nokia, etc.)
Perpetuating Network Divisions. To begin with, it is important to understand that
these 2.5G upgrades will perpetuate the deep divisions that already exist among rival 2G
cellular networks in the US. Originally the idea was that they were interim upgrades on the
way to a “grand reunion” of TDMA, CDMA, and GSM networks under the 3G/ wCDMA/
UMTS banner. (See Motorola’s version of this roadmap in Chart 33.) However, as we
© SHG 2002
- 70
© SHG 2002
saw in Chapter III, 3G’s timing and economics are in now in doubt. So the 2.5G upgrades
are beginning to look more like longer-term destinations than bridges.
After five years of standards work by ETSI,231 in the late 1990s the GSM Forum proclaimed
GPRS – “General Packet Radio Service” – as the next big thing for the world’s 535 GSM
cellular operators, a digital packet-switched overlay to their existing networks. In theory,
GPRS offers multiple advantages. First, it is supposed to be able to provide higher data
rates, with maximum throughput up to 171.2 kbps, enough to support applications like
video and audio streaming as well as Web browsing and email. 232 Second, unlike a circuitswitched network, but like i-mode, GPRS is supposedly “always on,” avoiding the horrible
latency – “WAP wait” – associated with surfing on 2G networks, and supporting real-time
applications like chat. Third, since GPRS is packet-switched, like i-mode, users will only
be charged for data packets actually delivered, rather than airtime. Fourth, once WAP 2.0 is
deployed, GPRS will support all the nice colors and graphics that i-mode already delivers.
Fifty, for 2G GSM networks, upgrading to GPRS is supposed to be relatively easy and
inexpensive.233 Finally, for 3G believers, the promise is that eventually there can also be a
smooth transition from GPRS to whatever variant of EDGE, wCDMA, UMTS, or
In Europe, where the GSM camp includes almost everyone, this argument has been widely
accepted. BT Cellnet launched the world’s first GPRS data network in June 2000. At last
count GPRS was being piloted by at least 59 European operators in 18 countries, including
15 of the top 20. Worldwide, there are now another 88 GPRS pilots in more than 30 other
countries – including 13 in China alone. So far, full-scale deployments have been limited by
factors like a shortage of working GPRS handsets. (See below.) Even so, and despite this
year’s depressed cell phone market, Motorola, the market leader in GPRS handsets, expects
to sell 5 million this year, and twice that many in 2002.234 This is consistent with a global
installed base for GPRS on the order of 20-30 million by the end of 2002.235 While this is a
tiny fraction of the global 400-450 million cell phone market, it is obviously a key growth
segment for the wireless data market, especially outside the US.
In the US and Canada, where until recently there were only a handful of small GSM
operators, Voicestream/Omnipoint (now owned by Deutsche Telecom) had already started
upgrading its GSM network in late 2000. But the real boost for GPRS’ US prospects came
when Cingular Wireless and AT&T Wireless announced plans to essentially double-upgrade
their TDMA networks, first to GSM, and then to GPRS, and both launched commercial
service in Seattle this summer. Depending on how their pilots go, these two carriers could
offer GPRS on most of the POPs by the end of 2002. 236
© SHG 2002
- 71
© SHG 2002
Speed Of Network
Meanwhile, the other key faction in the US cellular industry is following the Qualcommdesignated route to 2.5G, so-called “CDMA2000 1xRTT. ” This camp, which includes
Sprint PCS, Verizon, and probably Quest, Nextel, and Alltel, has been slower to do trials,
partly because CDMA2000
Chart 33.
1xRTT is a more expensive
Speed Tree to
Cellular Data upgrade than GPRS.237 Around
- The Original
the world, there are also far
2.5 G
2 mbps+++
fewer CDMA networks, though
1xRTT has already been
2 mbps
launched by SK Telcom and
1 mbps
LG Telcom in Korea,238 with
Japan’s KDDI also launching in
384 kbps
next year. In the US, Sprint
144 kbps
PCS is the 1xRTT frontrunner,
43-115 kbps
CDPD 19.2 kbps
with plans to deploy in “late
2001.” Apart from the fact that
iDEN TDMA 9.6-14.4 kbps
its upgrade costs and maximum
data rate – 144 kbps – are
slightly higher, CDMA2000
1xRTT offers the same nominal advantages as GPRS.
Source: Motorola (circa 1/2001)
C. Beyond the 2.5G Hype
At first glance, then, both these new networks appear to be formidable potential
competitors for all low-speed data networks, including ReFLEX. However, before we
concede too quickly, we need to remember our history lessons, especially the cases of imode and SMS in Chapter III, ReFLEX in Chapter IV and Mobitex in Chapter V.
As we saw there, competitive success in MDM depends less upon raw speed, cool colors,
or fancy multi-purpose designs than on a network’s ability to offer attributes that customers really
find useful. In the case of MDM, for many applications these include reliability (a joint
product of coverage, interoperability, battery life, error correction, and in-building
penetration); easy-to-use devices and applications platforms; a prolific applications
development channel; and affordable costs. From this standpoint, we believe that at last
for the foreseeable future, 2.5 G networks will have some major gaps.
“Always-On – Not.” One key implication of 2.5G’s “forked road” to 3G is that at
least in the US, there will not be a reliable, low-cost, interoperable nationwide 2.5G network
with great coverage and in-building penetration in place to compete with data-only networks
any time soon. Furthermore, it is unlikely that there will ever be a truly nationwide 2.5G
network, due to the “territorial imperatives” of these competing technologies.
© SHG 2002
- 72
© SHG 2002
Voice & Data
Dedicated Data
As shown in Chart 34, all
Chart 34. Low-Speed Data and Cellular Voice
the US cellular networks
Networks -- Coverage and Throughput
that are about to be
US Coverage
Max Speed (KBS)*
upgraded to 2.5G have
POPS Geography Now
56.0 (V.3?)
problems. To begin with,
19.2 (50% now
Cingular Wireless and
AT&T Wireless’s TDMA
networks together only
cover about 6 percent of
the country’s area and 38
percent of its population,
ReFLEX™’s 95 percent.
Source: The Strategis Group, Credit Suisse/First Boston, Bear Stearns, Motorola
Recent estimates also
indicate that to provide adequate data rates and in-building penetration, GPRS networks may
require at least 2.5 times as many base stations as GSM, especially in urban areas.239
Given the fact that they are competing vigorously with each other for voice customers (viz.
the rival pilots by Cingular and AT&T in Seattle!), it is also doubtful that these networks will
ever have roaming agreements that permit them to share network capacity and coverage,
even within the same families of cellular networks. (CDMA2000,GSM/GPRS). Nor have
these US network operators yet determined whether their handsets will be permitted to
receive messages from 2.5G-enabled PDAs that run on other people’s networks.240
Even apart from the fact that it will take time for these operators to upgrade their networks
to 2.5G, the result is likely to resemble the Balkanized situation that still plagues the US SMS
market. Given the low average density of the US market, and the increased number of base
stations required to insure reception for GPRS and especially for CDMA2000, it is also
unlikely that these networks will soon be able to match the in-building penetration of lowspeed data networks like ReFLEX. In short, chances are that for quite some time, many
suburban or rural areas will lack adequate coverage, many urban centers will lack adequate
penetration and highly variable data rates, and everyone will lack fully-interoperable data
" Birthing Pains/ Complex Devices. These are new, untested networks, and they are
experiencing many birthing problems. For example, as noted above,
Motorola Timeport
there is a shortage of affordable handsets for both 2.5G networks.241 7382i for
One problem has been that WAP 2.0, needed to match i-mode’s color
graphics and animation and support 2.5G’s higher speeds, security, and
xHTML-based applications, was only released on July 31, 2001, three
years after WAP 1.0. Another problem has been to make handsets with
acceptable battery life and “multi-mode” capability – backwards
compatibility with 2G networks. US GSM operators face special
challenges in this regard, because they run at 1900 MHz, compared
with 900 MHz/ 1800 MHz elsewhere.242 Devices designed for Europe
won’t work here unless they are specifically made with multimode capability, which is much
© SHG 2002
- 73
© SHG 2002
Motorola’s Accompli 008 for GPRS
more costly. Most 2.5G handset vendors have also felt
compelled to stuff their devices with new features like
PDA keyboards, calendaring and scheduling, color, pen
recognition systems, multimedia sound and graphics,
games, and even MP3 players, on top of regular voice
handsets.243 Packing all this functionality into integrated
mobile devices is not only difficult but expensive, so it is
not surprising that most first-generation 2.5G handsets
are likely to command at least $200-$400 retail price tags
in the US, even after carrier subsidies.244
Longer-term, the real issue is not whether or not there is any market for these conglomerate
devices. After all, even WAP 1.0 phones sold 5 million units in the US, though most of
them have probably never been used for browsing. For us the real issue is whether these
2.5G devices will command such an overwhelming share of the market that they will crowd
out lower-speed MDM devices entirely, just as 2G devices undermined the one-way paging
market. We suspect that, especially for enterprise applications, it will quite some time before
2.5G devices can match the reliability, cost, simplicity, battery life, specificity and
manageability of data-only network devices.
" Theoretical Vs. Actual Speeds. As noted above, in addition to “always on,” another
crucial part of the 2.5G value proposition is much higher data rates – the comparison is
often made to ISDN-BRI with two B channels, which delivers 128 kbps. Here again,
however, the actual average rates experienced by subscribers is likely to be much less than
the cellular operators are advertising.
In the case of GPRS, for example, the 171.2kbps data rate often cited is a theoretical
maximum for a single user who is permitted to take command of all eight timeslots on the GSM
network, without error correction. First, adding error correction, vital for wireless
communication, drops this to about 115 kbps. And no network operator in his right mind
will provide more than 1-2 slots on uplink and up to 4 on downlink to any GPRS data user,
because of the high opportunity costs of voice traffic – in fact today’s GPRS handsets don’t
even support it. That cuts the maximum to 50-60 kbps for downlinks and 15-30 kbps for
uplinks. Third, 2.5G data capacity is not dedicated, but is shared with all other users on the
network, including voice traffic. In dense urban areas, especially, data rates could vary
significantly by time of day.
So the actual data rates experienced by GPRS subscribers, especially in urban areas, is likely
to average just 10-20 kbps – not much more than is already available on a reliable, wide-area
basis from “low speed” data networks! As one recent analysis of GPRS concluded, “The
actual bandwidth is nowhere near the theoretical value…In reality, users can hardly expect
data rates greater than those provided by analog modems.” 245 Consistent with this,
operators and device vendors who are launching GPRS services and devices have been
careful to use “best efforts” language in the fine print, promising speeds “up to” maximums
in the 20 – 54 kbps range, and making no commitments about actual data rates.246
© SHG 2002
- 74
© SHG 2002
CDMA2000 networks may be less subject to congestion, but there is much less experience
with them to date. We suspect that their actual performance will also be below their
advertised 144 kbps rates.
Overall, therefore, the “higher speeds” claimed for 2.5G are just as overstated as the claims
about “always on.” At the very least, 2.5G certainly won’t open the door to a huge number
of new applications not already available from lower-speed wireless networks. On the one
hand, users won’t see much benefit at all on basic messaging applications like email, chat,
and information distribution. On the other hand, at these modest speeds, applications like
document sharing and even streaming multimedia will be painfully slow, especially to a US
audience that is increasingly accustomed to wired broadband, while downloading file
attachments, MP3 files, or movie clips will be almost unbearable. Somewhere in the middle,
perhaps, 2.5G may be compelling to cell phone users who are hungry for to do more Web
browsing on the run. But whether or not that is just an interesting little niche or a megasegment that justifies the billions operators are spending on this upgrade remains to be seen.
" Other Key Performance Issues.
There are several other performance issues that
prospective 2.5G customers also need to be keep an eye on.
Application Platforms/ Content Development. As we saw earlier, one key factor in
i-mode’s success has been its easy-to-use application development platform and its open
revenue-sharing model with respect to third-party applications. While WAP 2.0’s adoption of
xHTML is a huge step forward, and GPRS handset suppliers like Motorola are also
providing on-board support for J2ME, it is not yet clear that leading US 2.5G operators
have entirely given up on the “walled garden” approach. This issue is a by-product of the
basic fact that, unlike voice services, it is technically easy to offer data services across
networks -- and to do so without even owning a network. Is a Cingular GPRS subscriber,
for example, permitted to sign up for content from an AT&T Wireless-supported website?
Will all wireless Web developers be allowed to gain access to any carriers’ 2.5G subscribers,
for purposes of offering them new services and applications, without paying stiff fees?
These issues are important to enterprise customers as well as developers and consumers,
because they affect the overall economics of these new networks. To the extent that they
become seedbeds for an abundance of new wireless services, as opposed to semi-closed
operator fiefdoms, their chances of achieving better coverage, reliability, and costs are
" Other Missing Features. There are plethora of other technical shortcomings that
pertain to 2.5 G networks, most of which derive from the basic fact that they are semi-3G
networks trying to live in 2G bodies, with radios, channel allocations, paging mechanisms,
modulation schemes, and power controls that were designed for speech, not data. Unlike
ReFLEX™ or SMS, for example, GPRS has no message broadcasting capability, which is
essential to a whole class of information distribution and chat group applications. It also has
no native store-and-forward capability ---it has to rely on SMS to do that. Customers and
developers may be justifiably skittish about investing heavily in applications and terminals for
networks that even its strongest advocates agree should be replaced as soon as possible – if
the network equipment vendors had their way, as early as 2002. 247
© SHG 2002
- 75
© SHG 2002
" Pricing and Costs. The other key advantage that was promised for 2.5 G networks
was that services would be cheaper, because users would be charged only for data volumes,
not airtime. Indeed, the price
plans announced for GPRS by
AT&T Wireless, Cingular, and
Cingular Cingular AT&T Vodafone
Vodafone do provide for
Price per month $14.99
based on the number of Kb of
Price per incremental KB $0.07
data sent or received per
Price per incremental message $0.10
month, with some plans also
Voice minutes included
Imputed voice value
Net price per MB $29.98
messages sent.
Cost per message (100/month) $0.15
Cost per message (Heavy user)*
These initial plans are difficult
*Heavy user: averages 18.7 messages received, 8.3 sent per day, per Gallup
to compare without assuming
specific use patterns, but in
Source: corporate websites, Gallup Poll (7/2001), SHGanalysis
general they imply prices on
the order of $10 to $44 per MB of data, and $.10 to $.36 per message, depending on usage
volume. 249 (See Chart 35.) On a per MB basis this still looks pretty pricey -- even without
airtime charges, it would cost users at least $5-$12 to handle a typical day’s worth of Internet
email messages received and sent,250 or a handful of .jpg files. 251 For mobile surfers, every
page view’s worth of Web browsing costs 2-7 cents, somewhat higher than the average cost
per page view for wired analog modem surfing, but not off the mark, considering the value
of mobility. 252 However, it is still not clear how much surfing US mobile users will really
want to do, even at these improved prices, given the continuing dearth of compelling imode-like mobile content in the US. And users who really want to download music videos
to their MP3-ready mobile devices at affordable prices may just have to wait for 3G’s vast
new networks. Of course by then they may no longer be teenagers.
As noted in the charts in Appendix VI, these price levels per MB are also still well above the
marginal costs of ReFLEX™ and other low-speed networks. While some analysts have
recently argued that in some long run, all networks will compete with each other on a priceper – MB basis, we fundamentally disagree. For the foreseeable future, especially for
messaging customers, the fact is that there are simply many other elements of value beside
raw data throughput. Nevertheless, even on this basis, the per-MB prices used by the 2.5.
G operators leaves plenty of room for low-speed networks like ReFLEX to compete. 253
More important for MDM customers are average and marginal message prices. The unit
prices implied by the new 2.5 G pricing plans vary greatly with customer usage , but in
general, the plans are a marked improvement over circuit-switched data. However, they are
still are struggling to get unit message prices below 10 cents, and are often several times
that. 254 At least for these initial plans, given the fact that two-way data networks like
ReFLEX have much less message overhead, they can easily match these price levels,
especially given their capacity increases. 255
Overall, therefore, at least for the 2.5G pricing plans announced so far, it appears that the
main pricing benefits will be realized by mobile surfers, not MDM users. And even those
© SHG 2002
- 76
© SHG 2002
benefits depend on US operators adopting DoCoMo’s successful model for content
Of course for MDM customers the real issue is not pricing, but the total cost of ownership per
application, relative to application performance. Earlier we saw that many of the performance
advantages claimed for 2.5G networks and devices were spurious, and that these networks
still can’t match the coverage, in-building penetration, and many other key features of lowspeed networks like ReFLEX. We’ve also just seen that the average and marginal prices
per message of 2.5G networks can be readily matched by low-speed networks. But even
beyond service pricing, networks like ReFLEX can also provide lower device costs, lower
application costs for specific applications like field force management, information
distribution, and enterprise email, and much more manageable ownership costs. This is
especially true in the US, where enterprise customers should ponder carefully the
implications of issuing expensive 2.5G voice/data handsets to employees without “calling
party pays,” on the one hand, but with strict employer liability for accidents caused by
employees who are talking on company cell phones while driving, on the other.256
Telcos…Will Be Telcos. The sections above argued that in terms of
price/performance, low-speed networks are likely to maintain an advantage over 2.5G for
many MDM applications in purely technical and economic terms. However, there is another
very important non-technical reason why enterprise customers, in particular, might favor
solutions from low-speed data-only network providers. This is the fact that the leading
players in the US cellular voice/ 2.5G industry – especially the top four operators, Verizon,
Cingular, AT&T Wireless, and Sprint PCS – all come out of a regulated telephone
monopoly background. They are still struggling with the bad habits that it nurtured. In
particular, as one recent review of leading US wireless carriers concluded, they all “have a
long way to go to reach even a basic level of customer satisfaction….they can’t handle basic
things like service phone calls, billing and sales.”257 Another recent report on customer
satisfaction at Verizon and Cingular found that fully one-third of their wireless customers
were dissatisfied.258 As Forbes Magazine concluded only this month, “A confluence of
factors has conspired to create a business that is infamous for shoddy service, poor coverage,
and outright hostility toward its customers.”259
While in theory, improvements in customer satisfaction might be more easily achieved than
fundamental changes in network attributes, in practice organizations that have developed
bad bureaucratic habits over many decades usually take decades to change. The fact that
particular 2.5G networks will only be offered by one or two of these carriers in many US
markets for quite some time will only reinforce this behavior. Enterprise customers, in
particular, should be cautious about the wisdom of trusting their mission-critical wireless or
wired applications and services to companies that are still – with the possible exception of
Cingular – largely focused on voice services.
© SHG 2002
- 77
© SHG 2002
8. Summary – Comparative Advantages
A. Near-Term Competitors
The tale is told about the two hunters who encountered a hungry grizzly in the wild northern
woods, and were soon running for their lives. Just as they were about to take off, one turned
to the other and said, “How will we ever make it? Neither of us is fast enough to outrun
that bear! ” The other replied, “I don’t have to outrun the bear. I just have to outrun you.”
