Breakthrough Ideas (continued)
Disclaimer and Apologia: A respected technologist has informed me that my technical information is wrong in several spots. Because I am highly educated, the danger is that I may combine inaccurate information with misconceptions that sound accurate but are basically wrong. For instance, when I assumed a fixed phone was different from a mobile phone, I was mistaken. The software used to set up the call is different but not the hardware. I appreciate the help of any technologist who can detect such errors and correct them.
Also, yesterday I made a regrettable and preventable error by ascribing the wrong first name to the distinguished scientist and entrepreneur Dr. Andrew Viterbi. I apologize to Dr. Viterbi and any of his friends or admirers whom I may have offended or annoyed.
Moreover, I did not mean to imply that anyone would need my permission to hold a different opinion or to criticize my work. I am neither a technologist nor a financial analyst. I am strictly an amateur in this field, not an expert. My opinions are mere opinions; my beliefs are my own and often incorrect. I do not walk on water. As the jocular expression puts it, “Opinions are like noses, everyone has one.” I have a nose, and, I may be a bigger nose than most.)
The Economics of Mobile Wireless Data.
Wireless voice penetration rates slow when they reach 80% and higher. Yet, price per minute for voice continues to decline. In spite of increased usage, revenues from voice usage are declining. Thus, wireless data throughput necessarily becomes the key driver for long-term wireless economics. Not only is data use increasing around the world, where it will inevitably dominate in both wired and wireless networks, but also the use of data applications creates a stimulus effect that increases the use of voice minutes.
Assuming an equivalent quality of service, for a wireless carrier, the significant economic issues in selecting 3G technologies are the overall capacity of the RF system, the costs of associated infrastructure and operation, and the unit cost of delivery.
Average throughput for a 5 MHz sector of bandwidth varies as a function of the access mode and whether the mobile user is standing, walking, or driving. Simulations revealed average throughput in Kbps for either 1x or WCDMA were approximately similar: respectively 1,350 or 1,500 when Stationary; 1,050 or 1,100 when Pedestrian, and 1,050 or 900 when Vehicular. Whereas, throughput for 1x EV was far better: 4,200 when Stationary; 3,240 when Pedestrian; and 1,590 when Vehicular. In comparison, GPRS was theoretically 80 Kbps in each environment.
Qualcomm forecasts that mobile wireless data demand will exceed 200 megabytes per user per month by the year 2006. With 5 MHz of spectrum available for data traffic, they estimated the network cost to deliver data traffic would fall from several dollars per megabyte using 2G technologies to less than $0.03 using 1x EV-DO and to less than $0.07 for 1x and WCDMA. But, GPRS would cost approximately $0.42. A user consuming 200 megabytes of data per month would cost a 1x EV-DO operator $4 per month in network expense, compared to $12 per month with 1x, $14 per month with WCDMA, and $83 per month with GPRS. (The data above are from a Qualcomm white paper entitled “The Economics of Wireless Data.”). qualcomm.com
Because it has been optimized for data, 1xEV-DO yields significant throughput and cost-per-megabit advantages. This higher throughput enables richer, more compelling data service offerings at significantly lower costs.
Irwin Jacobs’ recent presentation at GoldmanSachs provided both voice capacity and data throughput figures that yielded interesting comparisons to their respective baselines. qualcomm.com
The Erlang is a standard measure of voice capacity. For comparative purposes 2G and 3G systems can be compared (using erlangs per sector in 10 MHz) to the baseline voice capacity of 1G AMPS (9 Erlangs/sector (E/s)). TDMA increased voice capacity (37 E/s) 4 times over AMPS. GSM increased capacity (31-51 E/s), either 3 times or 6 times (using the AMR vocoder) over AMPS baseline. CdmaOne (134 E/s) increased voice capacity 15 times. WCDMA (111-176 E/s) may increase voice capacity either 12 times or 20 times (using the AMR vocoder) over AMPS. CDMA2000 1x (231-329 E/s) uses Selectable Mode Vocoders, which can dynamically tradeoff voice capacity or voice quality to provide 26 times, 32 times, or 36 times the voice capacity of AMPS. In 2003, after adding dual diversity antennae, 1x Release A will provide 378-427 E/s, selectable for either 42 times or 47 times the voice capacity of the AMPS baseline. Clearly, 1x is the 3G-voice-leader¾the first-mover, offering not only the best voice capacity, but also continually expanding capacity.
