Table of Contents An Overview Defining What 3G May Deliver A Review of Conventional 3G Migration Paths Comparing the Migration Paths Summary & Conclusions Disclaimer Acknowledgements Preface Download pdf

3 A Review of Conventional 3G Migration Paths

3.1 Introduction

To understand the 3G migration options open to TDMA/IS-136 operators, we begin by examining the currently conceived migration paths for the major 2G technologies: GSM, cdmaOne, and TDMA/IS-136. This chapter reviews and expands parts of our earlier discussions. Following this review, we examine the possibility of CDMA2000 1X as a migration alternative for TDMA. We do not review PDC, a 2G technology that is unique to Japan and which will be superceded by UMTS and CDMA2000 1X.

One assumed migration for TDMA/IS-136 posits an initial transition to GSM, subsequent adoption of GPRS and EDGE and final deployment of UMTS, the generally accepted 3G standard for GSM.

Examining this assumed path uncovers what may be possible barriers to TDMA operators who deploy GSM. Because of these, some TDMA operators may find it useful to reevaluate the full cost benefits of the GSM option and to consider CDMA2000 1X (also called CDMA 1xRTT) as an alternative for enabling 3G services. This will be particularly so for operators who are assigned to 800 MHz spectrum.²⁴

3.2 The Migration Path for Current GSM Operators ²⁵

The migration path for current GSM operators envisions GPRS and EDGE enhancements to GSM technology followed by the transition to UMTS on new spectrum. The deployment of GPRS and EDGE may take place on 900, 1800, and/or 1900 MHz spectrum, on which GSM is, at present, deployed. The migration assumes the availability of multi-mode/multi-band handsets that will enable seamless interoperability between GSM (including GPRS-EDGE) and UMTS, which, into the foreseeable future, will be deployed on 1900 and 2100 MHz spectrum.

3.2.1 General Packet Radio Service

General Packet Radio Service (GPRS) is considered the first step in the 3G transition. GPRS enhances the GSM network by overlaying packet architecture onto the current circuit switched architecture. It enables GSM operators to gain experience with operating packet networks, billing for packet traffic, and delivering packet-based IP applications in what will be a mixed circuit switched-packet environment. In theory, GPRS enables mobile networks to connect to the Internet at speeds of up to 115 kbps. As we observed in our previous chapter, the likely reality will be rates of 10–40 kbps, although 50 kbps may be possible. As we observed in our previous chapter, BT is now experiencing data rates of only 8 kbps. This, however, should increase as the technology matures.

The cost to deploy GPRS is but a fraction of the cost to deploy UMTS. In concept, GSM operators will be able to incorporate GPRS infrastructure into future UMTS systems. This will mitigate the risk of GPRS becoming an orphaned technology and a stranded investment. In effect, this makes GPRS infrastructure (albeit not terminals) appear “free” to GSM operators who are planning the 3G transition. GPRS requires a dual-mode GSM-GPRS terminal. Once economies of manufacturing scale are reached, the earlier classes of GSM-GPRS terminals will be only marginally more expensive than conventional GSM terminals.²⁶

3.2.2 Enhanced Data [Rate] for Global Evolution

Enhanced Data [Rate] for Global Evolution (EDGE) is being positioned as a complement to GPRS. EDGE would enhance the air interface to the GSM Network. In theory, EDGE integrated with GPRS would enable data rates as high as 384 kbps. In the TDMA/IS-136 world, EDGE would likely be deployed as an integral part of a new GSM network. For this reason, deploying a “greenfield” GSM-EDGE network would, in concept, be simpler than integrating EDGE into an established GSM network.

However, deploying EDGE would require more than a software upgrade. EDGE has different modulation characteristics from either TDMA or GSM. Because of this, GSM-EDGE may require changes and/or additions to the hardware sub-systems of cell sites. These could include amplifiers, combiners, and isolators. EDGE may also require changes to established reuse patterns. This would imply changes to base station antennas. Of particular importance, EDGE has a 4–7 dB weaker link budget than GSM. Compensating for this would require additional base stations.²⁷ For such reasons, the implementation of EDGE may be more complex than some may have initially envisioned.