One key theme of this paper is that the most important competitor for low-speed networks
is not really 2.5G, much less 3G. As we began to see in Chapter III, the entire 3G vision
turns out to questionable, especially in the US. Indeed, it may be on the verge of becoming
the wireless equivalent of High-Definition Television (HDTV) – a technology that is
perpetually just around the corner, with no one quite sure what its value is, even though it
would cost a fortune to launch, including the costs of replacing all existing terminals. In 3G’s
case the situation is even worse, not only because it is vastly more expensive, but because the
technology will just not sit still – for example, wCDMA’s technical specification has
changed 240 times in the last two years!260 No wonder that aside from NTT DoCoMo,
service providers that had previously signed up to launch 3G services are now announcing
delays on a regular basis.261
Upon close inspection, as we’ve seen, the threat from 2.5G networks also turns out to be
overstated, especially for enterprises that demand reliable, affordable service – and which
enterprises do not? Our analysis of these purportedly “faster, always-on” networks showed
that they are neither that much faster in practice, nor anywhere near as “always on ” as
almost any one of today’s proven low-speed data networks. Nor are they less expensive.
More fundamentally, we’ve raised also serious questions about precisely what the “need for
speed” really is in the first place, especially for enterprise applications. Is it just about faster
surfing and multimedia downloads for cell phone users? What enterprises in their right minds
want to subsidize that – especially at the cost of poorer coverage and reliability?
Overall, we are deeply skeptical about what has become the central value proposition
behind the $350 billion+ 2.5 G and 3G cellular network and handset upgrades. For almost
all mission-critical MDM applications that we can think of, the fact is that these upgrades
will provide virtually no discernable improvements in application performance. Indeed, to
the extent that enterprises are seduced to adopt the data solutions promoted by the cellular
voice industry, actual MDM application performance is likely to suffer, even while the total
costs of ownership soars.
These doubts are consistent with a more general skepticism about the value of “broadband
services” that is just now becoming visible in the wired world, as well.262 As one analyst
recently put it, speaking about the demand for high-speed Internet services provided by
© SHG 2002
- 78
© SHG 2002
modems and DSL, “It hasn’t yet been proven that broadband is an essential service.”263 In
the computer world, too, much of the recent slowdown in worldwide PC sales may be due
to the fact that many customers – especially enterprise customers -- simply don’t see the
compelling need, from an applications standpoint, to upgrade their 3-4 year-old PCs to the
latest 1.5-2.0 GHz models.264 For enterprise MDM applications, where higher data rates are
often actually associated – as we’ve seen -- with a deterioration in service quality and coverage,
the case for skepticism about “speed fetishism” is even stronger.
For the immediate future, then, the two key competitor for enterprise MDM applications in
the US are likely to remain Reflex™ and Mobitex. Here, as we’ve shown, ReFLEX starts
with many strengths, and this year’s deployment of Version 2.7 by Skytel and Arch, plus
other new technical capabilities, could overcome almost all its relative disadvantages. As its
service providers proceed to roll out V.2.7, new devices, and other capabilities over the next
few months, we believe that enterprise customers and solutions providers who are seriously
contemplating the deployment of robust MDM solutions should carefully consider both
ReFLEX and Mobitex™ alternatives.
B. Longer Term Prospects – Device and Network Independence
Longer term – say, at least 5-10 years – there may indeed come a point when all the billions
that have been spent on these new cellular networks and the “forced upgrades” of hundreds
of millions of users to costly new integrated voice/data handsets finally yield pervasive
higher-speed cellular networks that offer low marginal costs, good coverage, high network
capacity, and perhaps even decent customer service. We don’t believe that these
investments are profitable ex ante for society. But this will certainly not be the first case
where a powerful global industry has been able to marshal hundreds of billions for
investments that turned out to be dubious and perhaps even unsafe.265
Eventually, therefore, there may well be more powerful high-speed networks and device
alternatives available for MDM applications. Even then, however, this won’t necessarily spell
the end for low-speed data networks like Mobitex™ and ReFLEX -- in fact, just the
Right now, if one wants to send an email to someone else’s wireless device, he actually has
to know which network it runs on, to address it – for example, [email protected], or
[email protected] Nor is there any easy way for recipients with multiple locations and
devices to forward messages on the fly to their current preferred device. At one point in the
day we might prefer to have all messages directed to our two-way pagers, because we’re
locked in a basement conference room; at others we might want them sent to our cell
phones, because our response demands a voice call. As we’ve seen, the proliferation of
wireless networks and devices is likely to continue for the foreseeable future, so this
problems will only get worse.
Fortunately. new software solutions are already becoming available to solve these problems,
providing unified addressing and device- and network-independence.266From the customer’s
standpoint this means that one no longer has to worry about stitching together connections
among all one’s various PDAs, PCs, and cell phones that run on multiple networks, in order
© SHG 2002
- 79
© SHG 2002
to get messages on the preferred device at hand. If I want to contact Sally, I just send a chat,
email or voicemail to her single address, and the system figures out how to get it to whatever
devices or workstations she wants it sent to at the moment, in whatever forms are feasible
on those devices – text, speech, speech-to-text, text-to-speech, real-time chat, images, or
even video.
This “device/network arbitrage” model of wireless services, in turn, will permit all these
various devices and networks to work together, specializing in what each of them does
best. Low-speed networks, for example, are likely to continue to provide superior reliability
and cost for quite some time; on the other hand, they were never meant to support –
VoiceNow™ aside -- voice traffic, browsing or multimedia downloads. The existence of
these new user-oriented service platforms means that their future is not an existence
problem, but essentially a pricing problem, a matter of sorting out what they do best and
choosing the appropriate resource allocations.
9. Conclusion
The objective of this white paper was not to review the business strategies or financial
prospects of leading low-speed network service providers like Arch Wireless, Skytel/MCI, or
Cingular Wireless. Obviously they have their work cut out for them. They must restructure
the one-way paging industry’s debts, aggressively invest in and promote their new network
capabilities, develop new channel partners for wireless devices, applications and solutions,
recruit new enterprise customers, and work much more effectively together. This will not be
easy, especially given the current economic environment.
But it is eminently doable. Assuming that the low-speed network industry can restructure its
debts and survive, we believe that two-way data networks like ReFLEX and Mobitex™
should actually have quite a bright future. Indeed, they could take the lead in introducing the
“software-defined” user-oriented MDM services described above to enterprise customer
market. That would let their operators specialize in what they do best – reliable, low-cost,
ubiquitous, if “slow,” MDM services. If they do that, like the copper cable network or the
DC-3 before them, they should be around for a very long time to come.
© SHG 2002
- 80
© SHG 2002
© SHG 2002
Appendix A: Key Technical Features,
ReFLEX Version 2.7
- 81
© SHG 2002
1. Introduction
The following Appendix takes a closer look at the most important new features in Version
2.7 of the ReFLEX protocol, including background scanning, auto-collapse (“Chat
Mode”), flexible control of the maximum inbound message length, unscheduled inbound
messaging, and the harmonization of ReFLEX variants. It also briefly examines the
efforts of the WCTP consortium to develop a new XML-based Internet gateway standard
for wireless messaging.
In examining the benefits of these new features, it is important to understand that there is
not a simple one-for-one alignment between the new features and their benefits. As
summarized in Chart 15 above, several new features affect more than one benefit, like
increased capacity or lower latency, while several benefits are the product of more than one
new feature. The overall result is truly a case of “the whole being greater than the sum of its
2. Background Scanning
One key driving force behind ReFLEX, as noted in Chapter IV, is the capacity
enhancements achieved through background scanning.
Mobile devices would start searching for a new channel only when they had completely lost
touch with the current control, in previous version of ReFLEX, even if other channels
had stronger signals. This meant that there was significant time between losing one channel
and acquiring a new channel, called "sub-zone drag", during which time the device would
not receive messages. This behavior means that the usual strategy of increasing network
capacity by using smaller and smaller sub-zones has a negative impact on the subscriber.
In Version 2.7, devices periodically scan for neighboring control channels in the background,
without interrupting normal operations. If the device finds a better channel, in terms of
significantly better signal strength or higher priority, it can request a transfer. This is usually
done using “make before break”, a concept similar to the soft hand-off used in PCS Phone
networks, where registration with a new channel is completed before communication with
the old channel is broken. Normally, this means that a device will always be registered with
the network, and capable of receiving messages.267 This permits mobile devices to move
quickly and efficiently across service areas with different control channels.
This change alone has a profound impact on ReFLEX networks by allowing cellular-like
functionality, while retaining the superior coverage and reliability offered by simulcast and
Sub-Zoning. Currently, ReFLEX operators have divided their networks into zones that
consist of many transmitters and receivers. Background scanning allows the operators,
through simple software reconfiguration, to sub-divide these zones and achieving significant
© SHG 2002
- 82
© SHG 2002
frequency re-use, multiplying network capacities several-fold. A sub-zone can be as small as
a single transmitter and receiver268, but there is a limit to the effectiveness of sub-zoning in
wide area coverage. To maintain the superior coverage advantages of simulcasting and
macrodiversity, primary zones for public networks typically include the number of
transmitters required to cover significant metropolitan geographic boundaries.269 There are
cases for secondary zones that benefit from being as small as one transmitter site. This will
allow ReFLEX operators who adopt Version 2.7 to multiply their national network
capacities more than three times their current limitations, with very little additional capital
“Hot Spot” Capacity Increases. Background scanning also helps ReFLEX operators to add
network capacity precisely where it is needed. ReFLEX has long had the ability to handle
multiple outbound and inbound channels within a single sub-zone. Version 2.7 enables
ReFLEX operators to add additional outbound and inbound channels only to those subzones where the traffic load warrants, and balance devices and traffic across those channels.
For carriers using linear transmitters and controllers, additional outbound channels can be
added with minimal capital expenditures. Operators can even overlay additional sub-zones
that completely overlap existing geographical coverage.
Campus Coverage. The background scanning feature of the ReFLEX protocol allows
operators to create special, dedicated networks to cover specific areas, such as corporate or
academic campuses, amusement parks or ski slopes. Devices can be programmed to register
when they are in range of this special, private zone, and utilize a public network otherwise.
The resulting private network, or campus, consisting of both outbound and inbound
channel(s), permits only authorized users to register. In fact, devices not associated with the
private network will not even recognize the campus exists. The campus can transmit
specialized private information relevant only to the specific private network without creating
any traffic on the wide area public network. At the same time, it can provide the private
network subscribers with the same coverage enjoyed by the subscribers of the wide area
The 'background scanning' enhancement to Version 2.7 of the ReFLEX protocol is one of
its most important features. It allows the network operators to make dramatic increases in
network capacity with minimal additional investment, while maintaining a high level of
customer service. This translates into lower cost and better performance for subscribers.
3. Auto-Collapse – “Chat Mode”
ReFLEX was initially designed to deliver low-cost mobile data, using small, inexpensive
devices that worked continuously for weeks on a single battery. One key design element
employed to achieve this is called “collapse.”270 Collapse provides the ability for the system
to let a device 'sleep' for periods of time. A collapse value of two, for example, allows a
receiver to be “available”, listening for its address, for only a quarter of the normal time. The
device saves energy by sleeping through the remainder of the time.
© SHG 2002
- 83
© SHG 2002
The trade-off made for the sake of battery saving was increased message latency, since the
device would only receive messages every 7.5 seconds (= 4 x 1.875 seconds, the standard
ReFLEX frame length), using a collapse value of two. If a device needed to send a
message, it requested a time allocation, and then waited for the next scheduled time slot to
wake up and receive its transmit time information. Then the receiver would fall asleep again
until its next scheduled wake-up, when it would receive a message acknowledgement. This
implies a time lag of more than 15 seconds, for a collapse value of two, from the time the
user presses “send” until the acknowledgment receipt. For higher collapse values (used to
obtain even longer battery life), the perceived wait could be much longer, on the order of 3060 seconds or more.
In ReFLEX version 2.7, these long latencies no longer exist because of a new feature
called “auto-collapse.” In auto-collapse mode, the device receiver stays awake after most
messaging events, for a device specified time, to look for any further messaging data. This is
typically 30 seconds. Compared with the example above, this could allow an
acknowledgement to be received just 5.625 seconds after “send” is pushed, regardless of the
collapse value. This reduces latency by about two-thirds, and even more for higher collapse
Additionally, Version 2.7 enables users, or applications, to initiate a “chat mode,” in which
the device automatically wakes up in every frame to look for messages, for four minutes. In
this mode, messages can theoretically be sent in every frame, with latency between devices
reduced to less than five seconds.271 In practice, perceived latency on public networks will
probably be a little longer. The reduced message latency of 'chat mode' on public networks
will be similar to that now achieved by Mobitex or Datatac networks. When combining this
feature with background scanning, which allows private networks, the latency will actually be
faster than what is now achieved with other mobile data networks, often achieving the
aforementioned 5.625 seconds.
This new “chat mode” could be very important for ReFLEX’s future. Not only does it
improve the performance of existing applications by greatly reducing latency associated with
sending longer messages, but it also allows new real-time-critical applications to be deployed
on ReFLEX networks, including instant messaging, wireless POS, and financial
4. Broadcasting Maximum Inbound Message Lengths
In Version 2.7, the network broadcasts, every minute, the maximum inbound message length
that it will accept. This feature can be dynamically set for each zone and control channel,
meaning that ReFLEX operators can manage inbound traffic to avoid network congestion
during peak usage times, while allowing larger, more rapid transfers when the network is
lightly loaded. This new more efficient way of managing network traffic will also help to
reduce perceived latency, and increase the effective network capacity.
© SHG 2002
- 84
© SHG 2002
This feature also has the added benefit of relieving device manufacturers of the need to
program fixed message length limits into their devices, and the burden of managing varying
device configurations among the carriers. Manufacturers can now more easily make devices
that will work across all ReFLEX networks.
5. Unscheduled Inbound Messaging
In a Version 2.6 ReFLEX network, inbound channels are divided into two separate logical
channels: (1) a “control” channel, randomly accessed using shared ALOHA slots, and (2)
the “data” channel, which is allocated by the network to specific devices that request
inbound capacity. The conventional procedure for device-initiated inbound messaging is as
The device sends an allocation request over the control channel using a shared
ALOHA slot;
The network responds by sending a command to the device indicating which time
slots in the “data channel” it has been allocated;
The device sends the inbound message in the allocated timeslots in the "data
This scheduling mechanism has tremendous advantages in terms of system capacity. The
scheduled data channel has virtually no collisions, and can be loaded to more than 80%
before any significant congestion is detectable. By comparison, a randomly accessed
ALOHA only network, similar to TCP/IP, can be no more than 30 percent utilized before
significant delays and congestion become apparent. ReFLEX networks utilize both
protocol types to improve both capacity and availability for the network. The cost of this
extra capacity is a small additional message latency, due to the time required for the
allocation of timeslots for transmission of messages in the scheduled data channel.
ReFLEX Version 2.7272 permits what are called ALOHA inbound messages, where in the
case of short inbound messages of less than 223 bytes, the device can access this channel
and immediately send, without having to wait for timeslot allocation. This further reduces
latency at a very small cost to overall network capacity. This feature works in combination
with the inherent ALOHA reuse of ReFLEX networks. ALOHA reuse allows messages
from different receivers in the same sub-zone, to be accepted when they arrive at that same
time, if they are uncorrupted.
© SHG 2002
- 85
© SHG 2002
VII. Appendix B: Detailed Comparisons,
Key Data Networks
© SHG 2002
- 86
© SHG 2002
1. Network Comparison Charts
Chart 26. Summary, Mobitex™vs. ReFLEX™
A. Technical “Speeds and Feeds”
Key Attributes
Mobitex™ V.2
ReFLEX™ 2.7
896-902 MHz TX
935-941 RX
896-902 MHz
929-932, 940-941 MHz
US Spectrum Allocated
500-750 KHz (depending on
>2 MHz
ReFLEX advantage -- capacity,
pricing potential
Maximum data speed,
single user
8.0 kbps
1.6-6.4 kbps down,
.8- 9.6 kbps up
ReFLEX™ 2.7 unifies platform
Mobitex v.3 -->32 Kbps
ReFLEX ™ v. 3.0 --> 56 kbps
1.2-2 kbps
3.2 kbps down
4.8 kbps up
ReFLEX™ adv. -- more efficient
12.5 KHz
12.5 Khz in; 10Khz or 12.5 Khz
R v 2.7 allows channel subdiv.
And reuse
TDMA (reverse channel)
“TDMA-like” (forward ch.)
Dynamic S-Aloha
Aloha + scheduling (see App
Frequency band
Typical throughput, single
Symmetrical data rates?
RF Channel spacing
Channel Access
Multi-user access
Full or half duplex
Packet length
© SHG 2001
Modulation Method
Native Protocols
Other transport protocols
supported (gateways)
© SHG 2002
Up to 512 bytes (MPAKs)
924 bytes
Proprietary --- gateway
support for TCP/IP
SNA, X.25, MTP/1, x.24 CCITT,
ReFLEX™ adv.
Neither support native TCP/IP
- 87
© SHG 2002
Chart 26. Summary, Mobitex™vs. ReFLEX™
B. Other Technical Performance Factors
Key Attributes
Reliable Reachability
-- Geographic coverage
-- Store-and-forward
Very Good
High for metro enterprises/
72% US pop
Very high - 95% US pop
Overall ReFLEX™adv.
Big ReFLEX™ adv.
Big ReFLEX™ adv., most areas
Simulcast, Macrodiversity
Best on latency
Best on broadcast
Yes (except Palm VII)
- -Broadcast/ simulcast
Big ReFLEX™ adv.
-- Latency (min. / typical
<5 - 15 sec.
<5-15 sec
Slight security adv.
Slight battery life adv.
Roughly equivalent
Sleep cycle
Very high - sleep cycle,
ReFLEX™ adv.
Single AA, 1-2 weeks
Single AA, 2-4 weeks
ReFLEX™ adv.
-- In-Building penetration
-- Other reliability
Messaging modes
-- “Always on”/ “push”
Other technical factors
--Battery saving
-- Typical battery life
© SHG 2001
Mobitex™ V.2
--Data compression
-- Native security
--End to end encryption?
© SHG 2002
ReFLEX™ adv.
Application dependent
Slight Mobitex™ adv.
Via 3rd party apps
Via 3rd party apps
Both can do if needed
- 88
© SHG 2002
Chart 26. Summary, Mobitex™vs ReFLEX™
C. Economic and Industry Factors
Key Attributes
Mobitex™ V.2
Device Economics
-- Device Selection
-- Device Cost
RIM 950/ 957, Palm VII
Advantra, Nixxo, Fine Telecom,
Glenayre(2.6), Mot (2.6),
Novatel (Palm cradle)
M advantage may be shrinking
Good (J2ME - Rim, Palm)
Very high - Ericsson pricing,
cellular structure
Low, for upgrades to FLEX™
networks -- high elsewhere
- Marginal Cost, New Capacity
Low- Mod (Cellular)
Higher before 2.7, similar with it
Marginal cost, Current Capacity
High- capacity constrained
“Zero” with 2.7 (up to 3-4 x
subscriber base)
Mod (trunking network)
Low- Moderate
Multiple 3rd party vendors
WCTP, Agea, Outr.Net
-- Proven Applications
Limited so far
M advantage - key R need
--- Developer Channel
Extensive (Aether, etc.)
