If the two-year technology recession had not intervened, 2002 might have been the year that reconfigurable computing found its application space in software-defined radio. Even with the slowdown, SDR is moving onto the industry's radar screen, as the SDR Forum leaves its roots in military applications and hosts its first trade show and technical conference in San Diego in November (www.sdrforum.org), with semiconductor proponents of adaptive computing as sponsors.
Meanwhile, legal barriers to SDR deployment have fallen. The FCC last September approved many uses of SDR within the network, a decision influenced to a great degree by the SDR Forum. The time should be ripe for bringing more reconfigurable digital signal processing into both basestations and handsets.
Indeed, when Pact XPP Technologies Inc. sponsored a reconfigurability forum in Santa Clara, Calif., last spring, SDR was touted as the application space that mattered. With digital cellular networks shifting to 2.5-generation data services yet continuing to rely on multiple air interfaces, the base for a broad market for reconfigurability seemed certain.
If only. For several months now, Forward Concepts Inc. president Will Strauss has argued that handset manufacturers are under such pressure to reduce production costs, they must aim for the simplest and lowest-priced baseband and RF components available. A baseband processor that handled GSM, TDMA and CDMA air interfaces might make sense if a high-end multimode phone had a significant user base. But Strauss maintains that the reconfigurable phone represents a minuscule portion of the market.
This might not be the case if carriers had been quick to roll out true 3G services. Many, particularly in Europe, would like to do so, based on the debt burden they face from billions of dollars' worth of 3G licenses they obtained at auction. But customer demand is not there for 3G broadband data, pushing 3G out into the latter part of the decade and even delaying 2.5G services like General Packet Radio Service.
Handset manufacturers are in a quandary as well. Practical competitors have shrunk to Nokia, Motorola, Ericsson and Samsung. Of these developers, Ericsson is relying on outsourcing, and its bonds were reduced to junk status in late July by Moody's Inc. While new Asian manufacturers could enter this space as 2.5G and 3G services take off, future players like Hutchison or Huawei likely will be under greater cost constraints than the current crop of handset OEMs.
For a time, the one certain application space for reconfigurable platforms appeared to be next-generation basestations for advanced digital cellular services. Because space and power dissipation are at a premium in basestations, and because these nodes employ banks of DSPs operating in parallel, designs would be well served by host processor boards that could be dynamically reconfigured to support the different channel capacities of TDMA, CDMA and GSM.
But that target market is in trouble, too. On July 31, In-Stat/MDR Inc. (Scottsdale, Ariz.) and Allied Business Intelligence Inc. (Oyster Bay, N.Y.) simultaneously released market studies of advanced basestations that held little solace for SDR advocates. ABI sees the bulk of new basestation designs moving to 2.5G and 3G support, but cautioned that the numbers will be affected by a slow buildout of infrastructure due to the economy.
In-Stat analyst Ray Jodoin was even more pessimistic, titling his study "Down Down Down We Go." Jodoin predicted that the compound annual growth rate for aggregated digital cellular markets will remain at 13.6 percent through 2006. While this may sound somewhat healthy compared with wireline markets, In-Stat predicts that the impact on infrastructure OEMs will be severe. At least two OEM suppliers in the basestation market will not survive the next year, In-Stat predicted, and possibly more will fail if carriers continue to consolidate. The 2002-03 basestation deployment rate in North America will trail behind global averages, the study said, and will be small enough to be almost negligible.
A basic gating factor in wireless communications remains similar to those in optical and copper-wireline communications, analyst Strauss said at the Pact XXP Technologies forum in the spring. Semiconductor developers may be able to show specific cost saving in moving to a new generation of technology. But if carriers have moved into crisis mode and halted all acquisition of infrastructure equipment, even the arguments for reduced capital and operations expenses in the network will fall on deaf ears.
Despite a bleak economic picture, contributors to this week's In Focus p make a case for the economic advantages of adaptive computing, and also reveal the competitive technical edge of SDR systems. Peter Varhol, a technologist based in Nashua, N.H., breaks down the infrastructure from an economics standpoint. And, Chris Dick, director of signal processing engineering at Xilinx Inc. (San Jose, Calif.) makes a case for the FPGA as a "naturally parallel-processing engine that can take advantage of the rich parallelism in a soft-radio PHY."
In our exclusive online coverage, Mike Buchanan, director of marketing at Elixent Limited, discusses reconfigurable algorithm processing as the enabler for SDR to become a cost-effective solution for handsets. Analog Devices' Zoran Zvonar and his colleagues, show how SDR "will come in a series of steps moving from hardware selectability to hardware reconfigurability and finally to true software reconfigurability." And, engineers from QuickSilver Technology explain the benefits of a adaptive computing in two compute intensive wireless areas: CDMA2000 and W-CDMA searcher acquisition sections.