EDGE would use the same frequencies as GSM-GPRS and would require a tri-mode GSM-GPRS-EDGE handset. Some observers point to the engineering challenge of imposing a packet architecture onto aggregated GSM time slots. The extent of this challenge is suggested by the delayed deliveries of commercial GPRS handsets and their so far limited data throughput. Not inconsistent with this observation, no vendor has yet demonstrated a prototype EDGE handset. For these reasons, it remains uncertain of how soon, if at all, EDGE handsets would become commercially available and if they become available what data rates they would deliver.²⁸ There is the further issue of how much GSM-GPRS-EDGE handsets would cost.

Given the above, even assuming that EDGE handsets become commercially available, some GSM operators will not adopt EDGE, but will migrate from GSM or GPRS directly to UMTS.

3.2.3 Universal Mobile Telephone Service

Universal Mobile Telephone Service (UMTS) is the accepted 3G standard for GSM operators. UMTS requires paired 5 MHz RF channels, four times as wide as the paired 1.25 MHz channels required for CDMA2000. For this reason, UMTS is sometimes referred to as “wideband CDMA” (W-CDMA). By migrating to UMTS, operators will gain access to additional spectrum as well as the greater capacity and expanded functionality of the new technology. UMTS incorporates a more efficient variable vocoder (codec). In common with CDMA2000 1X, this vocoder will increase the voice capacity of a given amount of spectrum. As we have already noted, outside of the Americas, UMTS is being deployed on the 1900 MHz (uplink) and 2100 MHz (downlink) frequencies. Because of this, some operators, primarily those in the Americas who now use the 1900 MHz frequencies for PCS, would be unable to migrate to UMTS.²⁹ Allocation of other frequencies for UMTS may or may not be possible. The well-publicized failures of U.S. operators to acquire frequencies at 700 MHz (occupied by TV broadcasters), 1700 MHz (occupied by the military), or 2500–2600 MHz (occupied by educational broadcasters) provide examples.

With the sole exception of Japan (which is constructing stand-alone UMTS networks), UMTS operators will use multi-mode and multi-band terminals. Such terminals will enable seamless handoff between what will be fully deployed GSM-GPRS (or GSM-GPRS- EDGE) networks and partially deployed UMTS networks. Seamless handoff will enable operators to buildout UMTS networks as the technology matures and as demand evolves, much as 800 MHz operators in North and South America built out digital networks as extensions to their analog networks during the mid- to late-1990s.

In sum, the transition to UMTS will enable the advantage of a gradual infrastructure investment closely tailored to demand. However, it will introduce the disadvantage of complex and expensive terminals.

3.3 The Migration Path for TDMA/IS-136 Operators

Originally, the migration from TDMA/IS-136 to 3G was to parallel that of GSM to 3G. TDMA operators were to have overlaid packet based GPRS onto their TDMA infrastructure. Following this, they were to have introduced an EDGE RF Interface. However, with AT&T’s adoption of GSM, that migration path changed. Currently, it is assumed that TDMA operators will first deploy a GSM network on to their currently assigned 800 and/or 1900 MHz spectrum. This will overlay or parallel their established TDMA network. Following, they will take the path of GSM operators and migrate their GSM networks to GPRS, possibly to EDGE, and finally to UMTS. In theory, this change in migration path will enable TDMA operators to benefit from the R&D advances and economies of scale enjoyed by the GSM world.

In concept, this approach makes sense. However, it presents at least four implementation challenges. All four apply to TDMA/IS-136 operators who occupy the 800 MHz frequencies. Two apply to TDMA operators who occupy the 1900 MHz frequencies.