WCTP should help
M advantage - key R need
RIM, Palm, Handspring, Nomadic,
Motorola (2.6), Advantra, Fine,
Nixxo, Glenayre
Sonik, TGA
-- Device Capability
Network Economics
-- Fixed Costs, Coverage
-- Operating Costs
-- Application/ Middleware
Development Tools
-- Network Vendor Support
-- Service providers
-- Installed Base/ Growth
© SHG 2002
R catching up
Only relevant to new networks
R2.7 narrows M lead
Tremendous R advantage
M Lead, R potential
Other Industry Factors
-- Device Vendor Support
R advantage
R strong cost advantage
Application Economics
© SHG 2001
M lead shrinking
M advantage -- R catching up
M lead/ R potential catchup
1 (US), 28 non US
3 (US), 5 non US
+.690mm, US - 45%/yr
+.400mm, non US
1.5 mm, US --+90%/yr
R playing catch up
Big M lead
R consolidation issues
Strong R. lead in US
- 89
© SHG 2002
Chart 27. Summary, Other Key Data Networks
A. Technical “Speeds and Feeds”
Key Attributes
Frequency band
US Spectrum Allocated
Maximum data speed, single
Typical throughput, single user
824-849 MHz TX
868-894.4 MHz RX
(AMPS networks)
855.8375 MHz
1900 MHz (GSM
PCS networks, US)
1900 MHz (CDMA
PCS Networks, US)
NA (shared with AMPS)
125-600 KHz
(varies by
NA (shared w PCS)
Na (shared with
CDMA cellular)
19.2 kbps (shared)
4.8 - 19.2 kbps,
depending on
38 -115 (shared)
0-9.6 (variable)
2.4 -8 kbps (4.39
kbps ave.)
7.2-14.4 kbps
60-72 kbps
RF Channel spacing
30 KHz
25 KHz
1.25 Mhz
Channel Access
Multi-user access
Full or half duplex
Symmetrical data rates?
24-928 (128 ave.)
Up to 256
Modulation Method
Native Protocols
(native Control
Protocol 1.2)
Other transport protocols
supported (gateways)
TCP/IP, multiple
© SHG 2001
Packet length
© SHG 2002
- 90
© SHG 2002
Chart 27. Summary, Other Key Data Networks
B. Other Technical Performance Factors
Key Attributes
Poorest choice
Spotty (55% of US) - 25
top MSAs
No interoperability
among 7 CDPD
Mod (“>427
SMSA,s, 79% of
US pop, >90% of
US businesses”)
>50% have 19.2k
43% US Pop TDMA/GSM base
National rollout by
AWS, 2001-2?
Sprint PCS, Verizon
deploying 2002?
US Pop 63% (CDMA)
-- Store-and-forward
-- In-Building penetration
Mod (same or
better than Mob)
None/ lots of delays in
-- “Always on”/ “push”
Yes (where
Yes (where coverage)
- -Broadcast/ simulcast
IP multicasting
-- Latency (min. / typical
Varies widely (<6
sec/ 20 sec)
<1 sec
Reliable Reachability
-- Geographic coverage
-- Other reliability
Messaging modes
© SHG 2001
Other technical factors
--Battery saving
Sleep cycle
Sleep cycle
Sleep cycle
-- Typical battery life
Single AA, <1 week
(poor R850 batt
--Data compression
-- Native security
--End to end encryption?
© SHG 2002
Very poor (bbone)
Yes (3rd party)
Yes (3rd party)
Yes (3rd party)
Yes(3rd party)
- 91
© SHG 2002
Chart 27. Summary, Other Key Data Networks
C. Economic and Industry Factors
Key Attributes
Cradles (Novatel,
Raven, Airlink)
RIM 850, Palm (cradel
for Palm V)
Shortage of devices that
work (Motorola)
Device Economics
-- Device Selection
Early - shortage of
devices that work
High, initially
Not quite ready
Low -- add on to
existing AMPS network
Higher than ReFLEX™ /
but no networks
Low cost upgrade to
GSM networks, but “opp
- Marginal Cost, New Capacity
Higher than ReFLEX™
Low - but opp cost
Marginal cost, Current Capacity
High - “Capacity Did
Prove Deficient”
Limited experience
Limited experience
IBM , Aether middleware
-- Proven Applications
IBM field sales (13k)
--- Developer Channel
Aether, ISPs, IBM
-- Device/ Vendor Support
RIM , WaveTek
Motorola, Ericsson
Kyocera, Ericsson,
Motorola, etc.
-- Network Vendor Support
CDPD Forum
Vert. Integrated ?
Lucent, Nortel,
Qualcomm, Nortel,
-- Service providers, channel
AWE, Verizon, 5
Motient (US); Aether,
Metrocall, Skytel,
AWE, Voicestream (US),
many Euro. GSM
-- Installed Base
.146 K (US only
.272mm (US), Canada
7 others (was 17)
-- Device Cost
-- Device Capability
High, initially
Not quite ready
Network Economics
-- Fixed Costs, Coverage
-- Operating Costs
Low cost upgrade
to CDMAOne nets
Very low
Application Economics
-- Application/ Middleware
Development Tools
© SHG 2001
Other Industry Factors
© SHG 2002
Sprint PCS,
Verizon, Quest
- 92
© SHG 2002
2. Devices Comparison, ReFLEX™, Mobitex™ and DataTAC™, 2001
Appendix Table 1. Two-Way Network Devices Comparison
© SH G 2001
Existing R eFLEX D evices
A dvantra
Tim eportP935
A R 1800
© SHG 2001
N ew R eFLEX D evices
G lenayre
Sled for Palm
A ctiveLink Fine Telecom
N ixxo
m 100
M obitex,D atatac D evices
R IM 857™ and
R IM 850™ and
R IM 950™
R IM 957™
M odel
N ow
N ow
N ow
N ow
Q 2 '02
Q 1 '02
Q 2 '02
N ow
N ow
R etailPrice
$100 -$125
R 2.6
D im ensions
W eight
R 2.6
R 2.7
R 2.6
R 2.7
R 2.7
7.9 C M x
3.189 X 2.146 3.75 X 2.85 X 1.2
8cm x 5.5cm x 2.953 X 2.559 X
5.4C M x 2.2 3.8 x 2.3 x 1.5
X 0.902 inches
0.866 inches
3.86 ounces
6.7 ounces
3.5 oz.
110 gr
110 g
Battery Type
1 AA Alkaline
1 N iM H
1 AA
2 AA
R echargeable
560m Ah)
Battery Life
3 w eeks
O ver1 w eek
2 m os
O ver7days
31 keys
49 keys + N avD isc
3 w eeks
StylusAttaches to
H andspring
33 + 4 Keys
EL Backlighting
R 2.7
Palm m 100
800 M H z
800 M H z
D ataTAC ®
D ataTAC ®
netw orks (R IM 850) netw orks (R IM
900 M H z M obitex
netw orks (R IM 950) 900 M H z M obitex
2.5 x 3.5 x 0.93
inches (LxW xD )
4.6 x 3.1 x 0.70
inches (LxW xD )
Palm m 100
< 5 oz.
3.6v Lithium O ne AA alkaline or
Ion orLithium
rechargeable AA
Polym er
N iM H (R IM 850)
O ne AA alkaline
7 days
3 w eeks
5.3 oz
1 w eek
O ptim ized
O ptim ized keyboard
keyboard +Thum b+Thum b-operated
trackw heel
trackw heel
O ptim ax EL
4 lines by 20 EL Backlighting
Electra Light
EL Backlighting 4 lines x 20
m onochrom e U ser-selectable 6 or U ser-selectable 16
D isplay Screen
D isplay
characters 7 Line X 23C har.
8 line display
or20 line display
9 Lines X 29C har.
display w ith
4 Line X
W ith backlight 160 * 120 pixels
back light.
20C har.
pixels.Palm V
G raphic
N ow
M otorola
M otorola
M icroprocessor
D ragonball???
PD 703017 (16M IP D ragonballVZ
(?? M H z)
(33M H z)
Program m able
N /A
W isdom O S 4.0
Palm O S 3.5
N /A
N /A
Palm O S 4.0
R IM Proprietary
R IM Proprietary
O TA Function
8 m egabyte
4 M B flash m em ory
5 M B flash
R AM w ith 8
TotalM em ory
4.5M B
8 m eg
plus 512 Kbytes
m em ory plus 512
m egabyte
Kbytes SR AM
flash m em ory
M essaging
2M B
8 m eg
M em ory for
M essaging +
M essaging +
M essaging +
Integrated PIM
Palm PIM
M essaging
M essaging
em ail/organizer
em ail/organizer
M essaging
softw are
softw are
(Address Book,
C alendar,etc.)
book, alendar
Tone,vibrate,onTone,vibrate oronscreen orLED
N otification
Vibrate orAudible
Vibrate oron-scre
i di
© SHG 2002
- 93
© SHG 2002
3. Network Cost Comparisons
Appendix Table 2. Comparative Network Capital Costs -- Coverage, Capacity Expansions, and Opportunity Costs
© SHG 2001
ReFlex 25
Coverage Cost per Cell or Sub-Zone Extension
City Center with in Building Coverage
Coverage Radius in kilometers
Area of Cell
Cost per Sq. Km.
Suburban with in Building Coverage
Coverage Radius in kilometers
Area of Cell
Cost per Sq. Km.
Suburban and Rural with Street Level Coverage
Coverage Radius in kilometers
Area of Cell
Cost per Sq. Km.
Mobitex 900
Datatac 4000
CDPD 800 MHz.
GSM GPRS 900 CDMA 2000
800 MHz.
Captital cost $/Additional kb/s Information Capacity
Busy Hour Capacity in kBytes per Base Station
Capital Cost per Kbyte per Hour
Cost per Megabyte Delivered
Costs for Adding Traffic Capacity
Cost of Additional Equipment at existing location(s)
Portion of capacity used for data
Costs of Equipment used at one Base Station
Cost of Equipment to Add Capacity
Capacity Added kb/s
Usable Information Capacity Added kb/s
y Cost per Megabyte Delivered for Voice/Data Networks
Voice channels used for data capability
Minutes of voice revenue lost per Megabyte delivered
Estimated net revenue per minute
Opportunity cost per Megabyte delivered
© SHG 2002
- 94
© SHG 2002
4. Applications Fit – Technology Requirements per Application Type (Illustrative)
Appendix Table 3. “Applications Fit” -Technical Requirements Per Application
Application Requirements
Point of
Field Force
Latency requirement: maximum
tolerable, typical
Throughput per active user required in
the network busy hour (kiloBytes)
either direction [assumptions]
< 0.01 kB
< 0.01 kB
1.8 kB
[6 x 0.3 kB]
1.8 kB
[3 x 0.6 kB]
1 kB
[4 X 1.0 kB]
200 kB
10,000 kB
Area coverage
Building penetration
Battery conservation
[depends on
Always on
Portability and Battery Life
Is location information required.
Does the terminal need to be hidden
10. Does QoS have to be controlled?
© SHG 2002
- 95
© SHG 2002
VIII. Appendix C:
Glossary of Technical Terms273
© SHG 2002
- 96
© SHG 2002
An address or username linked to a person or subscriber.
Analog refers to a representation of a quantity that varies over any
continuous range of values. Analog signals can be thought of as
pure in nature and not processed. Values are exact, but error
correction is not easy.
After French engineeer Jean-Maurice-Emile Baudot, (1845-1903),
who did pioneering work on early teleprinters. Initially used to
measure the transmission speed of telegraph, the baud rate is used
today to measure a data transmission speed with a modem. The
number of voltage or frequency transitions per second. At low
speeds only, baud may be equal to bits per second. The measure of
how frequently sound changes on a phone line. This used to be the
measure of speed of modems because they worked by brute force
and actually made a sound for each bit of information. Now,
modems work on a more sophisticated level. A 14.4 Kbps modem
actually uses 2400 baud, but can transmit 14.4 Kbps.
Bit Rate
The bits per second used to encode audio data in an MP3 or other
compressed audio file, in kilobits per second (kbps). Higher bit
rates typically mean better sound quality. Typically, bit rates range
between 96-256, but any rate is possible
This refers to the transfer of multiple signals over a single medium.
In slang terms, it is any Internet connection that allows for higher
transfer speeds than an analog modem, most often applied to cable
modem access. However, it is sometimes used to refer to DSL,
satellite and wireless Internet services as well.
© SHG 2002
8 bits. Think of it as a string of 1s and 0s that represents a number
from 0 to 255. For example '01100101' is one byte of information.
A measure of how fast some device communicates, usually in
thousands of bytes per second (KBps) or millions of bytes per
second (MBps). See also bits per second. With a capital B, you are
talking Bytes, which is equal to bits * 8.
Code-Division Multiple Access. Digital cellular technology using
spread-spectrum techniques. Unlike GSM and TDMA, CDMA does
not assign a specific frequency to each user. Instead, every channel
is more efficient using the full available spectrum. This is a 2G
digital wireless technology that allows multiple calls to share a radio
- 97
© SHG 2002
frequency 1.23MHz wide in the 800MHz - 1.9GHz band without
causing interference. This is accomplished by assigning each call a
unique code and varying its signal by that code to allow only the
caller and receiver with that code to communicate with each other.
The original CDMA standard allows transmission of up to 14.4 Kbps
per channel, with up to 8 channels being able to be utilized at once
for 115 Kbps speeds. Popular alternative definitions: “ Calls
Dropped Most Anywhere,” and “Customers Don’t Mean Anything.”
A multiplexed version of the IMT-2000 standard developed by the
ITU, and is part of 3G wireless technology. It increases wireless
data transmission speeds of the original CDMA standard to 144
Kbps using a single channel and 2Mbps using 16 channels.
Cellular Digital Packet Data. Data transmission technology
developed for use on cellular phone frequencies. CDPD uses
cellular channels (in the 800- to 900-MHz range) to transmit data in
packets, with data transfer rates of up to 19.2Kbps. Popular
alternative definition – “Capacity Did Prove Deficient.”
Cellular phone
A mobile, wireless telephone that communicates with a local
transmitter using a short-wave analog or digital transmission.
Cellular phone coverage is limited to areas where a cellular phone
can adequately communicate with a nearby transmission tower.
Chat Window
Window in which a person can enter a virtual room to participate in
a chat session. Technically, a chat room is really a channel, but the
term room is used to promote the chat metaphor.
Digital Phones
Phones using digital wireless service, as opposed to analog service.
Digital service offers improved quality, privacy, and additional voice
and data features; furthermore the efficiency of digital technology
means that digital service is often less expensive than analog
Electronic Mail. The transmission of messages over a
communication network. Messages are most often text notes, but
also may include file attachments. Most computer networks have email systems, but some are confined to a single system or network.
Many systems have gateways to other computer systems and the
Internet, enabling users to send electronic mail anywhere in the
Translation of data into a secret code. Encryption is the most
effective way to achieve data security. To read an encrypted file,
you must have access to a secret key or password that enables you to
decrypt it.
Buzzword referring to an intranet that is partially accessible to
authorized outsiders. Whereas an intranet resides behind a firewall
and is accessible only to members of the same organization, an
l l f
© SHG 2002
- 98
© SHG 2002
extranet provides various levels of accessibility to outsiders. Access
to an extranet is usually based on a valid username and password.
Federal Communications
Commission (FCC)
These are the people in the government who decide what's legal and
illegal to broadcast, including what frequencies are allowed to be
used by whom.
System designed to prevent unauthorized access to or from a private
network. Firewalls can be implemented in both hardware and
software, or a combination of both. Firewalls are frequently used to
prevent unauthorized Internet users from accessing private
networks connected to the Internet, especially intranets. All
messages entering or leaving the intranet pass through the firewall,
which examines each message and blocks those that do not meet
the specified security criteria.
Combination of hardware and software that links two different types
of networks. Gateways between e-mail systems, for example, allow
users on different e-mail systems to exchange messages.
Gigabit (Gb) - This refers to approximately 1 billion bits. More
exactly, it is 2^30 or 1,073,741,824 bits.
Gigabyte; 2 to the 30th power (1,073,741,824) bytes. One gigabyte is
equal to 1,024 megabytes.
General Packet Radio Service. A digital packet switched data
network that runs over GSM networks, and is capable of theoretical
data rates up to 171.2 kbps.
Group Messaging
© SHG 2002
Ability to send same message to several people.
Global Standard of Mobile Communication. Currently the leading
digital cellular technology. GSM systems use narrowband TDMA,
which allows eight simultaneous calls on the same radio frequency.
HyperText Markup Language - a standard language made for
typesetting, used for creating documents on the World Wide Web.
Included in the language are provisions for including pictures and
links to other pages.
Secure HyperText Transfer Protocol - a secure means of transferring
data over using the HTTP protocol. Typically, HTTP data is sent
over TCP/IP port 80, but HTTPS data is sent over port 443. This
standard was developed by Netscape for secure transactions, and
uses 40-bit encryption. . The HTTPS standard supports certificates.
A web server operator must get a digital certificate from third party
certificate provider that ensures that the web server in question is
valid. This certificate gets installed on the web server, and verifies
for a period of a year that that server is a proper secure server.
- 99
© SHG 2002
An easy-to-use online instant messaging program developed by
Mirabilis LTD. Pronounced as separate letters, so that it sounds like
"I-Seek-You," ICQ is similar to America OnLine's popular Buddy
List and Instant Messenger programs. It is used as a conferencing
tool by individuals on the Net to chat, e-mail, perform file transfers,
play computer games, and more.
Instant Messaging. Type of communication service enabling a user
to create a private chat room with another individual
Short for Internet Message Access Protocol, a protocol for retrieving
e-mail messages. The latest version, IMAP4, is similar to POP3 but
supports some additional features. For example, with IMAP4, you
can search through your e-mail messages for keywords while the
messages are still on mail server. You can then choose which
messages to download to your machine. IMAP was developed at
Stanford University in 1986.
The digital packet-based low-speed (9.6 kbps) Web browsing and
mobile messaging platform deployed in Japan by NTT DoCoMo in
February 1999.
A method of connection between two separate entities. For
example, a graphical user interface (GUI) is the part of a program
that connects the human user to the computer functions. Interfaces
can also connect programs and devices.
A network operating like the World Wide Web but having access
restricted to a limited group of authorized users (as employees of a
Integrated Services Digital Network - a digital line that is often
used to connect to the Internet. It generally come in two flavors: one
is a 56 Kbps version, which in actuality only uses half of the ISDN
line's bandwidth; the other is the 128 Kbps version, which uses both
the 56 Kbps channels on the line. However, that's only 112 Kbps--the
other 16 Kbps are an 8 Kbps back channel of each line.
Internet Protocol - a connectionless communications protocol that
forms part of the basis for the TCP/IP protocol suite. It is a fast
protocol, but it has no mechanism for sequencing or error
conditions. Error packets are simply lost. IP will basically just move
IP Secure - This is the IETF standard "secure IP" transport.
Typically, IPSEC is used in branch-VPN tunnels between routed
© SHG 2002
- 100
© SHG 2002
LAN segments, but it's destined to become the method for securing
IP traffic over IPv6.
