Shosteck Group White Paper on TDMA 3G Migration Paths  |
|||||||||||||||||||||||||||||||||||
 |
|||||||||||||||||||||||||||||||||||
|
GSM or CDMA: The Commercial and Technology
Challenges for TDMA Operators
2 Defining What 3G May Deliver 2.1 Introduction The mobile industry has been cursed by its own hyperbole. Whether driven by exuberance or guile, the press releases of manufacturers and network operators alike have set unrealistic expectations of what 2.5G and 3G can economically and commercially deliver, especially over the near term. This press release mantra has centered on 384 kbps or more in a pedestrian and/or mobile environment. Such rates would enable the promised full-motion video and multi-media in real time. Such data rates have not yet been delivered commercially, where costs are an
issue and the limitations of a real world radio frequency (RF) environment are
real. They can be delivered within idealized test sites or test networks where
a handful of friendly users on dedicated spectrum assures that interference
from other network traffic or inconvenient geography is minimal. However, in
the commercial world, costs must be considered and RF must be transmitted under
the imperfect conditions of frequency congestion, background RF noise, challenging
geography, and incomplete network Surprisingly, the first commercial network has achieved data rates close to the theoretical ideal. Korea’s SK Telecom (SKT) launched CDMA2000 1X service in October 2000. As of late May 2001, SKT had 180,000 CDMA2000 1X subscribers. The network provides data rates of 120 kbps in an ideal RF environment, but more typically 70–90 kbps.7 The experiences in Europe and the expectations in North America are more constrained. In both areas, members of the industry, in particular network operators, are beginning to recognize and acknowledge that while the theoretical throughput, and long-term potential, of new technologies may range from 115 kbps to 384 kbps, the near-term reality will be less. In the case of CDMA2000 1X, Richard Lynch, Executive Vice President and Chief Technology Officer of Verizon Wireless, the largest U.S. wireless operator, anticipates mobile data rates of 30–40 kbps over the near term, notwithstanding a theoretical expectation of 144 kbps.8 In the case of GSM-GPRS, David Williams, Vice President for Strategic Planning for Cingular, the second largest U.S. wireless operator, anticipates near-term data rates of 20–30 kbps, notwithstanding a theoretical expectation of 115 kbps.9 Some European observers look to GPRS delivering up to 50 kbps10 although others believe that 20 to 40 kbps will be more likely.11 Over the long term, the technologies will mature and the networks will be built. With these, the data rates will increase. However, until that time, the advantage of 3G will center not on high data rates but on its packet switched architecture. As we observed in our introduction, in contrast to circuit switching, packet switching enables instant and low-cost IP access to the Web and the plethora of applications and services available from it. As we discuss at the end of this chapter, regardless of the data rates that technologies may be able to deliver, economic and commercial concerns will drive operators to provide data rates that are less. John Roth, CEO of Nortel Networks, first advanced the cost benefits of packet architecture in February 1999 at the annual convention of the Cellular Telecommunications and Internet Association (CTIA). There, he announced Nortel’s goal of using packet architecture “to drive the overall cost of operating a mobile network down by an order of magnitude…from 37 cents today to only four cents within five years.”12 More than two years later, in May 2001, Conrad Labante, Nortel’s Director of Wireless Internet Business Strategy Development, reiterated Mr. Roth’s expectations, that packet architecture would reduce the costs of delivering data by a factor of 10. However, he still saw that day as at least five years into the future.13 Nokia, a leading proponent of 3G, doesn’t “believe that the data issue is data rates,” but rather relevant applications and services. In the initial stages of GPRS and EDGE deployment, Nokia sees data rates of 20–30 kbps as enabling “a lot for money making applications.” Nokia doubts that video will provide “money making applications in the beginning.”14 In sum, by focusing on high data rates, network operators may be misdirecting their resources. Into the foreseeable future, the key issues in deploying 2.5 and 3G will more likely center on the cost-benefits of given data rates for provisioning services and applications that subscribers will use and, thereby, generate revenues for network operators. We explore this further at the conclusion of this chapter. 2.2 Factors That Constrain High-Bandwidth Transmissions An issue of cost-benefits aside, multiple factors limit high-bandwidth transmissions in a real-world environment. Most, but not all, stem from technology immaturity. Some factors may be more relevant to one RF interface than to another. Over time and with sufficient R&D efforts, such factors will be overcome. Until they are overcome, such factors preclude consistently achieving anything near the data rates being claimed. We briefly discuss three of these factors below—power budgets and battery drain, latency, and bit error rate. These discussions are not exhaustive. Rather, they exemplify the range of challenges that wireless vendors and operators must surmount before high-bandwidth transmissions in real time can become a commercial reality. 