1. GSM equipment is, at present, available only for the 900, 1800, and 1900 MHz frequencies. Neither infrastructure nor terminals are yet available for the 800 MHz frequencies on which major TDMA/IS-136 operators have deployed. Nokia, Ericsson, Motorola, and Nortel have promised GSM 800 infrastructure. However, they have yet to produce it. This means that TDMA operators assigned to the 800 MHz spectrum cannot begin their 3G migration until vendors deliver GSM 800 infrastructure and, more importantly, GSM 800 terminals. In the U.S., this challenge would most affect Cingular, which of the TDMA operators holds the most 800 MHz spectrum. In Latin America, it will affect virtually all TDMA operators.
2. Because of cost and complexity, manufacturers are unlikely to produce AMPS-TDMA- GSM terminals. For example, the Siemens S47, due to be launched in the U.S. during the fourth quarter of 2001, will offer multi-band and dual-mode TDMA-GSM capabilities. Because of costs, it does not offer analog AMPS capability.³⁰ Siemens has no plans to produce AMPS-TDMA-GSM phones.³¹ However, many 800 MHz TDMA/IS-136 operators have not extended their TDMA/IS-136 networks into rural areas to provide the same coverage as their analog networks. As an example, as of year-end 2000, Canada’s Rogers AT&T Wireless reported analog coverage for 93 percent of Canada’s population, but digital coverage for only 83 percent.³² More importantly, in urban areas, the digital coverage of AMPS-TDMA networks may be incomplete. This is due to the greater attenuation of discernible voice conversations, which is characteristic of GSM and TDMA/IS-136. In such digital “holes,” whether rural or urban, as the TDMA call degrades, subscribers may hand off from TDMA to AMPS. Without analog capabilities, GSM-TDMA phones will preclude such hand-off. As a consequence, 800 MHz TDMA operators may have to deploy more dense (and, therefore, more expensive) GSM cell coverage than initially anticipated. Without such deployment, they may risk providing their new GSM subscribers with inferior service. Siemens acknowledges this challenge. However, Siemens foresees that TDMA operators will quickly build out GSM systems to cover such holes.³³ This coverage issue does not arise with cdmaOne. The CDMA RF link was designed to provide coverage equal to or greater than that of AMPS. Field tests have confirmed the design.³⁴
3. There is no current way to transfer the rich portfolio of GSM applications and services to a TDMA/IS-136 network. Thus, even with multi-mode GSM-TDMA phones, GSM subscribers will be unable to use their GSM services when they roam onto TDMA networks. Vendors and operators are exploring means of overcoming this challenge, under the acronym of GAIT. However, the question remains of if and when the necessary GAIT infrastructure and, especially, the GAIT handsets will be delivered. We discuss this further in our next chapter.
4. In most countries of North and South America, where TDMA/IS-136 is primarily deployed, the lower part of the 1900 MHz spectrum, which elsewhere is allocated to UMTS, is being used or is reserved for Personal Communications Services (PCS). This means that no spectrum is available for TDMA operators, whether using 800 or 1900 MHz, which would enable them to migrate beyond GSM-GPRS-EDGE to UMTS. Brazil is an exception. There, ANATEL, Brazil’s regulatory body has allocated PCS to the 1800 MHz GSM frequencies. While ANATEL has made no formal decision on 3G, its allocation of PCS services to 1800 MHz has left the 1900 and 2100 MHz frequencies open for UMTS.

AT&T, which precipitated the move of TDMA/IS-136 operators toward GSM, may be a special case. Unlike Cingular, the other major TDMA operator in the U.S., AT&T has amassed a disproportionate amount of 1900 MHz spectrum in the 12 largest metropolitan areas.³⁵ AT&T holds 1900 MHz licenses in 9 of these 12 largest markets.³⁶ Cingular holds 1900 MHz licenses in only four of them.³⁷ Thus, in adopting GSM as its migration path to 3G, AT&T may be responding to what it recognizes as a competitive advantage, one which Cingular, a major competitor, cannot match.

In sum, the concept of adopting GSM as the 3G migration path for TDMA/IS-136 is clear. However, in practice, there are barriers to doing so. For 800 MHz TDMA operators, there is the uncertain availability of GSM 800 equipment, in particular handsets, and the potential need for an unexpectedly dense GSM network. For both 800 and 1900 MHz TDMA operators, there is the potential inability of subscribers to use their GSM services when roaming on to TDMA networks. In addition, there is the uncertain future availability of spectrum on which UMTS can be deployed.