IPv6 (Internet Protocol version 6) - This is the current version of the
IP protocol that features a 128-bit addressing scheme, as opposed to
the 32-bit addressing scheme of IPv4, supporting a much higher
number of addresses. It also features other improvements over IPv4,
such as support for multicast and anycast addressing.
Internet Service Provider.
Java Virtual Machine - a program that runs under an operating
system and interprets Java programs. The Java Virtual Machine
ideally will not allow any harm to come to the computer because it
has no control of the operating system, and acts as if it is a separate
computer. Thus, if a malicious Java program were to crash the Java
Virtual Machine, the operating system would remain stable. Another
advantage of this mechanism is that different OS's can have their
own Java Virtual Machines that should act identically. Thus, Java
should be able to be run across different platforms easily with no
code change
1024 bits (2^1 bits)
Kilobits per second
A measure of data transfer. A 14.4 Kbps modem transfers data at
about 1.8 kilobytes per second or about 100 KB per minute.
1024 bytes (2^10 bytes)
Local Area Network.
The amount of time required for a message to be sent from a twoway wireless device, and the first byte of the response received.
(Since the time required for receipt of the entire message varies with
the size of the message, this definition controls for message size.)
A two-way low speed packet data network introduced by Ericsson in
Megabit (Mbit) - Roughly one million bits. Exactly 1,048,576 bits
(that's 2^20 bits).
Megabits per second - aka Mbps. This is a measure of throughput
roughly in millions of bits per second. More exactly, that is 2^20
(1,048,576) bits per second.
Megabyte - Roughly one million bytes. Exactly 1,048,576 bytes
(that's 1024 x 1024, or 2^20).
Megahertz (MHz) - One million hertz.
Numeric Pager
A wireless device that allows a person to receive a phone number.
Connected to the Internet.
© SHG 2002
- 101
© SHG 2002
A collection of information. It's often used to refer to the chunks of
information sent over computer networks.
A set of digital cellular technologies being deployed in the U.S.(1)
Peer to Peer (P2P)
A method of distributing files over a network. Using P2P client
software, a client can receive files from another client. Some P2P file
distribution systems require a centralized database of available files
(such as Napster), while other distribution systems like Gnutella are
POP (email)
Public Key Infrastructure. - This is the infrastructure needed to
support public key encryption. It requires a certificate authority to
issue and verify the public keys, a registration authority that verifies
the identity of a person or organization before a key is issued, a
certificate directory of the public keys and a certificate management
system. Public key encryption can be used to verify an identity or to
encrypt data or messages.
Short for Post Office Protocol, a protocol used to retrieve e-mail
from a mail server. Most e-mail applications (sometimes called an email client) use the POP protocol, although some can use the newer
IMAP (Internet Message Access Protocol).
There are two versions of POP. The first, called POP2, became a
standard in the mid-80's and requires SMTP to send messages. The
newer version, POP3, can be used with or without SMTP.
POP (access)
Short for Point of Presence, a telephone number that gives you dialup access. Internet Service Providers (ISPs) generally provide many
POPs so that users can make a local call to gain Internet access.
Personal Digital Assistant. Handheld device combining computing,
telephone/fax, and networking features. Some PDAs can function
as a cellular phone, fax sender, and personal organizer. Unlike
portable computers, most PDAs are pen-based, using a stylus rather
than a keyboard for input. Some PDAs can also react to voice input
by using voice recognition technologies.
Quality of Service -- an effort to provide different prioritization
levels for different types of traffic over a network. Various methods
are used to achieve quality of service, including the RSVP protocol.
For example, streaming video may have a higher priority than ICMP
traffic, as the consequences of interrupting streaming video are
more obvious than slowing down ICMP traffic.
© SHG 2002
Radio Frequency -- the range or frequencies between 10 kilocycles
per second to 300,000 megacycles per second in which radio waves
can be transmitted. It can also refer to a frequency used for a
specific radio station.
A two way low-speed paging overlay network introduced by
Motorola in 1994.
- 102
© SHG 2002
Response Time
Length of time it takes to send/receive text messages.
A device that connects any number of LANs.
Applications or systems that are able to scale to large amounts of
users. For example, a database that completely locks out every other
user when someone is using it is NOT scalable. The computer
system that runs ATM and bank transactions must be highly
scalable. Scalable
Refers to techniques for ensuring that data stored in a computer
cannot be read or compromised. Most security measures involve
data encryption and passwords. Data encryption is the translation of
data into a form that is unintelligible without a deciphering
mechanism. A password is a secret word or phrase that gives a user
access to a particular program or system.
Service Level Agreement s a promise of maintaining a consistent
level of data transfer over a network. Every ISP typically has a SLA
that states the promise of data availability that the ISP will provide
for their customer. Usually SLAs are only given to business
customers that pay more for their connections than home users.
Thus, business connections are typically more reliable and also cost
more. SLAs are important for companies that can lose millions of
dollars when their customers cannot access their webservers.
Sleep mode
The placement of a computing device into an inoperable mode
where less power is consumed by shutting down unnecessary
devices, but leaving all data in RAM. Typically, you return from
sleep mode by using the keyboard or mouse and devices are
switched back on. Sleep mode in its early incarnations was very
problematic in some PCs and would often crash programs and
operating systems that were not completely compatible with the
sleep mode capable by the PCs BIOS.
Short Messaging Service. Transmission of short text messages to
and from a mobile phone, wireless device or IP address. A method
of sending text messages that are 160 characters in length or shorter
over a mobile phone. More and more mobile phones are supporting
the sending and receiving of SMS messages.
SMTP Short for Simple Mail Transfer Protocol, a protocol for
sending e-mail messages between servers. Most e-mail systems that
send mail over the Internet use SMTP to send messages from one
server to another; the messages can then be retrieved with an e-mail
client using either POP or IMAP. In addition, SMTP is generally
used to send messages from a mail client to a mail server. This is
why you need to specify both the POP or IMAP server and the
SMTP server when you configure your e-mail application. The main
limitations of SMTP are that it is limited to 7 bit ASCII data,
handles limited message lengths, has to rely on MIME for
attachments, and often has inconsistent format translation. It uses
TCP/IP port 25 and allows for file attachments.
Systems Network Architecture - an IBM architecture for enterprise
© SHG 2002
- 103
© SHG 2002
computing systems. IBM has created a complete suite of programs
to work on their proprietary hardware for enterprise computing.
Electronic junk mail or junk newsgroup postings
Secure Sockets Layer -- a protocol specified by Netscape that
allows for "secure" passage of data. It uses public key encryption,
including digital certificates and digital signatures, to pass data
between a browser and a server. It is an open standard and is
supported by Netscape's Navigator and Microsoft's Internet
Network device that filters and forwards packets between LAN
segments. Switches operate at the data link layer (layer 2) and
therefore support any packet protocol.
2nd Generation Wireless - wireless technology used in the 1990s,
and still in use in the year 2000 and later. It features digital encoding
of voice and 3G features are slowly being added to form a sort of
2.5G version of digital wireless. Here today, perhaps gone tomorrow.
A second generation wireless technology (2G) with incomplete third
generation (3G) technology added to it. Imperfect, but infinitely
cheaper than 3G.
3rd Generation Wireless) - This refers to the phase of cellular
wireless communications that promises 2Mbps+ wireless data
transfer speeds, full roaming throughout Japan, US and Europe, as
well as enhanced multimedia capabilities and a standard features set
including cellular voice, e-mail, paging, and Web functionality.
Synonyms include “white elephant” and “HDTV.”
Time Division Multiple Access. Technology for delivering digital
wireless service using time division multiplexing. TDMA works by
dividing a radio frequency into time slots and then allocating slots
to multiple calls. In this way, a single frequency can support
multiple, simultaneous data channels. TDMA is used by the GSM
digital cellular system.
Text Messaging
Alphanumeric message delivered to a wireless device.
Text Pager
A wireless device that allows a person to receive a text message or
numeric message.
Thin Client
A desktop application
Transmission Control
Protocol (TCP)
Technology standard and format by which voice and data traffic is
handled and delivered over the Internet.
A communication channel between two points. It usually refers to
large-bandwidth telephone channels between switching centers that
handle many simultaneous voice and data signals.
Ultra Wide Band
© SHG 2002
An RF technology, now in development for commercial
applications, that uses short high energy bursts of radio frequency
energy, generating waveforms that can deliver extremely high data
- 104
© SHG 2002
rates – up to 1000 MB/s -- high resolution, and precision location
detection. They also have the interesting property that they use very
little spectrum, and are immune to multipath cancellation. The
FCC is now examining the impact of UWB network deployment on
interference with other applications, such as GPS and aircraft
tracking. Intel currently has a 60 person research group devoted to
UWB applications.
Uniform Resource Locator. Global address of documents and other
resources on the World Wide Web. The first part of the address
indicates what protocol to use, and the second part specifies the IP
address or the domain name where the resource is located.
Unique address for the receiver that corresponds directly to a
personal identifier.
Virtual Private Network
See VPN.
Virtual Private Network. Network constructed by using public
wires to connect nodes. VPN systems use encryption and other
security mechanisms to ensure that only authorized users can access
the network and that the data cannot be intercepted.
Wireless Access Protocol.
Wideband CDMA -- a 3G standard that increases the throughput of
data transmission of CDMA by using a wider 5 MHz carrier than
standard CDMA which uses a 200 KHz carrier. WCDMA allows for
data transfer rates as high as 2 Mbps.
Web Enabled
Able to receive internet content, communication or data.
Windows CE
A version of the Windows operating system designed for small
devices such as PDAs (or handheld PCs in the Microsoft
vernacular). The Windows CE graphical user interface (GUI) is
similar to Windows 95.
Wireless Transmission
See Transmission Protocol.
© SHG 2002
Wireless Markup Language (formerly HDML) – part of the
Wireless Application Protocol (WAP) and it allows text portions of
Web content to be separated from graphical content for display on
wireless devices.
A particular means of communicating text, graphics, and other
multimedia objects over the Internet. Web servers on the Internet
are set to respond to particular requests sent on TCP/IP port 80 by
sending HTML documents to the requester. The requester must
use a browser to receive this data. Think of the Internet as a 100-lane
highway, and the Web as one of those lanes. Of course, traffic in the
Web lane is probably very high compared to traffic in most other
World Wide Web Consortium -- an industry group created to
design and promote standards to increase the functionality of the
- 105
© SHG 2002
Web. The W3C was initially established in collaboration with
CERN, the creators of the World Wide Web. You can reach the W3C
at .
A packet-switching service that connects remote terminals to host
systems. X.25 has higher overhead than Frame Relay, but has been
around longer and is better supported. X.25 predates the OSI model.
Extensible Markup Language (XML) - XML is a standard created
by the W3C. It is a language with many simularities to HTML.
What XML adds is the ability to define custom tags, such as , and
define the meaning of those tags within the XML document itself.
XML will become more and more common as more browsers and
Web servers support the XML standard.
See, for example, Rick Perera, IDG News Service, May 18, 2001: “ At this time a year ago, Europe was
abuzz over the plans of high-flying telecommunication operators to roll out 3G (third generation) wireless
networks, with their promise of high-speed data transmission and nifty multimedia functions. Today those
same companies are limping financially. Having shelled out billions of dollars for UMTS (Universal Mobile
Telecommunications System) licenses in major European markets, they face problems raising the money
needed to build those networks One idea making the rounds is that multiple operators could share the same
infrastructure. There's no reason four companies, for example, should build four separate sets of
transmission networks in a given country. Why not build fewer base stations, masts, and antennas, as long
as there's enough capacity to handle everyone's customers?”
2 One Fall 2000 forecast by Cahners In-Stat was widely reported to predict that there would be more than
1.3 billion mobile Internet users by 2004. A closer reading of the forecasts shows that this figure doublecounted 607 million predicted SMS users and 783 million wireless Web subscribers, but the latter figure
was still more than 40 times the yearend 2000 worldwide total. In the U.S., for example, the number of
wireless phone users grew from 340,000 in 1984 to 16 million in 1993, and then soared, to more than 79
million in 1999 and more than 100 million by yearend 2000. Global adoption was even more explosive,
with more than 55 million by 1993 and 650 million worldwide cellular wireless users by yearend 2000.
Data from Wireless Survey Results, Cellular Telecommunications Industry Association (CTIA), December
2000. GSM Association,, July 2000.
3 At the peak of all the excitement in mid-2000, there were a flurry of wild-eyed forecasts , often based on
nothing more than sheer optimism. In addition to this September 2000 forecast by the Yankee Group, which
forecast nearly 700 million mobile Internet users by 2004, there were also forecasts by Ovum (2000) – 407
mobile wireless users by 2004; ARC (2000) – 803 million users by 2005; and Mobile Lifestreams – 400
million mobile Internet users by 2004. One of the most widely misquoted estimates was Cahners In-Stat, in
© SHG 2002
- 106
© SHG 2002
September 2000, which was reported to have predicted more than 1.3 billion mobile Internet users by 2004.
On closer inspection this figure turns out to have double-counted a 609 million figure for the SMS users.
But the remaining 783 million estimate for the number of world wide wireless Web users in 2004 was, in
retrospect, still very aggressive.
4 Metricom’s service, based on wirelessly-enabling PC Internet traffic, started with 32 kbps in 13 cities, and
started to upgrade and expand these systems to deliver 128 kbps service in most major metropolitan areas.
In July 2001 it ran out of funding and declared bankruptcy.
5 Considering, for example, whether to upgrade now to GPRS, wait around for EDGE, go to GPRS and
then go to EDGE, convert over to CDMA2000, or wait still longer for W-CDMA.
6 By now the harsh characterizations are legion – “WAP is crap;” “WAP means “where are the phones?;”
“WAP is a trap;” “WAP Lash;” “WAP’s killer app is killing time;” “WAP is the DOS of cell phones;” and
so on.
7 WAP applications are often excruciating, because they run on circuit-switched data networks, where
dialup connections have to be made. The contrast is striking with I-mode, which runs on a digital packetbased network that is “always on,” even though throughput is only 9.6kbps. As a result, most WAP phone
services to date have been unsuccessful. Unwired Planet’s (later, then OpenWave) first
customer for WAP-like applications was AT&T Wireless. The first version of its WAP service, PocketNet,
was launched in 1999, and was a flop. A December 2000 study in the UK found that 70 percent of WAP
phone users in a panel who were given free phones to use for one week wanted to give them back.
Download times just to check news headlines or the weather, for example, averaged more than a minute.
“WAP Usability Report,” Nielsen Norman Group Report, December 2000. A December 2000 Accenture
survey of 3189 adult mobile phone users in the US and Europe found that only 15 percent used their Webenabled phones to browse the Web, mainly because service was slow, expensive, and difficult to use.
Jupiter (March 2001) reported that less than 20 percent of US subscribers who had Web phones with WAP
browsers were ever using them for Internet services, and that for Sprint PCS, less than 10 percent of their
customers ever accessed the wireless Web.
Meta Group, August 8, 2001, “Study Finds Corporate Users Giving Up on WAP-Enabled Phones. “
9 By now there have been numerous sharp critiques of WAP’s technical and business strategy since the
WAP Forum was first organized by, Ericsson, Motorola, and Nokia in July 1997. See, for
example, Mike Banahan, “Underwhelmed by WAP --- Impressions from the Coal Face,” May 27, 2000,; Meg McGinity, “WAP Lash,” Interactive Week, July 28, 2000; Keri Schreiner, “WAP
2.0:Mature Enough for Flight?” IEEE, Nov-Dec. 2000; Rohit Khare, “W*Effect Considered Harmful,” 4K
Associates, April 9, 1999; and Mohsen Banan, “The WAP Trap,” May 26,2000, www.
10 At last count there were more than 30 rival “mobile middleware” software companies in contention.
11 Morgan Stanley, “Wireless Data – State of the Union,” May 2001. Another source gives even lower
number figures for WAP users as of November 2000: 4 million in Japan, 2 to 3 million in Korea, 200,000
in the US, and 1-2 million in Europe., November 2000.
12 Cahners In-Stat (March 2001) estimates that the volume of “m-commerce” in the US in 2000 was only
$264 million. While it still estimates that this will grow to $25 billion by 2005, others are not so sanguine.
Jupiter Media Matrix (July 2001), for example, estimates that only .01% of the 110 million US mobile
phone users in the year 2000 purchased something over their phones, and that the total volume of “mcommerce” by way of Web-enabled phones will only reach $3.6 billion by 2005. Among the key factors
retarding the growth of Web phone-based factors are concerns about security and privacy, and the
restrictive “form factors” of existing wireless handset devices.
13 Location-based e-commerce services have also been slow to take off, especially in the US, where the FCC
has delayed the mandate for E-911 services for wireless carriers until October 2001. Positioning equipment
hardware, content, and software companies have, accordingly, been adversely affected.
14 See, for example, the 1996 forecast by NTT DoCoMo analyst K.Kinoshita that there would be 10 million
wireless mobile users in Japan by 2000. In fact the number was close to 65 million, of whom 29 million
were using Web phones. See Mobile Communications Handbook, 1996. (CRC Press, Boca Raton, 1996),
15 Chart 4B includes leading North American wireless data services providers like Arch, Omnisky, Motient,
and GoAmerica, as well as solutions providers Aether and 724 Solutions.
© SHG 2002
- 107
© SHG 2002
Venture capital is reported to have funded just 39 wireless deals, for a total of $500 million in the first
quarter of 2001, compared with 73 deals totaling $1.4 billion during the first quarter of 2000. Red Herring,
June 2001.
17 See, for example, ComputerWorld, 3/22/2001: “US Wireless Industry Eyeing Japan’s I-mode Success.”
18 Total Telecom, 7/16/2001, figures for June 30, 2001 NTT DoCoMo subscribers and market share. Morgan
Stanley Japan Telecom report, June 2001. Since I mode subscriptions and phones are sold together through
retailers, and subscriptions only cost about $2.40 per month, three may be a gap between the actual number
of I mode users and the number of subscribers. eMarketer, “I-mode: subscribers, users, and the area
between,” estimates that about 20 percent of i-mode subscribers may not actually use the service. However,
the growth in actual traffic and subscribers is still dramatic.
19 “DoCoMo” is a brand name, similar to the phrase for “anywhere” in Japanese.
20 Morgan Stanley (6/2001) projects that DoCoMo will account for 796 billion yen of recurring profit this
year, more than 100 percent of NTT’s operating profit (given its breakeven status in the wireline business
and its large losses on its investments in Verio, a hosting company.
21 Consistent with this, as implied in Chart 6, relative market valuations for Japan’s leading wireless data
providers like NTT, Japan Telecom, and KDDI have actually increased in the last year, compared with
those of wireless service providers in the US and Europe, despite Japan’s continued economic woes. From
May 7, 2001 to July 21, DoCoMo’s share price fell by 37 percent, mainly because of concerns about its
overseas investments like KPN, the Java handset recall, its delay of I-mode entry in Europe, and the delay
of its new 3G service.
22 By June 30, 2001, there were 63.39 million cellular subscribers in Japan, a 50 percent population
penetration ratio, and a 77% household penetration ratio. Total Telecom, 7/16/2001, SHG analysis. US cell
phone household penetration is for yearend 2000, a relatively high 51% estimate from Dataquest
(12/21/2000). Merrill Lynch Research (6/2000) reported that US cell phone population penetration was 24
percent; the comparable figure for Japan is about 60 percent. Wired (June 2001) says that US cell phone
penetration is only about 40 percent.