2.2.1 Power Budgets and Battery Drain For GSM, conventional voice is transmitted on a “time slot” at 9.6 kbps. High bandwidth is commonly advertised as up to 384 kbps—40 times more than 9.6 kbps. Regardless of RF technology, such transmissions require greater power budgets, and therefore produce more battery drain. This will pose a problem for all new technologies, albeit less so for CDMA2000 1X and UMTS than for GSM, GPRS, and EDGE. The transmission power required for all CDMA technologies, including CDMA2000 1X and UMTS (also a CDMA technology), is continuously adjusted to the bits being transmitted. If, at a particular moment, few bits are being transmitted, as in a short message, less power is used. If more bits are being transmitted, as in a video clip, more power is used. This is commonly called a variable data rate. Thus, CDMA2000 1X and UMTS are more efficient, in that they allocate only the RF power needed to deliver the bits being transmitted. Battery drain will vary accordingly. That said, average battery drain will be relatively less than for non-CDMA technologies. GSM technologies, including GPRS and EDGE, also vary their power output to match the number of bits being transmitted. However, they are less adaptive in how they match power output to bits rates. In concept, they can add or subtract up to eight time slots, depending on the nature of the transmission. Because of this less refined matching of power output to bit rate, the battery drain will generally be higher than for CDMA-based technologies that transmit the same content. This phenomenon is apparently occurring on the GPRS network recently launched by BT Cellnet, where subscribers are complaining about battery drain.15 2.2.2 Latency Latency describes the delay of a transmission from the time it enters the network until the time it leaves. Low latency means short delays. High latency means long delays. Latency may occur in the handset or in the network. Latency that occurs in the handset or between the handset and the base station is called access latency. Latency that occurs from the base station through the network is called network latency. Low latency is essential for real-time transmissions. These include live voice conversations (but not voice mail messages, which are time insensitive) and live two-way video (but not entertainment video clips, which also are time insensitive). Latency is not a phenomenon only of mobile networks. It is an outcome of all the networks, terminals, and devices through which transmissions may pass and the bottlenecks (and, therefore, delays) they may encounter. At home, users of broadband Internet connections experience latency as delays in downloading websites during peak traffic hours (often in early evenings and during inclement weather). Such delays are due to overloading bandwidth at the network periphery. More important are delays due to overloading bandwidth at the network core. A user in New York may download a website that is hosted in Seattle. Depending on traffic loading and transmissions costs at the moment, the download may travel from Seattle to Los Angeles to Denver to Houston to Chicago and finally to New York. It may use fiber networks owned by Qwest, AT&T, and/or Verizon. At each switching point, and in particular at the juncture of each network, it will encounter delay. Each of these delays increases the latency. This means that even if a mobile network is configured to provide low latency, the operator cannot guarantee low latency for end-users who use their mobile devices to access other networks or who use their terminals in a noisy, and thereby latency inducing, RF environment. 2.2.3 Bit Error Rate Voice is more forgiving of transmission errors or high bit error rates (BER) than is data (also called non-voice content).16 The reason for this forgiveness comes from the ability of the human brain to reconstruct the missing parts of conversations. If a syllable or even a word is dropped, the brain assumes the missing information from the context of the conversation and reconstructs it. This enables people to communicate through the noise or the break-up of a marginally understandable mobile radio call. The transmission of data does not provide this reconstruction advantage. If non-voice content is lost, it can only be recovered, if at all, through sophisticated error correction algorithms. Such error corrections add overhead to RF transmissions. This overhead slows the true data rate, or transmission rate, of the desired content. The greater the number of errors, the more error correction required and the slower the true data rate. Thus, in an imperfect RF environment, any RF technology will deliver but a fraction of its theoretical peak data rate. In addition to complaints of excessive battery drain on its GPRS system, BT Cellnet is also experiencing lower than expected throughput on its GPRS system, in some cases as low as 8 kbps.17 As we observed earlier, as other operators follow BT’s lead, slower than expected data rates will be the norm, not the exception. To one extent or another, each of these issues applies to all transitional and 3G technologies. We present them as considerations, which may be useful for network operators in evaluating their migration alternatives. 