Not to be overlooked are the issues of handset costs. Dual-mode GSM-TDMA handsets will be inherently expensive. They will be useful for only a niche market. As such, they will not benefit from the economies of scale, which, in general, will be enjoyed by the GSM community. We discuss these economies in our next chapter.

3.4 The Migration Path for cdmaOne Operators

The initial migration path for cdmaOne operators extends from the current cdmaOne (also called CDMA/IS-95 or CDMA/IS-95-A), optionally to CDMA/IS-95-B (deployed only in Japan, Korea, and recently Peru), and then to CDMA/IS-95-C or CDMA 1xRTT (One Times Radio Transmission Technology). CDMA 1xRTT is frequently abbreviated to CDMA2000 1X or sometimes CDMA2000 1x.

The next evolutionary step is to CDMA2000 1x EV-DO (Evolution-Data Only). While the timing of its deployment is not yet firm, Sprint anticipates its commercial availability in early 2003. Sprint is cautious in approaching EV-DO, wishing to observe how the data market may unfold before making a commitment.³⁸ Others anticipate that EV-DO will become commercially available in late 2002. CDMA2000 1x EV-DV (Evolution-Data and Voice) lies further in the future. During the first quarter of 2001, Motorola demonstrated the technology and together with Nokia, Philips Semiconductors, and Texas Instruments proposed a standard. At that time, the group anticipated that the standard would be set in May 2001.³⁹ The end of year 2001 now appears more likely.

3.4.1 cdmaOne/IS-95-A

cdmaOne/IS-95-A supports circuit-switched voice and circuit- or packet-switched data at speeds of up to 14.4 kbps. Due to the early focus of vendors and operators on voice, cdmaOne/IS-95-A historically has been used exclusively for circuit-switched voice and recently for a small amount of circuit-switched data.

3.4.2 cdmaOne/IS-95-B

cdmaOne/IS-95-B supports circuit-switched voice and packet-switched data. KDDI in Japan and SK Telecom in Korea have deployed it since 1999. It provides theoretical data rates of up to 115 kbps, with generally experienced rates of 64 kbps. cdmaOne/IS-95-B is now being superceded by the higher capacity and faster CDMA2000 1X and is unlikely to be deployed elsewhere.

3.4.3 CDMA2000 1X

CDMA2000 1X was historically called phase one of the 3G-migration for cdmaOne. As we observed in Chapter One, the ITU considers it to be 3G. It supports circuit-switched voice and packet-switched data on the same RF channel. Theoretically, CDMA2000 1X, Release A, enables data rates of up to 307 kbps or above, depending on RF environment. The former is a 10-fold increase over the 14.4 kbps provided by cdmaOne and complies with the accepted performance standard for 3G.

In October 2000, Korea’s SK Telecom (SKT), using Samsung equipment, launched the world’s first commercial CDMA2000 1X service on its currently occupied 800 MHz spectrum. At that time, it announced plans to cover all urban regions of the country during the second quarter of 2002.⁴⁰ In May 2001, Korea’s LG Telecom (LGT) and Korea Telecom Freetel (KTF) launched commercial CDMA2000 1X on their currently occupied spectrum. At that time, LGT stated that it would cover the entire country by year-end.⁴¹ In the U.S., Verizon and Sprint PCS will deploy CDMA2000 1X by year-end 2001.⁴²

The Korean launches illustrate both the flexibility and full commercial availability of CDMA2000 1X. SKT and KTF have received 3G licenses to deploy UMTS on 1900 and 2100 MHz spectrum. Nonetheless, pending commercial delivery and testing of UMTS infrastructure and handsets, they are first deploying CDMA2000 1X on their current spectrum.

3.4.4 CDMA2000 1x EV-DO

CDMA2000 1x EV-DO supports packet-switched voice and packet-switched high-speed data on separate RF channels. The voice channel facilitates the low latency necessary for transmitting two-way conversations. The data channel enables the flexible routing and low-cost transmission advantages of a packet network. CDMA2000 1x EV-DO provides theoretical data speeds of up to 2.4 Mbps. In concept, using separate channels for voice and data requires more bandwidth than using a combined channel. In practice, the spectrum disadvantage diminishes as data traffic increases. This will be especially so for operators with larger spectrum assignments and large data throughput.