23 www.editorand, June 7, 2000.
24 It costs about $700 to have a new wired phone installed in Japan, compared with only $60 wireless phone
activation fees.
25 An October 2000 survey showed that more than 25 percent of Japanese commuters to work or school
spend at least an hour each way per day, compared with less than 8 percent in North America and six
percent in Europe, and that more than 60 percent of Japanese commuters use public transportation,
compared with just 16 percent in North America and 23 percent in Europe. Cars, on the other hand, were
used by 71 percent of North American commuters and 58 percent of European commuters, but just 24
percent of Japanese commuters. Schauweckers’ Guide to Japan, November 2000. While cell phones can
obviously be used in cars as well as on trains and buses, the relatively long commute times have helped to
encourage cell phone/ handyphone adoption in Japan.
26 ACNielsen (Japan), July 2000 survey – 38.2 percent of Japanese households had PCs.
27 There is wide variation in statistics on PC penetration, Internet penetration, and broadband penetration by
country, but there is general agreement on the overall relative patterns. Gartner Dataquest reported as of
January 2001 that US PC penetration was “over 63 percent” (The Wall Street Journal, 1/19/2001). Arbitron,
June 5, 2000, reported that on a survey of 50 US cities with an average PC penetration of 54 percent. The
US appears to have passed Japan’s current level of PC penetration back in 1997. Dataquest, reported in The
Washington Post, 2/11/1999.
28 “Wired” includes dial-up, ISDN, cable modem, and DSL connections.
29 Japan’s Ministry of Post and Telecommunications reported in June 2001 that as of the end of March 2001,
there were about 17.25 million dial-up Internet users, 785,000 cable modem subscribers, and 112,000 DSL
users. In addition, there are about another 1.25 million ISDN users not included in these figures. All told,
assuming that the ISDN users have been omitted from the dial up users, this amounts to about 19.5 percent
wired Internet penetration. Of these, about 9 million are from homes. ACNielson, 2/2000, reported that
about 45 percent of households with PCs in Japan had Internet connections. It gave a lower number, 8.7
million, for the number of PCs at home. Other measures of Internet penetration in Japan are consistent with
these estimates, though there is substantial variation in the absolute measures used. For example, eMarketer
(May 2001) estimates that as of yearend 2000, there were 17.7 million “Internet users” over the age of 14
in Japan, for a population penetration rate of 19.7 percent. Morgan Stanley, which used a broader measure
© SHG 2002
- 108
© SHG 2002
for al age groups, estimated that there were 29 million Internet users, while Nielsen/NetRatings estimated
28.3 million. See eMarketer, The eJapan Report, May 2001. All reports agreed that Japan’s Internet
penetration rate was relatively low – compared with, say, the US (63%), Singapore (41%) Australia
(31.7%), and South Korea (21%). If our thesis is correct, these countries should all also have lower cell
phone penetration rates than Japan (77%) – and indeed, it appears they do: US (52%), Singapore (70%),
Australia, (69%), and South Korea (58 %). For the US, see The Pew Foundation Internet Report, January
2001. which reported that in December 2000, 56 percent of Americans, and 52 percent of US households,
had Internet access. Arbitron, June 2001, reported that “nearly 60 percent” of Americans had Internet
access. Forrester (quote in The Wall Street Journal, July 16, 2001, p. B1), reports that 63 percent of US
households have PCs and 57 percent have Internet connections. Gartner Dataquest has predicted that 75
percent of Americans will be online by 2004, while Strategy Analytics has predicted that the figure will be
91 percent. As usual, all such forecasts should be taken with a grain of salt, but the overall pattern of very
high US wireline Internet penetration is consistent.
30 See Forrester, supra. Web access in both countries is stratified by income group – among those with
incomes greater than $75,000 a year, Internet penetration rises to more than 83 percent in the US. UPI, July
17, 2001.
31 Measure of average hours per month online vary significantly by survey, but they all agree that Americans
spend at last 50-100 percent per month more time online now than Japanese or Europeans. See Media
Matrix (April 13, 2001); Nielson NetRatings, June 2001, which showed that the average Japanese Internet
users spent 9 hours per month online, compared with 35 hours for Canadians, 7 hours for Germans, and 6
hours for residents of the UK.
32 See eMarketer, The eJapan Report, May 2001, 38, which compares the cost of 40 hours per month of
Internet use in Japan ($49) with the US ($35).
33 Nielson/NetRatings, July 2001, reports that in July 2001, about 43 million office workers, or 32 percent of
the US employed labor force, had Internet connections at work. Wall Street Journal, July 16, 2001, B1. For
Japan, the figure is about 27 percent. A.C.Nielson, February 2000, found that 11. 8 million Japanese
workers had PC connections at work.
34 By cable modem, optical fiber, fixed wireless, ISDN, or DSL connections.
35 For the US numbers, see Arbitron, “Broadband Revolution Part Two,” June 21 2001. See also the June
2001 report by Japan’s Ministry of Post and Telecommunications, supra.
36 Gartner estimates for yearend 2001. Data on household broadband penetration in the US and Europe are
from Strategy Analytics (6/12/2001). The most recent actual estimates for Japan are from the Ministry of
Post and Telecommunications’ most recent data on DSL and cable modems for March 31, 2001, which
showed just 790,000 cable modem users and 120,000 DSL users.. In addition, there are also ISDN users and
some fixed wireless users. The government’s target for broadband is to have 30 million users by 2005.
37 Strategy Analytics, 6/12/2001, forecasts US broadband household penetration in 2005 at “greater than 50
percent,” compared with about 25 percent in Europe and 14 percent in Japan. Gartner is more bullish about
Japanese broadband, predicting 30 percent broadband penetration by 2005.
38 eMarketer, The eJapan Report, May 2005, 17. As noted above, the total number of Web phone subscribers
in Japan passed the 40 million mark in July 2001. As of March 2001 there were about 18 million wired PC
connections to the Internet in Japan. Of course the total amount of Internet activity by wired versus wireless
devices is not necessarily proportionate to the number of devices.
39 Ministry of Public Management, Home Affairs, Posts, and Telecommunications, 2000, cited in
eJapan Report, supra, 44.
40 As of January 1997, there were 10.3 million one-way paging customers in Japan, a population penetration
rate of 8.2 percent. The number of one-way customers was already falling – a year earlier, it had stood at
10.8 million. Compared with other Asian countries like Korea (30 percent) and Taiwan (17 percent), the
penetration rate for one-way paging was relatively low. Asian Technology Information Program, August
1997. NTT DoComo accounted for about 58 percent of subscribers at this point, and used a nation-wide
FLEX network.
Earlier we saw that about 31% of the country’s 130 million adults, or 40.3 million, have cell phones.
Only about 10.4 million, or about 8 percent, had pagers.
42 As developed for the US market, both ReFLEX™ and Mobitex™ required frequencies in the 900 MHz
“narrowband PCS” range, which the US FCC licensed to US paging companies in 1994. These nationwide
© SHG 2002
- 109
© SHG 2002
frequencies were not available in Japan. The 450 MHz spectrum that was available had very poor inbuilding penetration – critical to Japan’s densely-populated urban markets.
43 The $6.3 million ReFLEX™ network was sold to Tokyo Web Link Inc, partly owned by Japan Telecom, a
leading NTT competitor.
44 In the US market many PDAs have been sold as an adjunct to desktop PCs.
45 ComputerChannel, 5.10.2001, reported annual sales of 912,000 PDAs in Japan, a 20 percent increase over
2000. It forecast 2.1 million units sold for 2005, but also admitted that PDAs faced sharp competition from
Web phones in the Japanese market.
46 Yankee Group, 2001; SHG analysis of Cingular, Omnisky, and GoAmerica PDA wireless subscribers.
47 A recent survey of 1480 respondents by the Nikkei Business Daily reported that 39 percent preferred
PDAs as Internet terminal devices, compared with 27 percent for Web phones and 25 percent for PCs.
(Reported in Wireless Industry News, 5/25.2001).
48 Industry Standard, July 19 2001. Reuters, August 22, 2001, reported that Sharp announced it would make
so-called “3G PDAs” for NTT DoCoMo.
49 GSM Association, July 2001. In the Philippines, where SMS messaging has become a very low cost
alternative to voice calls, about 5 million cell phone users send an average of 40 million SMS messages per
day, or 240 messages per month!
50 IDC, “Mobile Date Services,” op.cit., May 2001, estimates that SMS traffic in Europe will continue to
grow by nearly 19 percent a year in Western Europe through the year 2004, and then start to decline,
presumably because of 3G services. As noted below in this white paper, we have serious doubts about the
3G’s adoption rate and economic viability, and would be quite willing to bet that SMS, on the other hand,
will have its useful life prolonged by new technologies like T9 intelligent text and chat boards, and the
adoption of prepaid billing models, and the growth of handset personalization services.
51 Assuming a 132 character message,
and 128 bit packets, an SMS message is the equivalent of about
$.058 cents per Kb. This is about 291 times the price per kb of a one minute $.10 voice call. See Morgan
Stanley (May 2001), op. cit.
52 IDC, Mobile Data Services, op.cit., May 2001.
53 Vodafone D2 says that SMS messaging alone now accounts for 16 percent of its revenue, while Sonera
reports an 11 percent revenue share. Both have marketed new SMS services like information push and
mobile handset personalization, and now have higher ARPUs than their competitors
54 See IDC, Mobile Data Services and Applications: Forecast and Analysis, 2000-2005. (, May,
55 SMS latency and unpredictability arises from delays that are inherent in its queuing model, and
fundamentally, the fact that it has to compete with voice traffic. It relies on the cellular network’s control
channel for capacity. One recent US study by Mspect Inc. reported on the results of sending 30,000 sample
SMS messages on six different US cellular networks. It found that an average of 13 percent of the
messages sent took more than ten minutes to arrive, and that the fraction of messages received within 30
seconds varied significantly by network , from a low of 46 percent to 98 percent. See Computer World, June
20, 2001.
56 At first, the Group Speciale Mobile GSM) standard, later the Global System for Mobile Communications
– (GSM, minus the C).
57 This meant that
message recipients would not be charged for person-to-person messages. however,
subscribers who sign up to receive news services are indeed billed for SMS messages received, on a prepaid
basis. But not the recipients of messages sent by others.
58 These included the Nordic Mobile Telephone System, Germany’s Total Access Communication System,
France’s RadioComm2000, Italy’s RTMI/RTMS, and several others.
59 Unlike cellular operators in the US, Europe’s operators provide third-party SMS hubs for internetworking
traffic across individual networks. This is precisely the basic Internet’s model. It provides a striking
contrast to the Balkanized structure of non-interoperable cellular and wireless data network that exists in
the US.
60 The British Post Office’s POCSAG standard, and the ERMES standard fostered by ETSI.
61 European paging operators were required to deploy ERMES networks, and not allowed to use the FLEX
standard, even though ERMES devices were significantly more expensive and less diverse. Operators in the
UK and some other European countries were finally allowed to deploy FLEX networks in 1998, but at that
point it was a desperation move to try to salvage something from the dying European paging industry.
© SHG 2002
- 110
© SHG 2002
Given the absence of calling-party-pays in the US, for example, before the introduction of digital cell
phones with caller i.d., cell phone users often carried both pagers and cell phones to help them control their
phone bills. The European paging industry tried to adopt its own sender-pays rules in response. But while
this created a short-term surge in pager sales, it did not help the long-term condition of the European paging
industry,. This may have been also partly due to the high rates (greater than $1 per message) that were
charged to maintain ARPU levels.
63 As of July 2001, for example, the UK’s Mobitex™ operator Transcomm PLC, which bought RAM
Mobile Data’s UK subsidiary in December 2000, claimed 30,000 customers.
64 A subsidiary of Deutsche Telecom
65 For example, BT’s price per SMS message sent is about $.12, compared with $.02 for Verizon.
66 In the US in 2000, about 40 percent of the 40 million
new cell phones shipped were Web-enabled.
Naviglobe, 3.28.2001.
67 This study reported that 58 percent of wireless subscribers in Europe were using SMS messaging,
compared with just 11 percent in the US. Wireless Week, May 28, 2001.
68 Morgan Stanley, “Wireless Data: State of the Nation,” op.cit.; our analysis. See also Strategis Group,
“State of the U.S. Paging and Advanced Messaging Industry,” March 2001, 40
69 Communications Weekly International, July 16, 2001.
Uniform Resource Locator. See Glossary.
71 See
72 Red Herring, “Wireless A La Mode, June 12, 2000.
73 Takeshi Natsumo, Gateway Business Developer, DoCoMo, Communications Week International, July 26,
74 “Hima tsubushi” – killing time – is a pervasive feature of Japan’s long commutes and crowded facilities.
75 As of June 2001, I-mode was charging .3 yen per incremental packet. With 128 bytes per packet, and
126.3 yen per dollar, this implies a price per megabyte of $18.56. Even I-mode’s new 3G FOMA service,
launched in pilot mode in June and is being priced more aggressively to attract users, will charge .05 yen
per byte, or $3 per MB. For comparison’s sake, ReFLEX™ service providers offer plans that are the
equivalent of about $4.50 per MB for heavy use. Current packet-switched data services offered by AT&T
Wireless vary from $5.50 to $46 per MB; Verizon’s are much lower, at $2.40 to $5.50 per MB. SMS
services, when looked at this way, are astronomically expensive – from $10 to $110 per MB in the US, and
from $60 to $130 per MB in Europe. Of course most customers have not been taught to think in terms of
how much they pay per MB of data; they are accustomed to think in terms of minutes of use, for cellular
networks, or cost per message. In some time frame, as networks migrate to packet-based digital, and
services become more fungible, the assumption made by some analysts is that customers and competitors
will both become much more aware of these per-MB price differentials. For the sake of comparison, the
price per MB equivalent for a minute of voice traffic, assuming $.10/minute and an 8 kbps codec, is only
about $.20 per MB – less than 1/300th of what carriers are effectively earning per MB with SMS messaging.
See Morgan Stanley, “Wireless Data Services – The State of the Union.” (May 2001.)
76 Communications Week International, July 16,2001; eJapan Report, May 2001, op.cit.
77 Both WML and WMLScript have to be learned from scratch, and are very different from HTML.
Ordinary HTML pages have to be completely rewritten in WML to be available to wireless devices for
WAP services.
78 Though it made the client-side WAP browsers free, OpenWave, formerly,
proposed to charge carriers a license for its WAP gateway of $20 per potential subscriber per year, whether
or not they actually subscribed to Web services. In July 200 Geoworks, a San Francisco-based software
company that has fought a patent battle with OpenWave over the WAP Gateway, starting seeking $20, 000
per server from major corporations that were using WAP phones.
79 NTT DoCoMo’s I mode service was launched on a digital packet-switched variant of a PDC/ P TDMA
network at 800 Mhz, delivering 9.6 kbps. KDDI, one leading competitor, offered its “au” service with
WAP on a CDMAOne network, at 14.4 kbps. Japan Telecom, the other leading competitor, offers WAPbased services at 14.4 kbps on a circuit-switched PDC network. For a discussion of WAP’s security
vulnerabilities, which were inherent in the gateway model, see R. Khare, op.cit. The WAP Forum claims to
have fixed these security deficiencies in Release 2.0, due this month. With respect to circuit-switched
networks, one of I-modes core advantages, despite its slower nominal throughput, is the fact that it is
“always on.”
© SHG 2002
- 111
© SHG 2002
For a discussion, see Banan, op.cit. At least nine members of the WAP Forum have made declarations of
“intellectual property rights” that may be covered by the WAP Forum’s standards, including Motorola,
Nokia,, Entrust, Geoworks, NEC, Diversinet, and an individual named Behouz Vezuan, a Fin
who claims to be the “inventor” of WAP.
81 See the discussion of the Openwave v. Geoworks litigation, supra in footnote 36.
82 In January 2001 NTT DoCoMo, Telecom Italia, and KPN Mobile announced that they would be
launching i-mode services this year in Europe in the Netherlands, Belgium, and Italy over new GPRS
networks. However, in July 2001, they delayed this to “some time in 2002,” because GPRS handsets are
still scarce, and WAP 2.0 has still not been delivered.
83 The I-appli service was launched on January 26, 2001, but Java-enabled phones from NEC and Sony were
not available until the end of March. That makes the growth of this service even more dramatic.
84 Strictly speaking, Java and j2ME are not an “operating systems; “indeed, Sun Microsystems makes a great
deal out of the claim that Java runs across all other devices and operating systems, and is “OS agnostic.”
Still, especially in the mobile world, it is viewed as competing with other mobile operating systems like
Symbian’s EPOC, supported by Nokia, Motorola, and Ericsson for GSM phones; Microsoft’s Stinger, now
about to be launched into service on Samsung-based phones by Telefonica and Australia’s Telstra; the Palm
OS, now running on CDMA2000 phones from Kyocera and Samsung, and being brought into service by
Sprint PCS, and Qualcomm’s BREW, about to appear on Samsung and Kyocera CDMA phones for new
service with Verizon and KDDI. So far J2ME has the market lead among cellular operators because of its
relationship with NTT DoCoMo. KDDI and J-Phone are also looking at introducing Java phones, as are
FarEastTone in Taiwan and SmartTone in Hong Kong. In late 2001 it will also become available on new
Motorola phones available from Nextel in the US, with Java phones also planned for introduction by Sprint
PCS and Bell South. In Europe, Telefonica in Spain, One 2 One in the UK, But Java’s lead may eventually
be challenged, mainly by GSM’s use of EPOC.
85 In February 2001 DoCoMo had to recall 230,000 Java-enabled handsets because of operating problems,
and again in May 2001 it had to recall 420,000 Java-enabled Panasonic 503i handsets. The problems were
quickly fixed, and Java handset sales have continued to grow at more than 4 percent a month.
86 Japan Internet Report, 7/2001.
87 Interview with Sun-Japan executive, Communications Week, July 16,2001.
88 KDDI website, 6.14.2001.
89 The 505- joint venture, Mobimagic, was formed in mid-1999 between NTT Mobile Communcations and
Microsoft. NTT has also been expanding its relationship with Microsoft in other areas – for example, in
March 2001 the two companies announced a deal whereby NTT would host the forthcoming “Xbox” on
broadband for online gaming -- to Sony’s surprise.
90 In May 2001 NTT invested in AOL Japan, and AOL and NTT DoCoMo announced an agreement that
would permit AOL users to get their AOL email on I-mode. Cnet, May 21, 2001.
91 In August 2000, AOL adopted cHTML as its worldwide standard for wireless services, partnering with
DoCoMo to help it develop services outside Japan for its 32 million customers. Since 23 million of these
AOL customers reside in the US (as of June 2001), the implementation of this agreement obviously depends
on NTT DoCoMo’s ability to work with US service providers, like AT&T Wireless and perhaps dedicated
Java-enabled network providers as well. As of July 2001, NTT has acquired 15 percent of KPN, 16 percent
of AT&T Wireless, 20 percent of KG Telecom in Taiwan, 20 percent of Hutchinson 3G in the UK, and
14.5 percent of SK Telecom in Korea. This year DoCoMo also announced plans to migrate I-mode services
to Europe and the US, though some of these plans have been delayed. See footnote 44 above.