For average throughput in kbps for 10 MHz sectors of spectrum, average data throughput comparisons used GPRS as a baseline. The baseline GPRS provides 261 kbps average throughputs in 10 MHz. EDGE provides 783 kbps, which is a 3 times improvement. WCDMA provides 1800 Kbps, 7 times as much as GPRS. CDMA 20001x provides 2450 kbps, 8 times or 9 times (with antenna diversity) times as much as GPRS. And, 1xEV-DO provides 8,028 kbps, 27 times or 31 times (with dual diversity) that of GPRS. Thus, because its simplified the architecture and optimized it for data only, not only does 1xEV-DO have a tremendous throughput advantage, but also no mobile wireless competitor is scheduled to emerge for years.
These simulations were based on defined theoretical assumptions rather than comparative testing in operating networks, which would favor Qualcomm’s voice and data solutions even more because its engineering implementations wring out rates that are closer to their theoretical ideals because of their beautiful algorithms. In particular, the ugly sister, GPRS (and its big brother, to-be-or-not-to-be EDGE), has a substantial disadvantage in throughput and costs because: (a) it is a software extension of 2G voice technology; (b) as a time-division technology, it is limited to 8 slots, but only 4 can be used because of collisions and interference; (c) each time-slot is limited to 10 kbps because more sophisticated coding schemes require better signal to noise ratios; and (d) its 3/9 or 4/12 frequency reuse pattern limits the number of RF carriers. After troubled roll outs in 90 markets in numbers so small that they are not reported, GPRS continues to run hot, eat batteries, emit radiation, and multiply base stations while under performing technically and economically.
Given the controlled one-step migration of CDMA to 3G contrasted with the eventual necessity of changing TDMA/GSM to the 3G CDMA RAN, after adding the additional sunk costs of following either a one- or two-step path from GSM or a 3-step path from TDMA, what are the comparative costs of upgrading? According to Sprint PCS, who offered capital cost comparisons at the last 3G World Congress, on the one hand, the upgrade to either 1x or 1x-EV-DO each cost less than 10% of CDMA’s original 2G-network cost. On the other hand, using Sprint’s estimates taken from the Yankee Group, the upgrade cost from TDMA or GSM to GSM/GPRS is 28%, to Edge is 51% more, and to WCDMA/UMTS, it is another 90% of TDMA/GSM’s 2G-network cost. However, in practice, more base stations than expected are required, reflecting the geometric problems in GSM RAN designs discussed earlier. Once again, costs of capital expenditures and operation (because of all those new GSM base stations additions or the migration to new WCDMA RANs) favor CDMA2000.
Capital expenditures depend on both intergenerational compatibility and spectral efficiency. Given dense data traffic, ideally, GPRS requires 42 cells/sq Km; WCDMA and 1x requires 3, and 1xEV-DO, only 1. Poor frequency reuse, failure to synchronize, and ugly technology are economic handicaps. The outcomes of these comparisons of estimated voice or data capacity, capital expenditures, and network cost per bit, all favor Qualcomm’s CDMA2000 solutions. For data, the remarkable cost/performance ratio of 1xEV-DO becomes better than that of dial-up modems and competitive with the long-awaited, slow to roll out, fixed broadband solutions.
Usable Wireless Internet.
In the early 1990’s, wireline Internet data volumes averaged one Megabyte/user/month. Average data volume increased to about 200 M/u/m by 1999 from the combined effects of introducing Netscape’s graphical browser interface and the steady increase in wire line modem-data-transfer speeds. Assuming a temporal pattern of growth similar to the Internet’s, wireless mobile data volume is projected to exceed 200 M/u/m by year 2006.