Of particular value—and in some cases not fully recognized—the migration of cdmaOne to CDMA2000 1X and beyond provides a more flexible use of spectrum compared to the migrations from GSM to UMTS and TDMA/IS-136 through GSM to UMTS. Under present concepts, GSM will not be available for the 1900 and 2100 MHz frequencies allocated to UMTS. UMTS will not be available for the 800, 900, 1800, and 1900 MHz frequencies allocated to GSM. However, operators can deploy CDMA2000 1x EV-DO (and eventually EV-DV) either on newly available 1900 and 2100 MHz spectrum and/or on currently assigned 800 and/or 1900 MHz spectrum.⁴³ As we observed earlier, SKT, LGT, and KTF in Korea have deployed CDMA2000 1X on current spectrum. The Japanese operator KDDI intends to deploy CDMA2000 1X on newly available spectrum. Most operators will deploy CDMA2000 1X on current spectrum.

This flexible use of spectrum is an advantage of CDMA2000 1X. By enabling operators to use their current spectrum, CDMA2000 1X can save them the overt costs of bidding for new 3G spectrum or, in the case of “beauty contests,” the covert costs of petitioning for it. The latter can be considerable, especially when they include onerous conditions for network construction. Sweden, for example, did not charge for 3G licenses. However, it did require each license recipient to spend what would have been $3 billion or more for constructing full nationwide networks within two years of the license award.⁴⁴ (The regulator has since eased this burden by allowing the license recipients to share up to 70 percent of the 3G infrastructure.)

Operators who deploy CDMA2000 1X on currently assigned spectrum do not gain the added capacity that new spectrum provides. However, this disadvantage is to some extent overcome by the more efficient coding algorithm, which CDMA2000 1X deploys. This algorithm doubles the theoretical capacity of cdmaOne, although in practice the capacity gain without voice degradation will be closer to 50 percent.⁴⁵ EDGE and UMTS will also deploy a more efficient coding algorithm and realize the associated capacity gains. Because GPRS is a network architecture, not an RF interface, it cannot provide capacity gains.

3.5 CDMA2000 1X as an Alternative for TDMA/IS-136 Operators

Our preceding discussion has briefly compared the currently assumed 3G migration paths for TDMA/IS-136 and cdmaOne. It uncovered four potential barriers for TDMA operators who may adopt GSM.

1. The uncertain availability of GSM 800 infrastructure and, especially, TDMA-GSM terminals.
2. The uncertainty of whether TDMA-GSM terminals will enable access to GSM applications and services on TDMA networks.
3. The possible need to deploy a denser GSM network than initially envisioned.
4. The uncertain availability of spectrum at 1900 and 2100 MHz that would enable migration to UMTS.

TDMA/IS-136 operators who occupy the 800 MHz frequencies will face all four barriers.

For TDMA/IS-136 operators who occupy the 1900 MHz frequencies, GSM infrastructure is available and Siemens, at least, is promising TDMA-GSM terminals by year-end. These operators have deployed networks to provide acceptable digital coverage. However, in common with TDMA 800 operators, even when TDMA-GSM terminals become available, it is uncertain whether they will enable subscribers to access GSM applications and services on TDMA networks. Likewise, TDMA 800 operators face the uncertain availability of UMTS spectrum.

By enabling the 3G transition on currently occupied spectrum, whether at 800 or 1900 MHz, CDMA2000 1X overcomes these uncertainties. CDMA2000 1X infrastructure and terminals are available for 800 and 1900 MHz frequencies. All CDMA2000 1X terminals for 800 MHz have an analog mode, thereby assuring coverage of non-digital network holes. Operators already hold their spectrum. Finally, the 50 percent or greater efficiency of CDMA2000 1X and its derivatives provides capacity for higher bandwidth applications as well as more voice.

These are meaningful advantages. For these reasons, TDMA/IS-136 operators may find it useful to reevaluate the full cost-benefits of GSM versus cdmaOne and to consider cdmaOne and its derivatives for enabling 3G services.