92 TNS Interactive (2001).
93 InfoComm Research (2000), reported in eJapan Report, eMarketer (May 2001), 59.
94 O’Reilly, September 20, 2000; CommWeb, April 16, 2001.
95 Supra.
96 Morgan Stanley Japan Telecom Report, July 21, 2001.
97 DoCoMo had previously asserted that I-mode actually boosted cellular voice traffic per users by about 10
percent See CommWeb, April 16, 2001.
98 See, for example, the recent analysis by Spectrum Strategy (July 2001), “3G Madness – Time for
Moderation,” which argued that even IF it cost $10 billion of fixed cost to deploy a 3G network (assuming
just $3 billion to build the network, $6.3 billion to buy the license, and another $.700 billion to launch and
market services), such an investment might be able to avoid bankruptcy (note: though not be very
© SHG 2002
- 112
© SHG 2002
profitable) IF it could get 9 million subscribers to pay an average ARPU of $25 per month for data services
by the year 2012, in nominal dollars. Typical data service ARPUs for cellular customers now, by
comparison, are around $2.
There are many things wrong with this analysis, but the assumption that (allowing for a 5 percent annual
inflation rate) the real-dollar value of ARPUs for data would rise by an average of nearly 20 percent a year
over the next 12 years seems especially dubious.
99 For example, the elongated mini-brick design of most cell phones, which
contributes to cramped
keyboards and tiny screens, and the use of numeric keypads, are both artifacts of the fact that the devices
were primarily designed to reach from ear to mouth.
100 Depending on network concurrency, this probably translates into perhaps 64kbps – 96 kbps per user in
actual throughput.
101 The FOMA pilot service runs on a $760 handset, which provides full motion video at up to 64 kbps.
Initial reports indicate that the service and the handset are buggy. The screen is small, battery life is short, it
can’t communicate with PCs or non-FOMA handsets, there is too little memory, one can’t do a call while
surfing, and the device gets hot. See the Japan Internet Report, July 2001, for a review.
102 The FOMA service has been initially priced at the same 300 yen per month ($2.40) as I-mode’s basic
service. But its cost per packet delivered is just .05 yen, compared with .3 yen for I-mode. This may partly
reflect the expectation that the network will have greater capacity and lower costs, but it may also reflect a
strategy to drive penetration. Of course NTT DoCoMo also expects that MBs per customer will much
higher if its new 3G multimedia applications are successful – with 30 seconds of MPEG4 –compressed
video requiring an average of 3 MB capacity, even FOMA’s service would cost about $19 per minute of
downloaded video. This is well below the $113 that I-mode pricing would imply. A typical 50 kilobyte
JPEG still picture, on the other hand, would cost just $.16.
103 Communications Week International, July 16, 2001. On September 2 2001 DoCoMo announced pricing
for its FOMA commercial service, still on track to launch October 1. The monthly premium, above voice
service, is 8000 yen (about $64), four times the monthly fee for i-mode, and FOMA handsets will cost
subscribers about 50,000 yen ($404).
104 Note that both competitors are adopting non-PDC technologies, partly to avoid dependence on NTTowned network technology.
105 One recent estimate is that I-mode is now at about 75-80 percent of peak network capacity, with its
customer base still growing at 4-5 percent a month. Japan Inc’s Wireless Watch, July 16 2001.
106 Japan Inc’s’ Wireless Watch, July 16, 2001. According to the report, DoCoMo is doubling the capacity of
each base station from 6 to 12 simultaneous sessions. In the words of one observer, “this cannot be cheap.”
107 Supra.
108 In 2000, the US accounted for more than 60 percent of the 1.5 mm unit video camera market. Computer
World, June 2000.
109 As of April 2001, there were already 16 million wired broadband users in the US. One-third of them were
in five major cities – New York, LA, San Francisco, Boston, and Seattle. Nielsen/Netratings, May 2001.
These are arguably precisely the “Tokyo-like” urban environments where high-speed mobile wireless
broadband would be most in demand. For the umber of cell sites required for 3G coverage, see Crown
Castle (antenna company), interview with M. Scheuppert , SVP, May 2001. As one commentator in the UK
put it recently, “3G almost requires a base station on every corner.”
110 US population density in 2000 averaged just 30 per square kilometer, compared with 334 in Japan, 476 in
South Korea, 130 in China, 241 in the UK, and 230 in Germany. Even if we omit Alaska and the most
empty Western states, average density rises to just 75 per square kilometer. Only about 41 percent of
Americans live in urban areas, compared with 78 percent in Japan, 87 percent in Germany, and 76 percent
in the UK. CIA Factbook, July 2000.
111 Yankee Group (2001) – cell site data per country, cited in ComputerWorld, June 21, 2001. There are
about 80,000 cell sites in the US, divided among about 10 leading carriers.
112 By now, after twenty years of deployment about 91 percent of the population has access to three or more
cellular operators, and 75 percent have access to five or more.IDG data, cited in ComputerWorld, June 21,
113 As of yearend 2000, there were roughly 230,000 fixed wireless Internet subscribers in North America,
compared with roughly 2.3 million DSL subscribers and 4.8 million cable modem subscribers. Cable
Datacom News, June 1, 2001; eMarketer, March 2001; DSL Prime News, March 2001. About sixty percent
© SHG 2002
- 113
© SHG 2002
of the fixed wireless customers were businesses; we don’t have a breakout for the DSL and cable modem
customer bases, but we suspect that they were overwhelmingly residential. eMarketer, op.cit, predicted an
installed base for fixed wireless in the US of just 3.86 million subscribers, compared to 16-20 million cable
modem users, by 2003. Strategis Group, May 2001, predicted just 2.4 million users in the US by 2003 and
5.4 million by 2005. Interestingly, it also foresaw an $8.6 billion market in Europe, where wired broadband
is farther behind. These forecasts were made without taking into the account the possibility of the Sprint
Broadband nationwide service discussed below. Some possible revivals by players at other frequencies like
the unlicensed 60 MHz band (e-Xpedient) and the 28GHz-39GHz LMDS bands in the US has also been
114 For example, the development new non-line of site technologies, “smart antennas,” and low-cost ASICs
that will considerably reduce the cost of both receivers and base stations. Together, these technology
developments may make it possible to deploy integrated receivers in homes and businesses without
external antennas, at less than $300 per home, in the price range of DSL and cable modems, and also permit
the service to be deployed mainly by self-provisioning. Together, this would significantly enhance the
economics of fixed wireless access. Typical fixed wireless services would include bi-directional bandwidth
of 384 kbps to 12 Mbps or more, depending on base station deployments and whether the spectrum used is
licensed or unlicensed. Typical multipoint services have offered up to 1-2 Mbps of shared bandwidth, bidirectionally Most US service providers to date have used unlicensed spread spectrum at 2.4 GHz. For
example WorkNet, a fixed wireless ISP that launched service in 2000, was able to deliver stable Internet
connections at 2 – 4.4 Mbps to business customers in the 2.4 GHz band.
115 Among the leading fixed wireless technology alternatives are Spike Broadband, Soma Networks, and
ArrayNet. SHG industry interviews, July 2001. One motive for IXCs like Sprint, Worldcom, and AT&T, as
well as RBOCs like Verizon that are going national to offer fixed wireless is to compete with the
broadband access strategies of RBOCs (DSL) and AT&T Broadband (cablemodem). Another reason may
be to do an “end run” around RBOC access charges, assuming that telephone services can be provided over
fixed wireless access.
116 Otherwise known as IMT-2000 systems, after the standards that govern them.
117 This difference in policy is also due to variations in local market influences. Europe’s leading 3G network
vendors, Alcatel, Siemens, Nokia and Ericsson, have a strong interest in preserving their significant GSM
customer franchise. Because 3G technology is late and expensive to deploy, many GSM networks have
been looking for alternatives. Technically speaking, CDMA2000, US-based Qualcomm’s technology, can
also be configured as an upgrade to GSM or TDMA networks. However, it requires more bandwidth – 1.25
Mhz, vs. 200 kHz on the GSM circuit switched network – than GPRS, the European vendors’ preferred
alternative. The “only new spectrum for 3G” rule effectively makes it difficult for European operators to
reuse their existing GSM spectrum for CDMA2000. For a given level of projected demand for 3G services,
this reduces the supply of spectrum frequency on the market, and guarantees that Europe’s regulators
received higher bids for 3G licenses.
118 Estimate by Gary Rhodes, former Assistant Secretary of Commerce, Nov. 15, 2000. Adding this to the
roughly 189 MHz of existing spectrum available for 3G in the US, the total would come to about 349
MHz. This compares with the 300 MHz that Japanese regulators have reserved for it, 395 MHz in
Germany, and 364 MHz in UK. See FCC (March 2001), op.cit.
119 In January 2001, the FCC reauctioned about $16.9 billion of 1900 MHz spectrum, about $15.8 billion of
which it had previously auctioned to NextWave Communications, which failed to meet payment conditions.
In May 2001 NextWave successfully sued the FCC, getting the auction overturned. Winners of the second
auction, like Verizon, were counting on it to provide them with the additional spectrum required for 3G.
The matter is now on appeal, and the parties are bargaining.
120 The most important licensed-spectrum versions of fixed wireless in the US occupy frequencies that are
also important to 3G. For example, MMDS fixed wireless service is located in the 2.5- 2.69 MHz band. As
of June 2001, Sprint had acquired MMDS licenses in 90 US markets covering 30 million households, and
had filed new license applications for 45 more, while MCI/Worldcom had licenses in 78 markets and was
seeking them in 30 others. US Federal Communications Commission, “Spectrum Study of the 2500-2690
Mhz Band,” (Washington, D.C., March 30, 2001)One recent FCC study indicated that it could cost up to
$19 billion and take ten years to clear this 2.5-2.69 Ghz band of MMDS carriers – even apart from new
services they may be launching.
© SHG 2002
- 114
© SHG 2002
121 The FCC is having a hard time persuading these players, who basically got their spectrum for nothing, to
give up any frequencies either in the MMDS range, the “milband” range at 1755-1850 MHz, or in the 700
MHz, where in the mid-1990s TV broadcasters were given free spectrum for digital TV services that in
most cases have never been launched. The FCC had originally set a deadline of fall 2001 for auctions in
the 700 MHz range, but those have been repeatedly delayed. It also faces a legal deadline of September
2002 for auctioning other 3G frequencies, but is also behind on that schedule.
122 As identified by the World Radio Conference – 2000 and the WARC-1992, among the possible frequency
candidates for 3G are 698-746 Mhz, 747 – 762 Mhz, 777-791 Mhz, 806 – 960 Mhz, 1710-1855 Mhz.,
1850-1990 Mhz, 2110-2150 Mhz, 2160-2165 Mhz, and 2500-2690 Mhz. The most sought after are 17101855 Mhz, because that would harmonize the US with the rest of the world, or 2500- 2690 Mhz. However,
the former are heavily populated by US military and other US government agencies, while the latter have
the MMDS problem. See FCC (March 2001), op.cit., and US Department of Commerce, National
Telecommunications and Information Administration, “the Potential for Accommodating Third Generation
Mobile Systems in the 1710-1850 Mhz Band,” (Washington, D.C., March 2001).
For example, W-CDMA’s version 3G is expected to deliver up to 2 Mbps of shared bandwidth in
stationary applications. Current fixed point-multipoint wireless technology can easily delivery 4-8 Mbps or
more up to 18-20 miles or more from base stations.
124 UWC-136, or EDGE, provides a specification that calls for three levels of upgrade – the first provides
for enhancements to 30Khz channels for advanced voice/data, the second adds a 200kHz carrier component
for high-speed data to 384 Kbps (“136HS Outdoor”), and the third adds a 1.6 MHz carrier component for
high speed indoor data, to 2 Mbps. Source: FCC (March, 2001), op.cit., 2-6.
125 Estimates for cost/MB are from Morgan Stanley , “Wireless Data Services,”op.cit., 12. These estimates
are based on a recent study by Qualcomm, and may be biased in CDMA’s favor. They also assume that all
four networks will be deployed by then, and that average traffic per user of 205 MB per month, which
Morgan says will appear by 2005. At current compression rates, that translates into more than 3.5 hours of
downloaded video per month.
126 eMarketer, February, 2001. A more bullish forecast by Cahners In-Stat estimates that for the world as a
whole, 3G’s market share of global wireless market will be 50 percent by 2005.
127 As noted, NTT DoCoMo’s initial FOMA service offers just 64 kbps of mobile video, with up to 384 kbps
of shared bandwidth to come later one. Metricom’s Richochet service offered 128 kbps Internet access to
mobile laptop users in 13 cities, and planned to do a national rollout.
128 While Metricom declared bankruptcy in June 200l and is attempting to continue network and commercial
operations to its 40,000 subscribers through its restructuring, it will require a significant capital investment
in order to survive. In the fixed wireless broadband space there have also been several recent dramatic
failures. Winstar, which bit the dust in April 2001, had raised over $1 billion from investors like Microsoft
and CSFB to launch, among other services , high speed fixed wireless in metropolitan areas for businesses,
using LMDS technology at 38 GHz. Teligent, which filed for Chapter 11 in May 2001, was also focused on
using a combination of its own fixed wireless technology and DSL to provide business access. By the time
it closed, it had sold more than 500,000 connections, but only 36,000 customers had signed up for its high
speed voice/data over IP Internet service., May 21, 2001. WorkNet, a St-Louis-based
company funded in part by UBS Capital, launched a $53 million nationwide build out of its wireless ISP
business in mid 2000, using its own direct sequence spread spectrum multipoint fixed wireless technology
that could operate in unlicensed bands at 2.4 GHz and 5.6 GHz. The service worked fine, but customer
adoption proved slow, and it ran out of money, partly because it ran into the fall 2000 capital crunch. There
have also been a number of recent LMDS business failures in Europe, characterized by high infrastructure
cost and few customers.
129 For example, WorkNet, a US provider of high-speeded fixed wireless services to businesses that went
bust this year, found that among more than 800 business customers, less than ten percent were willing to
sign up for more than 384 Kbps of shared bandwidth, even though connections up to 1.5 Mbps were
available. And those that did sign up were often just sharing bandwidth among multiple users, as a cheaper
substitute for separate dial-up connections. SHG Interview, Sanjay Jain, former WorkNet CEO, June 2001.
130 See Wired, April 28, 2001, and For ultra-high bandwidth, see
131 One recent forecast for “wireless LAN hotspots” in the US estimated that there would be 6300 in the US
by yearend 2001, and as many as 114,000 by 2006, servicing up to 20 million wireless laptops and PDAs.
© SHG 2002
- 115
© SHG 2002
See BWCS Consulting, July 30, 2001. This might cut significantly into 3G traffic, since WLAN access is
already 5-6 times faster than 3G.
See the brief history of the videophone written by the US Air Force Communications Agency, Office of
the Historian. Interview with Sheldon Hochreiser, AT&T Corporate Historian, September 15, 2000, CNN.
133 See Total Telecom, “UMTS: What’s That?,” July 18 2001.
134 See Universal Mobile Telecommunications System (UMTS) Forum, The UMTS Third Generation
System – Structuring the Service Revenues Opportunities. (Report No. 9), September 2000,, p.2.
135 Gigabit Ethernet, now being deployed for business users in leading metropolitan markets by companies
like Yipes, can provider data rates up to 10Gbps See, for example, The Yankee Group “, Metro GigE
Providers,” April, 2001.
136 See, for example, PacketVideo and GPIX, two companies that are now developing two-way video
streaming applications for wireless devices.
137 With voice coders running at 8 kbps, and a video application requiring 800 kbps or more, the video has to
generate at least 100 times the revenue per minute as the voice call in order to justify the carrier’s
opportunity cost. As noted above, 1 minutes of video could easily require 6Mb of bandwidth, or 700-800
times the bandwidth of a 1 minute voice call.
138 See Richard Dalton, “The Year of Desktop Videoconferencing,” Byte, December 27, 2000. In 1999, the
worldwide market for ISDN room-based videoconferencing units was just 93,600.
139 There has also recently been progress toward making black-and-white videoconferencing available over
both wired and low-speed (9.6 kbps or less ) wireless connections. See the description of Microsoft’s new
Portrait low-speed wireless video product, USA Today, August 8, 2001.
140 One key implication of this for supporters of
two-way data networks, including ReFLEX™ and
Mobitex™, is that they should also consider making them more interoperable. See below.
141 Cellular Digital Packet Data.
142 Gwcom’s Planet™ is yet another paging-based two way data network. Gwcom, a US-based company
with a strong focus on China, raised venture money in 2000 to launch services in China, and then turned to
a focus on the wireless ASP market.
143 The first version of “Personal Air Communications Technology,” a purported ReFLEX™ competitor,
was released by AT& T Wireless, Ericsson, Pacific Communication Sciences, and other pACT™ alliance
partners in October 1995 about 2 years behind ReFLEX™, aimed at the narrowband PCS market in the US.
Despite purported advantages over ReFLEX™ like location detection, spectral efficiency, and symmetrical
send and receive speeds, it never got any adoption.
144 Note that this omits low-speed two-way data- only networks that are used primarily for telemetry, or
device to device applications. These include the Nexus™ and DataTrak™ networks, as well as the analog
control channel Aeris™ and Cellemetry™ technologies. Note also that an unidentified portion of the twoway data subscribers reported in Chart 11 may be telemetry subscribers – for example, MCI/Worldcom’s
Skytel™ network is reported to have a large number of telemetry endpoints in service.
145 As of July 2001.
146 Red Herring, July 16, 2001.
147 See above, footnote 8.
148 Strategis Group, “State of the U.S. Paging and Advanced Messaging Industry, 2001,” March 2001, 37. As
of yearend 2000, the country’s 37.64 million one-way paging subscribers were divided among service
providers as follows: Arch – 35 percent; Weblink Wireless (direct) – 5 percent; Verizon (resale of
Weblink’s ReFLEX™ network – 8 percent; Metrocall (resale of Weblink Wireless’s ReFLEX™ network –
16 percent; Skytel – 3 percent; all others – 33 percent. Thus directly or indirectly, the national networks run
by Arch and Weblink account for about 64 percent of all one-way paging subscribers. This represents a
major two-way conversion opportunity, as discussed below.
149 See footnote 61 above.
150 “POCSAG,” the “Post Office Code Standardization Advisory Group” standard paging protocol, was a
digital code introduced by a group of international engineers working with the British Post Office in 197681, and was adopted by the ITU as an international standard in 1981. By the late 1980s it accounted for a
majority of the world’s pagers, and still is the predominant standard in some markets, notably in Asia. The
Golay Sequential Code (GSC), named after the legendary MJ Golay, who published his famous half-page
article on error correction in 1946, was another digital code introduced by Motorola in 1983., but it was
© SHG 2002
- 116
© SHG 2002
slower than POCSAG and never achieved much market success. Standard Golay paging receivers operated
at just 300/600 bits/second, while POCSAG ran at 512/1200/2400 bits per second, with 2400 bps in a 12.5
KHz channel. Motorola was an early leader in the production of both POCSAG and Golay pagers.