Because increased usability and higher data rates combine to drive penetration and usage, with 1xEV-DO, the wireless Internet becomes a reality. Useable access requires availability of higher data rates, low cost delivery of high bandwidth, and ubiquitous coverage. For example, it take 28.8 kbps to use graphics, 56 kbps to make the Web fully accessible, 128 kbps for multimedia and large downloads, and 384 kbps for almost full network transparency. And, the price of all-you-can-use bandwidth must approach a competitive fixed price of $40 to $50 per month.
Increased usability stems from continual improvements in wireless access, applications, and devices. However, igniting indirect network effects always poses a chicken-or-egg dilemma; in this instance, increasing usability requires simultaneously improving all three complementary elements.
Qualcomm’s trident tactics for shaping increasing returns to scale among these three complementors includes: (a) increase usable access by improving performance, lowering cost of delivery, and promoting ubiquity; (b) increase usable applications by enabling access to full media-rich content, providing a common BREW development platform and applications interface, and enabling unique applications using its Multimedia Launchpad and SnapTrack position location technologies; and (c) increase usable devices by enabling the availability of a wide range of wirelessly enabled devices, improving their cost/performance ratios, and making high capability devices increasingly affordable. Sustained advances in its ASICs, whose cost/performance ratio, pursuant to Moore’s Law, plunges off a Cliff of Costs, facilitate a burgeoning growth of usable devices.
In 2002, Qualcomm believes wireless data volume can be enabled by the prevalence of: (a) MPEG4 compression; (b) Color-Screen mobiles and PDAs; (c) laptop modem cards; (d) emerging data applications; (e) Extensible Markup Language; and (f) low-cost media-rich services. In 2003, data volume drivers include: (a) complex voice recognition; (b) widespread deployment of position location technology; (c) mini-keyboards; (d) reasonable cost phone-PDAs; and (e) ubiquitous High Data Rate coverage.
CDMA2000 1xEV-DO can deliver: (a) a 3 megabyte MP3 file in three minutes at a cost under $.07; (b) a 6 Mb medium-resolution 2-minute video clip at about $.13; (c) 205 Mb a month at a cost of just $4.50; and (d) in areas underserved by broadband alternatives, wireless Internet at $.22 per Mb. This is only the beginning.
From an end-user’s perspective, what counts is not data-in-itself, but what you can do using it. Given abundant access and a supply of usable devices, applications will multiply, fostering a wireless WEB lifestyle of human connectivity and interaction. Best of all is always having access to what you need or want on the road or in daily life when and where you need it.
Dr. Paul Jacobs contends that the Wireless Internet is greater, even grander, than the wired Internet because it not only includes it, but also is mobile, providing new time- and location-sensitivities that extend well beyond the tethered and bound Internet simply because they are always on and always with you.
Breakthrough ideas create wealth by engendering multiple innovative business solutions. Qualcomm shapes a data market with diverse opportunities, including new competitive spaces: (1) mobile business applications; (2) consumer info-tainment; (3) digital picture sharing; (4) streaming multimedia; (5) wireless Internet; (6) position location applications; (7) push-to-talk QChat; (8) movie e-mail; (9) safety-security; (10) telemetry; (11) vehicular telematics; (12) VOIP; (13) wireless connection and integration with wi fi and BlueTooth networks; and (14) all wireless IP networks.
As a thought experiment, imagine various combinations--recipes--of technologies as building blocks--software-as-codified-knowledge--and then add this Greek chorus refrain--always on, always with you. For example, imagine a 9/11 firefighter at ground zero with push-to-talk QChat plus streaming media of breaking news events--always on, always with you. Imagine a grandmother forced off the road into a Florida swamp with position location--always on, always with you. An off-duty soldier in Afghanistan with photo or movie-email of the birth of a child or his kid’s first soccer goal or homerun--always on, always with you. Imagine a nanny cam capturing a video of someone abusing your child--always on, always with you. Imagine me smiling with the Library of Congress to download--always on, always with you. Imagine medical smart cards, telemetry, and help-lines for 911-emergencies at remote auto/train/airplane crash sites--always on, always with you. Looking ahead five or ten years, what innovative business ideas can you imagine as--always on, always with you? |