Our following chapter compares the GSM and cdmaOne alternatives in further detail.

 

24.	“The radio spectrum is that portion of the electromagnet energy spectrum used by radio waves. ‘Radio services’ are categories of radio use. Setting aside frequency ranges, or bands, in the radio spectrum to or radio services is a government function called spectrum allocation or frequency allocation.” From Bennett Z. Kobb, Spectrum Guide, New Signals Press, Falls Church, Virginia, 1994, p. 8. In this white paper, we use the terms “spectrum” and “frequency” interchangeably.
25.	The Shosteck Group, Third Generation Wireless (3G): Why, When, and How It Will Happen, Wheaton Maryland, November 1999, pp. 194-205.
26.	GPRS terminals are defined in terms of “classes.” Each class specifies a permutation of up-link and downlink time-slots, which the terminal aggregates to provide higher data rates. The GPRS standard specifies 29 different classes, although this number will not be manufactured. The earlier classes of GPRS terminals will specify lower data rates and, once produced in volume, will likely cost little more than GSM handsets. The later classes will specify higher data rates, which will require more components. These will likely cost more noticeably more than GSM handsets.
27.	“EDGE400 Compensates EDGE’s 4-7dB Weaker Link Budget,” Slide presentation, (GSM400General.PTT/ver 2.0 3.1.2000, Jla), Nokia, Helsinki, 1999.
28.	“Handsets Hold the Key to Survival of Edge,” 3G Mobile, December 12, 2000, pp. 1-2.
29.	Only part of the allocated PCS frequencies overlap the allocated UMTS frequencies. Depending on frequency assignment, some PCS operators may be able to deploy UMTS at 1900 MHz and others not.
30.	Press release, “Siemens Challenges U.S. Mobile Phone Market with GSM/TDMA Handset to Capitalize on Carriers’ Interest in Globally Dominant GSM Standard,” Siemens AG, Las Vegas, March 21, 2001.
31.	Personal communication, Martin Fichter, Director of Products Marketing, U.S., Siemens, San Diego, California, April 19, 2001.
32.	2000 Annual Report 2000, Rogers-AT&T Wireless, Toronto, 2001, p. 1.
33.	Personal communication, Martin Fichter, Director of Products Marketing U.S., Siemens, San Diego, California, April 19, 2001.
34.	Personal communication, Dr. Charles Wheatley, Senior Vice President – Technology, QUALCOMM Inc., San Diego, California, May 25, 2001.
35.	These include Atlanta, Baltimore, Boston, Chicago, Dallas, Detroit, Houston, Los Angeles, New York, Philadelphia, San Francisco, and Washington DC.
36.	Personal communication, informed source, AT&T Wireless, Seattle, Washington, May 31, 2001.
37.	Personal communication, Jeff Cannon, Director of Investor Relations, Cingular, Atlanta, Georgia, May 31, 2001.
38.	Personal communication, Oliver Valente, Chief Technology Officer, Sprint PCS, Kansas City, April 24, 2001.
39.	Press release, “Motorola Successfully Demonstrates Industry-First CDMA 1xEV-DV Solution in Lab,” Motorola, Inc., Arlington Heights, Illinois, March 19, 2001 and “Motorola Transmits Live Video in CDMA 1xEV-DV Test,” RCR Wireless News, March 26, 2001, p. 33.
40.	Hung Song and John S. Csapo, 3G in Korea, CDG Digevent Series, Samsung, Seoul, April 17, 2001.
41.	Nathan Lynch, “South Korea Delivers on 2.5G Promise,” WapWeek, May 3, 2001.
42.	Dan Meyer, “Sprint Says It Has Ample Spectrum for 3G Services,” RCR Wireless News, March 26, 2001, p. 34.
43.	In the unique case of Korea, currently allocated spectrum is at 1700 MHz.
44.	Almar Latour, “Sweden Shocks Telia by Rejecting Its Bid for New Wireless Licenses,” The Wall Street Journal, December 18, 2000.
45.	Personal communication, informed source, Seoul, Korea, May 29, 2001.