151 “ERMES,” the Enhanced, or European, Radio Messaging System, a constant 6250 bit/second one way
paging system, was an ETSI-backed European standard for digital one-way paging that supposed to do for
paging what GSM did for cellular telephony in Europe. It was introduced in 1990, offering more capacity
than POCSAG and better roaming capability. Commercial ERMES systems were adopted in France,
Finland and Sweden in 1994, and several other countries, mainly in Europe and the Middle East. But as
noted in Chapter III, most of the European paging operators were severely hurt by “calling party pays,” and
were never very successful. As noted in Chapter III, paging had a difficult time competing in Europe,
especially after the introduction of “calling party pays” rules for cellular telephones and the success of the
GSM standard.
152 From 1.6 kbps to 6.4 kbps.
153 While POCSAG accounted for a dominant share of numeric pagers through the mid-1990s, it was were
increasingly unable to deal with capacity and reliability problems created by the dramatic growth of the
paging market. It basically suffered from three problems. First, its data rate was relatively slow – only 2400
bps per 12.5 KHz of paging spectrum. This lowered its system capacity by requiring more time to transmit
a given amount of information. Second, its protocol was relatively inefficient, with lots of messaging
overhead for preamble messages and synchronization words. Third, it had little fade protection, which
meant frequent retransmissions, especially in mobile applications. FLEX™ was conveniently designed to
be overlaid on to POCSAG, GOLAY and ERMES systems on the same RF channel, side by side – for
example, FLEX™ 1600bps could operate in conjunction with a POCSAG 1200bps systems software
upgrade to the paging terminal and FLEX™ pagers. FLEX™’s key advantages included first and foremost
higher speeds (6400 bps in a 12.5KHz channel for paging, at the FCC’s specified 929-932MHz band, vs.
POCSAG’s 2400 bps maximum), and much better error correction. Higher speeds, plus lower latency, in
turn, meant more users per channel. -- a FLEX™ 4200 system had at least 4-5 times the network capacity
as a POCSAG 2400 system, for numeric paging, supporting up to 600,000 pagers per channel, compared
with POCSAG’s 120,000. Mats Frisk, Ericsson Review No. 1, (1997), “Personal Air communications
technology, ”5. FLEX™ also offered longer battery life, due to the fact that it was a fully synchronous
paging code, allowing the endpoint device to engage only when a message was available, whereas
POCSAG’s was asynchronous, requiring a startup preamble signal to let the system know that a message
was coming. FLEX™ also supported more than 5 billion addresses, whereas POCSAG only supported 2
million. Finally, FLEX™) also had much better fade protection. because it provided for data interleaving.
ReFLEX™ later built on all these FLEX™ advantages.
154 Skytel at the time was a subsidiary of MTEL, based in Jackson, Mississippi. MTEL was acquired by Bell
South Wireless. See below.
155 At the PCS’94 Conference, September 1994. See Motorola Press Release, “Motorola Conducts Industry’s
First Public Demonstration of Two-Way Paging at PCS
’94,” September 22, 1994.
156 See below. It also enhanced capacity by permitting channel layering and sub-zoning.
157 While there had been earlier designs for two-way messaging, Motorola had the clear lead in two-way
licensed paging networks. Nexus Telecommunications Systems had developed a two way system that used a
return channel in the unlicensed 900 Mhz band.
158 The N-PCS spectrum was licensed in blocks of up to 50 KHz. The FCC’s first auction, in July 1994,
raised an unprecedented $614 million, for eleven 10-year nationwide N-PCS licenses.
159 These are Skytel, Arch Wireless, and Weblink Wireless. Skytel appeared at the 1994 auction as the
National Wireless Network, owned by Destineer, a subsidiary of MTEL, which had been awarded a
Pioneers Preference by the FCC in 1992, giving it narrowband PCS spectrum before the auctions. Microsoft
also helped to finance MTEL/ Skytel network. Skytel was later sold to MCI/Worldcom in October 1999.
NWN acquired one of the 50/ 50khz paired national licenses. Arch Wireless bought the Paging Network
Inc., of Plano Texas, acquiring two paired 50/50KHz and one unpaired 50 KHz national N-PCS licenses,
out of the 11 national licenses auctioned in July 1994. Arch acquired PageNet’s assets in 2000. Weblink
Wireless was known as PageMart until December 1999. At the July 1994 FCC N-PCS, PageMart acquired
one of the 50/12.5 KHz national licenses. Note that Craig McCaw’s KDM Messaging Co., later sold to
AT&T, and Airtouch also acquired three national N-PCS licenses, but never fully used them.
© SHG 2002
- 117
© SHG 2002
160 ReFLEX50, deployed by Skytel, permitted messages to be transmitted at speeds up to 25.6kbps on four
6.4 kbps channels in a 50 KHz outbound channel, could receive messages at 9.6 kbps in a 12.5 KHz
inbound channel. In theory this was about twice as fast as ReFLEX25. ReFLEX25 was able to transmit
messages at speeds up to 12.8 kbps with a 50 KHz channel, and .6.4kbps in a 12.5 KHz channel. Its return
channel used 12.5 KHz to transmit at data rates up to 6.4kbps.
161 In addition to the cost of its $80 million N-PCS spectrum, MTEL/Skytel also spent several hundred
million on the cost of developing and building out this network, and $60 million on a network operations
center. All this practically bankrupted the company, and MTEL was compelled to sell out to
MCI/Worldcom in 1999.
162 ReFLEX50 only has a return data rate of 9600 bps – unlike ReFLEX25, it doesn’t have the flexibility to
use lower data rates in areas where coverage is more important than capacity. It also used higher-power
transmitters, which required a very large number of receivers, resulting in a high receiver: transmitter ratio,
on the order of 5:1. This gives Skytel somewhat less flexibility than other ReFLEX operators in reengineering their system to take advantage of ReFLEX Version 2.7, since the high power transmitters
covers a large geographical area, making sub-zoning more difficult. The advent of smart antennas for
ReFLEX (from vendors such as Wireless Online) has recently permitted the balancing of the outbound and
inbound link budgets, reducing the receiver: transmitter ratio to nearly 1:1 for ReFLEX 50, and
eliminating the need for many receiver sites, potentially yielding much lower operating costs for Skytel.
163 The Tenor™ VoiceCoder pagers, supplied by Motorola, weighted 5.5 ounces, had batteries that could last
6 weeks, and stored up to three minutes of voice messages. PageNet offered the pagers for $230, or $10 per
month. It preferred to call them “portable answering machines.”
164 InFLEXION™ required a national 50KHz channel to deliver voice and data at up to 112kbps. PageNet
began testing its VoiceNow™ service in 1995, and launched it commercially in February 1997, with plans
to roll it out nationally by the end of 1997. At the time it was the country’s largest paging operator, with
more than 9 million direct and indirect subscribers. PageNet had developed the service jointly with
Motorola, and had a six month exclusive on the service. Long Distance Digest News, February 24, 1997.
George Mannes, “New pagers let you send messages anywhere,” Popular Mechanics, February 1996.
165 ConXus reportedly spent up to $500 million on a nationwide “voice paging” network, using
InFLEXION™ technology.
Motorola’s voice paging deal with PageNet was announced on April 21, 1997. ConXus also rolled out
trials of voice paging in selected markets in 1997-98.
Glenayre was stuck with about $49 million of ConXus receivables for network infrastructure. See the
May 19,1999, Glenayre press release.
168 The device was a little bulky, but was battery powered and had a Qwerty keyboard and a legible screen.
In July 2001 the original Pagewriter 2000 was added to the Smithsonian’s permanent collection, a tribute to
its role as the first two-way messaging device.
Motorola’s email “VClient,”, for example, introduced in June 1998, provided connectivity to Lotus
Notes™ mail and Microsoft Exchange™ mail. However, it required that users not only run Motorola’s own
Messaging Server on their networks, but also Motorola’s Wisdom OS on their devices. As of 2001,
Motorola’s own Pagewriter 2000x and P935 are the only devices running the Wisdom OS.
Motorola Press Releases,” “Motorola Announces First Generation ReFLEX Chipset Solution,”
September 22, 1998; “Motorola Enables Narrowband PCS Wireless Data Applications with Industry First
ReFLEX™ Chipset,” Nov.2, 1998. See
See Motorola Press Release,”Motorola and Glenayre Sign MOU for Paging Infrastructure
Development,” April 20, 1999. The MOU was the first comprehensive agreement providing Glenayre with
the right to manufacture and sell of Motorola’s paging products, including those pertaining to ReFLEX. A
subsequent expansion of the agreement, announced December 29, 1999, expanded the license to include
ReFLEX wireless modules and chipsets as well paging devices, and also licensed Motorola’s GOTAP
(Generic Over the Air Programming) protocol to Glenayre. During the summer of 1999, Glenayre and
Motorola also jointly hosted the first ReFLEX developers’ conference.
172 CNET, March 22, 2000, for 1999 cell phone numbers in the US.
173 As of the end of 1999, ReFLEX™ had about 750,000 subscribers in the US.
174 As of July 200l, both Motorola and Glenayre continue to be involved in the development of Version 2.7,
however. As noted, Motorola’s decision to turn network equipment for ReFLEX™ over to Glenayre
© SHG 2002
- 118
© SHG 2002
occurred in 1998. Faced with the fundamental fact that few operators in world were upgrading their
FLEX™ networks to ReFLEX™, GlenAyre, in turn, decided to exit the ReFLEX™ network business in
May 2001. Motorola continues to make devices for the network, and to license the ReFLEX protocol to
other manufacturers. Glenayre and Motorola are both doing some development work on Version 2.7 of the
ReFLEX™ protocols
175 Internet email that relies on the SMTP protocol is delivered on a “best efforts” basis, with no automatic
provision for delivery confirmation. Senders may at best request delivery confirmation, but delivery times
can be highly variable, and confirmation is optional on the part of the recipient.
176 This means that after locating a device, the network sends messages for it only to the transmitters in its
locality. ReFLEX™’s architecture provides for inbound receiver to receive the reverse channel messages
from the pagers. It receives on 929-942 MHz and transmits at 896-902 MHz. ReFLEX25 transmitted at 800,
1600, or 6400 bps, and received at 1600, 3200 or 6400 bps. ReFLEX 50 transmitted at 9600 bps.
Estimates vary for this, but some indicate that cellular systems may have as much as forty times the
infrastructure cost per customer at full capacity as ReFLEX.
See Appendix A for more details on frequency reuse and sub-zoning.
179 Link budgets on receive are increased significantly by macrodiversity on receive.
180 The WCTP consortium includes Arch, Metrocall, Skytel, Weblink Wireless, Motorola, Glenayre, Verizon
Wireless, RTS Wireless (now part of Aether), and Mobilesys. See
181 Roaming and interoperability may bring some additional effective increases in capacity to ReFLEX
network, by way of more efficient sharing of capacity across geographic regions.
182 For examples, see and
Another “free” popular instant messaging client developed by Mirabilis, and subsequently acquired by
184 See Appendix A for more details.
“Latency” is defined here and in the glossary as the amount of time it takes for a user to send a message
and receive back the first byte of the response from the network. This varies greatly, depending on specific
conditions, and whether it is operating in a LAN or a WAN environment. For a WAN environment, V. 2.7
is expected to reduce “normal” latency to 7.3 seconds (1.9 inbound, 5.4 outbound) to 13.9 seconds (3.8
inbound, 10.1 outbound). For an “OASIS LAN” situation, in a corporate campus situation for example,
latency may be reduced to as little as 3.5-7.2 seconds. Interviews with Arch Wireless engineers, JulyAugust 2001.
186 Arch Wireless. WebLink Wireless, SkyTel, Metrocall, Motorola, Glenayre, TGA, Verizon Wireless, RTS
Wireless and MobileSys.
Simple Mail Transfer Protocol, oldest and simplest Internet protocol. See the Glossary.
189 For WCTP to provide access to other networks, this would require someone to develop a WCTP gateway
to those networks.
190 For example, there is no need to send an entire IP address over the air – the gateway can do a translation
between standard IP addressing and a much shorter network-specific addressing. Other issues include
TCP/IP’s methods of stepping data rates up and down, which are completely inappropriate for wireless
191 Sun’s Java Two Micro-Edition™ offers a write-once, run anywhere cross-platform application OS that is
designed to provide local processing power. See the discussion of J2ME above in Chapter III.
192 Among the many proprietary, special purpose two-way data networks that we will NOT examine closely
here are Nexus™, Geotek™, Teletrak™, Qualcomm’s Omnitracs™, Siemens/Securicor’s Datatrak,,™
Nextel’s SMR data (based on Motorola’s MIR technology), other mobile data services over SMR (Racom,
Southern Company, Chadmoore), Metricom’s Ricochet™ (an unlicensed spread spectrum technology
operating in the ISM band), and RadioMail (really a gateway service).
193 See Chart 10 for some of those two-way data networks that we will NOT be examining. These include
Metricom’s Ricochet™, Siemen’s Datatrak™, the low-speed analog control channel technologies used by
Aeris and Cellemetry, and Nexus™. While SMS is also available in the US, and is a theoretical competitor,
as noted in Chapter III, its current utilization is very limited because of interoperability issues, so we’ve also
decided to omit it from the short list of serious competitors. See also Chart 13 for the growth of digital CSD,
compared with the leading data-only networks.
© SHG 2002
- 119
© SHG 2002
194 In North America, the first Mobitex™ network was launched in Nova Scotia by Rogers Cantel in 1988.
Rogers Cantel, a subsidiary of Rogers Communications, only expanded its Mobitex™ network to
nationwide coverage in Canada by July 1998. RAM Broadcasting Corp. was formed that same year in New
York. The US network only commenced operations in 1991, however.
195 As of 2001, there are public Mobitex™ networks in 14 countries, including the US, Canada, the UK,
Venezuela, Chile, Korea, Singapore, Indonesia, Australia, Turkey, Belgium, the Netherlands, Finland, and
two in Sweden. There are also private Mobitex™ networks in 8 countries, including Austria (n=6),
Denmark, France, Germany, Italy, Poland, Nigeria, and Australia. There are plans to build new Mobitex™
networks in Brazil and China. Of 29 Mobitex™ public and private networks that have been built, northern
Europe accounts for 16 of them. (Austria alone has 6, Sweden 2).
196 RAM Mobile Data was a subsidiary of RAM Broadcasting Corp., New York. BellSouth Corp. took a
49% stake in 1992. In October 1997, it acquired 100 % of RAM Broadcasting Corp. Before the acquisition,
RAM Mobile Data, in partnership with Bell South International, also launched Mobitex™-based joint
ventures in the Netherlands, the UK, Belgium, and Singapore. The Bell South/RAM Mobile Data JV
launched in the Netherlands in 1993, the UK in 1994, and Belgium and Singapore in 1995.
197 Canada’s Mobitex™ network also grew to 900 base stations during this period. By comparison, Arch
Wireless now also has about 2500 ReFLEX base stations in the US, and will be adding another 700-800
by yearend 2001.
198 Cingular Wireless is a privately–held joint venture of SBC and BellSouth Corp, formed in April 2000.
The business unit formerly know as Bell South Wireless Data, owner of what used to be RAM Broadcasting
and RAM Mobile Data, is now a Cingular subsidiary.
199 Data from Cingular web site;, “A Short History of Wireless Data,” (1996).
200 These are yearend subscriber numbers.
201 RAM arranged with Intel to produce the Intel Wireless PCMIA Modem for Mobitex™ in 1995, but it
sold few units. Ericsson GE Communications, a JV, produced the Modidem AT wireless external modem,
and Motorola also produced a bulky Mobitex modem in the early 1990s.
202 The Palm VII was announced by Palm in December 1998, but did not appear in commercial quantities
until May 1999, at a relatively expensive list price of $599. Early reviews were mixed, especially noting its
inability to serve as a paging like notification device because it was not “always on,” because it couldn’t
receive corporate email or do unfettered Web browsing, and because the wireless services on Bell South
Cingular and Motient were initially quite expensive. PC Magazine, October 6, 1999; Wired, May 21, 1999.
Research In Motion’s original 900 series [email protected] Pager, a two-way product with a Qwerty keyboard
that it first produced for RAM Mobile Data under contract in December 1997, was upgraded to the 950, a
smaller, device with more memory (4 MB of flash, 512 Kbytes of SRAM),, a 2 watt transmitter, a 32-bit
Intel386™ processor, better battery life, a thumb roller wheel that operated similar to a PC mouse, and a
clear LCD display (with backlighting and a 6-8 line display) in 1998. It began shipping to BellSouth
customers in August 1998 and to Rogers Cantel in September 1998. Despite its commitment to ReFLEX™,
PageNET also began reselling the BellSouth RIM-based service in March 1999.That same month, RIM also
signed a contract to deliver an equivalent version, the RIM 850 [email protected] Pager, available to American
Mobile Satellite Corp. (later Motient, as of April 2000) for the Ardis DataTAC™ network that it had
acquired from Motorola in March 1998. It delivered the devices in May 1999, and this time Skytel, another
supposed member of the ReFLEX™ alliance, agreed to resell the American Mobile/Motient service on the
RIM 850. RIM’s even more popular “Blackberry” devices for these two networks – the RIM 957 for
Mobitex and the 857 for Ardis/DataTAC™ -- was announced in January 1999, but did not start shipping in
quantity until mid-year, as its Blackberry Enterprise Server software became available.
203 RAM Mobile Data’s initial messaging charges were incredibly expensive -- $135 per user per month for
unlimited messaging, with $25 and $75 packages available that put stiff ceilings on usage. “Speedy
Wireless Nets,” Network Computing, June 20, 1998, 38. Under Bell South/Cingular, prices finally fall. For
example, the monthly service plan prices for the RIM messaging monthly plans were reduced by about 40
percent from 1999 to 2001. As of 1999, Bell South’s unlimited service plan for the RIM 950 [email protected]
Pager was $99.95 a month, stepping down to a minimum of 25,000 characters per month for $24.95. By
August 2001 it was charging $59.95 for unlimited service, a forty percent reduction, and its entry level price
was $9.95 for 15,000 character, a 34 % per bit reduction. Cingular only introduced an unlimited service
plan for the Palm VII in mid-2000.
204 See for example “Palm VII: A Definite Lemon,” Wired, May 21, 1999.
© SHG 2002
- 120
© SHG 2002
205 UPS rejected both RAM Mobile Data and Ardis in February 1993, in favor of a custom circuit-switched
cellular system that it deployed in conjunction with Southwest Bell, GTE, and PacBell. The loss of this
potential 200,000 endpoint system was a major disappointment for RAM. Motorola designed and
manufactured the custom endpoint devices that UPS deployed on this network, which used cellular modems
to interface with UPS’s own packet-switched network.
206 The Mobitex™ packet-based radio protocol groups data in packets of up to 512 byte packets, and the
maximum message takes 907 ms. So the effective data rate is 4.516 kbps = 512 x (8 kbps/.907 bps).