2.3 The Cost to Deliver Theoretical Data Rates Into the foreseeable future, the cost to deliver theoretical data rates will prove economically and commercially unsustainable. To provide anything near ubiquitous and consistent 384 kbps in real time will require enormous investment by operators. Not to be forgotten, operators must expand RF capacity to provide such data rates for non-voice content without compromising the RF capacity needed to carry conventional voice traffic. If a network is at capacity, adding data capabilities without expanding network capacity will undermine the quality of service for conventional voice users.18 If data traffic causes voice service to degrade, it will alienate voice subscribers. In the case of CDMA2000 1X, the costs of networks capable of carrying high data rates without degrading voice quality have not been published. The UMTS costs have been published as an outcome of the “beauty contests” for 3G licenses. Without shared networks, these would be as much as $400 per population covered, although $200 per population covered might be sufficient.19 Such costs would be three to five times those needed to provide acceptable voice service. By way of comparison, from 1983 through December 2000, the cumulative investment of all U.S. operators (up to seven per market) had reached $89.6 billion.20 Given a population of 276 million, this equated to $325 per population covered. At present, capital expenditures of $200 to $400 per population covered are commercially non-viable. Given the flight of capital from the telecommunications industry, this situation is unlikely to change soon. Thus, into the foreseeable future, mobile operators will invest less in 3G infrastructure than required to deliver 384 kbps on a ubiquitous and consistent basis. This precludes providing the publicized nirvana of full-motion video and multi-media in real time.21 This means that over the near term, the commercial relevance of ubiquitous 3G must stem from something other than high bandwidth. In most respects, this will focus on the cost perspectives of Nortel and the applications and services perspectives of Nokia. Other vendors may or may not share these perspectives. 2.4 An Alternative to the High Data Rate 3G Business Model In 10 years and beyond, the general provision of high-bandwidth wireless traffic will plausibly become commercially viable. However, over the next five years, and likely longer, the high costs of the required infrastructure will preclude operators from providing 384 kbps, or anything near it, on a ubiquitous and consistent basis. This will prevent ubiquitous and consistent delivery of full-motion video and multi-media in real time. The business case for 3G will not come from providing massively high bandwidth. Rather, it will come from providing (1) cost-efficient mixed voice and data traffic and (2) immediate, open, and low cost connectivity to the Internet. The latter will provide access to infinite sources of infinite applications and services. Most will be available through relatively low bandwidth. The success of NTT DoCoMo’s I-Mode points to this potential. Notwithstanding data rates of only 9.6 kbps, I-Mode has been adopted by most of DoCoMo’s subscribers. Average revenue per user (ARPU) has increased by up to 30 percent or more.22 This is not to say that a data rate of 9.6 kbps will be sufficient. Over time, it will not be. As 3G networks and their 2.5G proxies are launched, they will enable bandwidths greater than 9.6 kbps, albeit at less than the rates promised by industry hyperbole. As we pointed out earlier, operators who have spoken, place these in the range of 20–40 kbps. Into the foreseeable future, this may emerge as an industry benchmark. Data rates greater than this will prove an extra dividend. At present, SKT is unique among world operators as the only one with meaningful experience in the operation of a commercial 3G network. Based on this experience, SKT is perhaps more aware and more thoughtful than other operators over the trade-off between the provision of commercially feasible services, on the one hand, and high data rates, on the other. As we observed earlier, SKT is now delivering 70–90 kbps to 180,000 commercial CDMA2000 1X subscribers. This is out of an approximate customer base of 12 million. Notwithstanding this technical achievement, SKT understands that economics and not technology will limit the data rates that its customers eventually experience. SKT observes that with only 180,000 3G customers, it does not yet have sufficient traffic to experience problems with loading or deployment. However, SKT foresees an eventual trade-off in voice versus data capacity. For this reason, SKT foresees economical data rates of 30–50 kbps—what it will be able to deliver cost-effectively for every customer while maintaining acceptable quality of service. SKT believes that this will be the case whether on its CDMA2000 1X network or on a UMTS network. SKT does not see cost differences between the infrastructure for the two 3G technologies. SKT’s 3G problem is not technical but commercial “converting 3G services to money.” To do so, it is striving to provide “reasonably good” quality of voice and data services in order to “satisfy…customers reasonably not absolutely.” SKT perceives that attempts to provide high data rates and absolute customer satisfaction are dangers. To do so, it would spend too much money, thus reducing its chance for profits.23 In sum, SKT sees the central 3G issue not as high data rates but as profitable data rates. SKT believes that a focus on data rates, and the technologies that may deliver them, detracts from what for SKT is their basic issue—to generate profits through providing 3G services. To the extent that other operators recognize and adopt this point of view, they may shift their long-term perspective. Rather than focusing on a distant and commercially uncertain goal of data rates of 144 kbps and beyond, they may refocus on an immediate and commercially profitable goal of data rates at 30–50 kbps. To the extent that this refocus takes place, current TDMA/IS-136 operators may pay greater attention to cdmaOne as a transition to 3G. 
|
|||||||||||||||||||||||||||||||||||
 |
 |
|
|
||||
|
||||
|
|