207 See the network service test reported in Network Computing, July 10, 2000, which concluded that the
typical Cingular Mobitex™ network user would average just 2 kbps in actual throughput. See also Pete
Edmundson, RIM, “Security Issues in Wireless Environments,” Wireless Internet for E Commerce
Conference, April 28,2000, which reports that since the 8 kbps is shared bandwidth, Mobitex users average
just 1.2 kbps.
208 These are yearend subscriber estimates for 1999 and 2000. Of the 688,400 Mobitex™-based customers as
of July 2001, about 225,000 were using Palm.Net, 65,000 sold by RIM’s Blackberry service, 353,000 were
sold by Cingular, and about 45,000 were sold by Aether, a wireless solutions company.
209 About 14 of the 29 Mobitex™ networks in the world today are private networks, for individual
companies or members-only associations.
210 China is reportedly also considering a new Mobitex™ in the 800 MHz range.
211 CDPD can also be packet switched.
Online transaction processing applications – for example, credit card processing.
213 For reasons not clear, the designers of the original Palm VII lost sight of this vital paging-like Mobitex™
capability, requiring the user to take action, raise his aerial and essentially contact the network in order to
receive any messages. RIM, in contrast, built pager-like always-on capability into all its devices. Palm is
now reported to be working on a “RIM killer” that will mimic this always-on feature.
214 With 5 percent concurrency, this implies about 1500-2000 concurrent users per Mobitex™ base station.
215 DataTAC™’s original protocol, still in use for close to half its network, only runs at a maximum of 4.8
216 This is the costs of establishing coverage in any large service area, apart from spectrum costs.
217 IBM’s legendary Systems Network Architecture, of Sears Roebuck fame – perhaps the most lucrative
proprietary protocol in history.
218 Regular POP3 mail, for example, has to be forwarded to a user’s account.
219 See for example the 1997 claim by Bell South Wireless Data: “Mobitex protocol provides a high level of
security. Data transmissions over a wireless packet switched network are much more difficult to capture
than voice transmissions over a cellular voice network. Unlike conversations in the cellular environment,
which are continuous and easily monitored by unsophisticated eavesdroppers, messages in a packet
switched environment are sent in bursts. "Reading" such messages is only possible if the RF interface can
be de-scrambled, a process requiring a level of personnel skill and software sophistication that is
prohibitive. In addition, Mobitex is compatible with customer selected security packages, thus enabling the
user to choose additional security for select messages.” See “The Inherent Insecurity of Data Over Mobitex
Wireless Packet Data Networks,” March 14, 1997, [email protected], newsgroup.
220 See, for example, Alison Campbell, “Mobitex vs. GPRS,”, July 2001: “Mobitex
is also said to be an extremely secure network, a primary requirement for trading exchanges. It is one of the
only networks in the UK used by the Police and emergency services without encryption (although
encryption can be built in).”
221 For example, the combination of frequency agile modems, bit interleaving, and data scrambling built into
the Mobitex™ protocols. They send data over the airlink in short bursts at up to 8 kbps using Gaussian
Minimum Shift Keying (GMSK) modulation, which is encoded and interleaved for error correction, then
scrambled. For efficient channel access, Mobitex™ also uses a TDMA method with a modified slotted
Aloha channel access algorithm, which also complicates monitoring.
222 See “The Inherent Insecurity of Data Over Mobitex Wireless Packet Data Networks,” March 14, 1997,
[email protected], newsgroup.
ReFLEX’s RC4 encryption does comply with the National Institute of Health’s guidelines for health
care data privacy. Basically it amounts to using RC4 to generate secret keys, which all protected wireless
devices and network access points (base stations) share in common. But RC4 – the most widely used secret
key cipher in software applications – has recently been shown to be vulnerable to attack, especially in a
© SHG 2002
- 121
© SHG 2002
wireless context where attackers are able to scoop up lots of encrypted data for analysis. For a recent
example of an attack on 802.11b’s security protocol, which also relies heavily on RC4, see AT&T Labs
Technical Report TD-4ZCPZZ, “Using the Fluhrer, Mantin,, and Shamir Attack to Break WEP,” August 6,
224 For example, Palm.Net uses Certicom’s Elliptic Curve Cryptography for end-to-end encryption for the
Palm VII services that it offers on Mobitex™.
225 Other device hardware manufacturers for the Mobitex™ network include Maxon, Nomadic, CNI, and
Ericsson itself. Middleware providers including RIM, Palm, Cingular, Nettech, Mobix, Infowave, and
226 Our estimate is that ReFLEX’s national coverage is about 95% of the US population, compared with
Mobitex’s 72 percent. Both networks also provide national coverage in Canada, the US’s largest trading
partner. ReFLEX also provides it in Mexico, the second largest US trading partner.
227 The Gemstar TV Guide application was announced on June 4, 2001. Advantra will supply the modems,
and Thomson will manufacture the TV sets.
The former Phillips Electronics subsidiary, now owned by Punch International.
229 For example, CDPD – “Cellular Digital Packet Data+ -- has very low capacity expansion costs, because it
is an overlay network that rides on existing AMPS cellular systems. It is also a full-duplex transfer mode
system, unlike Mobitex™, DataTAC™ or ReFLEX™, allowing its modems to talk and listen at the same
time, which reduces latency. It is also capable of relatively high throughput, at least when voice traffic
permits it, at speeds up to 19.2 kbps, and provides native IP support, unlike all the others. Unfortunately
coverage is lousy, there is no interoperability among the 7 US carriers that offer CDPD services, there’s a
shortage of low-cost devices for it, and it has also managed to earn the moniker, “Capacity Did Prove
Deficient.” Its strengths and weaknesses partly reflect the fact that it was developed in the early 1990s by a
coalition of analog cellular operators (plus IBM, which holds the patents!) that was concerned about
generating higher ARPUs from excess capacity in their voice systems.
Motient’s DataTAC 4000 network also has pretty good coverage, and about half of the US business
population is capable of providing a shared maximum throughput of 19.2 kbps, in the 70-odd cities where it
has implemented its “Radio Data Link Access Protocol.” It is also pretty good at in-building penetration. On
the other hand, its slotted “Digital Sense Multiple Access” (DSMA) protocol reportedly has serious
problems with latency because of the way base stations and mobile devices register with each other.
Ardis’s strengths and weaknesses also reflect its origins. The US Ardis DataTAC™ 4000 network that is
now owned by Motient, had its roots in a proprietary data-only network that built for IBM’s field sales
force in 1983 by Motorola, and was jointly owned by the two companies until 1994. IBM remains the
largest customer to this day, but Motorola and IBM parted ways on their wireless data JV in 1994, when
Motorola bought out Big Blue for $100 million. In 1997 it sold the network to what was then the American
Mobile Satellite Corp, which changed its name to more catchy but perhaps less meaningful “Motient” in
230 See Boeing acquired McDonnell Douglas, the DC-3’s creator, in 1997.
231 ETSI’s GPRS standards effort for GSM networks started in 1994. In the words of one participant, it
followed the “standard telecommunication model – four years to write the standard, followed by years of
implementation, after which we see if it works.”
232 The GSM Forum had been supported a higher-speed circuit-switched technology called HSCSD until
roughly 1998, when it lost ground to GPRS, largely because GSM finally realized the value of a packetswitched network for data – viz, the Internet example.
233 For example, one early report indicated that it only cost Voicestream about $50 million upgrade its US
GSM network. It turns out that this was probably a serious understatement, because it assumed that
Voicestream could provide adequate GPRS coverage with its existing GSM base stations, while in fact it
may require at least 2-3 times as many base stations to get adequate data rates and building penetration.
ETSI, SMG2, 2001. In any case, Voicestream’s required investments will probably be much lower than
those required of AT&T Wireless and Cingular Wireless to upgrade their TDMA networks to GSM and
then add GPRS -- much less Verizon and Sprint PCS, since CDMA2000 requires even more investment in
new base stations and perhaps more spectrum. From the standpoint of total system economics, the
investment cost of replacing non-2.5G handsets will also be substantial – at least equal to the cost of the
© SHG 2002
- 122
© SHG 2002
network upgrades. It is appropriate to include “handset capital” (as well as application capital) in the
accounting, when we consider the cost-benefits of upgrading to 2.5G – after all, consumers (or investors)
will ultimately have to foot the whole bill.
234, August 22, 2001.
235 Dresdner Kleinwort Benson, 2001 estimate
236 In July 2001
AT&T Wireless launched a GPRS pilot in Seattle, offering Motorola’s new Timeport
7382i phone, with plans to upgrade 40 percent of its POPs to GPRS this year, and the rest scheduled for
2002.The AT&T Wireless GPRS pilot started in Seattle on July 17, 2001. In August 2001, Cingular also
launched a Seattle GPRS pilot.
237 For some carriers it also requires more spectrum. Unlike GPRS, however, the CDMA2000 upgrade
supposedly pays for itself very quickly, not necessarily because of increased data traffic, but because it
automatically doubles the voice channel capacity of a carrier’s CDMAOne cellular voice network.
238 SKTelcom launched its CDMA2000 platform in December 2000. LG Telcom followed suit in June 2001.
239 ETSI, Special Mobile Group 2, 2001.
240 This is referred to as the “mobile termination” issue. Given the fact that GPRS customers will be paying
for their services based on data received rather than airtime, the cellular operators argue that permitting
handsets to remain “open” could leave customers vulnerable to junk mail. On the other hand, it would also
permit them to receive data from services that don’t pass through the “walled gardens” run by the operators.
241 Our friends at Motorola delivered the first GPRS handset, the Timeport 260, for the European market in
March 2000. That was widely panned, but Motorola has generally been very aggressive with GPRS, and
now has at least six GPRS handset models in the market, than any other vendor. Ericsson followed in June
2000 with the R520, but it was withdrawn for battery life problems in April 2001. In June 2001 it launched
the T39. Siemens has also recently produced a GPRS handset, the S45. Conspicuously late in joining the
GPRS bandwagon has been the overall cellular handset market leader, Nokia, which apparently bet very
heavily on a rapid transition to 3G, and is playing catch-up. It reportedly plans to introduce the 8390 GPRS
phone in the US by yearend 2001. Until then, Motorola will be a dominant player on GPRS handsets. The
shortage of handsets has already held up GPRS commercial launches. For example, in August 2001
Sweden’s Tele2 pushed back its GPRS launch to later in the fall.
242 As of July 2001, the GPRS handset shortage had become a real problem for European operators. Only
Motorola had one commercially available; Ericsson’s T39, finally delivered in June 2001 after delays, is
still not available in commercial quantities.
243 An alternative tack, taken by vendors like RIM, Compaq (iPaq), and Palm, is to start with basic PDAs,
add card slots for 2.5G modems, and deliver voice through headsets. This may prove more successful.
244 AT&T Wireless’ price for the Motorola Timeport 7382i GPRS phone in its Seattle trial is $199.99, but
this is a subsidized rate.
245 Some software companies, indeed, have identified this relatively slow unadjusted performance as an
opportunity to offer middleware that boosts 2.5G performance. See, for example, the analysis provided of
GPRS speed issues by, White Paper on GPRS, August 2001.
246 For example, Novatel’s new GPRS modem, to ship this fall, supports speeds “up to 53.6 kbps.” In
Europe, where GPRS services have been tariffed for some time, no operator is offering a service greater
than 40 kbps (T-D1 in Germany), most are in the 20-28 kbps range (Viag – 26.8kbps; E-Plus – 20 kbps; D2
Vodafone – 28 kbps), and all use “up to” language to qualify these services.
247 The reader may be surprised to find that the GSM Forum’s own website has really quite critical things to
say about GPRS, for example, relative to the “next big thing,” which it takes to be EDGE or wCDMA. For
example, one white paper on the site makes the point that GPRS’s modulation scheme, GPSK, is decidedly
inferior to EDGE’s 8PSK scheme, resulting in lower bit See
248 For example, AT&T Wireless’s GPRS trial, started in July 2001, offered 1 MB of data sent or received
plus 400 voice minutes for $50 per month, plus incremental data for 3 cents per kilobyte, plus $199.95 for
the phone. Cingular’s pricing for its Seattle trial of its GPRS-based “wireless Internet Express” service
started at $14.99 for up to 100 messages or 500 Kb per month, increasing to $21.99 for up to 500 messages
per month, plus 10 cents per additional message and 7 cents per additional kilobyte.
249 We have examined two usage patterns in providing the estimates for Chart 35. One pattern assumes that a
user stays within the limits of his monthly data allocation – typically 500KB for all these plans – and sends
100 messages a month.. At the other extreme, we assume, based on data from a Gallup Poll of US email
users taken on July 24, 2001, that users reflect the average behavior of the 72% of American adults who
© SHG 2002
- 123
© SHG 2002
now get email at both home and at work, and that they now try to use their GPRS wireless for all their
monthly email traffic. According to this poll, these users now average about 18.7 messages received per
day, of about 6 kb each, and send about 8.3 messages per day. The resulting estimates show their average
costs per message if they converted all this messaging activity to wireless GPRS devices, and did no further
browsing. The resulting estimates show that for heavy messaging, the AT&T and Vodafone plans are much
cheaper. But all of these plans result in average costs per data message no less than ten cents, now about the
average cost of a minute of voice service, and in most cases much higher.
It should be noted that two-way wireless data usage patterns are very similar to these average email users -in July, 2001, the median Arch two-way customer was sending or receiving an average of 18 messages per
day. On the other hand, a sample of 1086 users among PageNet’s November 2000 Mobitex subscribers
showed that they averaged just 123.Kb of messages per month. We would expect two-way data customers
to have much shorter average message lengths than email users, so perhaps these data are not inconsistent.
250 Gallup Poll of email use, July 24, 2001, supra, implies average total Internet email traffic per user per day
for employees who also have email at home of 174 kb per day, compared with the .5MB per month
provided by these GPRS pricing plans. The $5-$12 range sited in the text as an average cost for one day’s
email assumes is based on the assumption that this use is after the first .5MB are exhausted.
251 Assuming an average per .jpg file of 50 KB.
252 This assumes that each wireless page view equals about 1 kb of data, so that these plans provide up to
500 page view per month. This obviously can vary a great deal, depending on Web content and WAP’s
translation capabilities. According the Nielsen/NetRatings (July 2001), the average US internet user that
month did 33 sessions per month, visited 21 sites per session, and viewed 36 pages per session, for a total of
1188 page views per user per month. Assuming that the median user still surfed for $19.95 over an analog
modem, this implies a cost per page view – ignoring the value of email , chat, and other web services – of
about 1.7 cents, somewhat below the average price per page view implied by the initial GRPS pricing
253 See, for example, Morgan Stanley (June 2001), op. cit.
254 For example, above the initial 500 KB offered by AT&T, incremental KBs cost 3 cents. Assuming that
messages average 6 KB each, this implies a marginal cost per message of 18 cents.
255 For example,
one basic Arch Wireless plan offers 200,000 characters per month for $40, and typical
discounts for corporate accounts are at least 20-30 percent lower. Because of ReFLEX’s greater payload
efficiency, each message only averages about ..5 kilobytes or less, so this equates to about 400 messages
per month, at an average cost of $.10 per message. An alternative Arch plan provides unlimited messaging
for $60 per month.. Even before corporate discounts, for the heavy users discussed above who use wireless
devices for all their messaging, this implies an average cost per message of just 7.4 cents, compared with
the $.12-$.36 unit message prices for GRPS shown in Chart 35.
256 This may be another advantage of wireless data-only devices. At least 85 percent of those who have cell
phones try to use them while driving, leading to accidents as often as drunken driving. A 1997 New
England Journal of Medicine study showed that drivers are four times more likely to have automobile
accidents while using cellular phones, and that the risk was the same when drivers used "hands-free"
phones. US case law has already held that employers can liable for employees who have accidents because
of their use of company-provided cell phones, whether or not the phones are being used for business or
personal calls. See Ford & Harrison, Management Update, December 1999.(Vol. 22, No.1): “If cellular
phones are provided by a company or if cellular phone use is a necessary component of a job, employers
can be liable for problems created by employees’ use of cell phones while driving. Employers can incur
liability whether or not the call is personal or business related. Now, at least one lawsuit indicates that
employers should consider banning their employees from talking on cellular telephones while driving.” See
Roberts v. Smith Barney (ED Pa., No. 97-CV-2727, 2/12/99). There is now a nationwide wave of state
legislation on cell phone use while driving, including a recent New York State law that banned the use of
handheld phones while driving. Federal legislation has also been propose, though the cellular operators are
lobbying fiercely against it. See “Drive Talking: Cell Phone Debate Set to Heat Up,” The New York Times,
September 2, 2001.
257 Pete Blackstone, CEO,, April 24, 2001.
258 WitSound View, Survey of Customer Satisfaction Among Leading Wireless Carriers, March 2001.
© SHG 2002
- 124
© SHG 2002
Forbes Magazine, “Cell Hell: Why Wireless Service is a Mess and What You Can Do About It,”
September 17, 2001, 114-117.
260 Michael Uhm, Spectrum Signal Processing, August 2001.
261 See, for example, Vodafone’s July 20, 2001, announcement that it was delaying the European roll-out of
3G services until 2003. On August 22, 2001, the first 3G-related bankruptcy occurred – Broadband Mobile,
jointly owned by Italy’s Enitel and Finland’s Sonera, announced that it would be unable to build out the 3G
licenses it won in Norway, Germany, and Italy, and filed for bankruptcy.
262 See, for example, Christopher Stern, “Demand for Broadband Cooling,” The Washington Post, August
29, 2001.
263 Cynthia Brumfield, Broadband Intelligence Inc., quoted in The Washington Post, supra.
264 The early history of the PC is instructive in this regard. Until there were compelling applications for the
IBM PC – Lotus 123’s easy to use spreadsheet in particular – its early sales, especially to business
customers, were sluggish. Only after Lotus 123 was launched in the fall of 1980 did IBM PC sales really
take off.
265 Cf. the nuclear power industry in the 1950s and 1960s, which also spent hundreds of billions on power
plants around the globe, many of which are now mothballed, and which at one point even contemplated
building nuclear-powered cars, airplanes, ships, and toasters! The nuclear industry analogy also turns out to
be at least somewhat similar on safety grounds – while the jury is still out on the long-term effects of cell
phone electromagnetic emissions, there is no question, as noted above, that vehicle accident rates are much
higher because of them. Indeed, even allowing for the risk of catastrophic accidents like Chernobyl that
obviously don’t apply to the cellular industry, it is likely that there have been far more casualties due to cell
phone-induced driving accidents than to nuclear power accidents.
266 See, for example,
It is also possible to configure a “Break before Make”, in which the device would break from its old zone
before requesting registration in the new zone. This would typically happen in the case of a pager roaming
between network providers, rather than changing zones within a single network.
See “Campus Coverage” below.
“Hot spot” zones can be smaller, however -- see “Hot Spot Coverage” below.
See “ReFLEX Wireless Data Technology”, published by WebLink Wireless in August 2000, for a more
complete description.
To achieve this would also require all the other components of the network to work this quickly. In
many current ReFLEX networks, the transmit controllers work-ahead up to 4 seconds. To achieve a lower
latency, these controllers would need to be upgraded.
This feature may not appear until Version 2.7.2, due in early 2002.
Compiled from general sources, including,, and
© SHG 2002
- 125

Similar documents


Report this document