AT&T, Nortel plot '4G' wireless nets

AT&T, Nortel plot '4G' wireless nets
DENVER — Amid projections of a boom in wireless data, top technologists from AT&T Labs and Nortel Networks sketched out their first efforts to define fourth-generation cellular networks in separate presentations this week.

The companies also pointed to key technologies — such as software radios, wideband transceivers and a new breed of power amplifier — needed to deliver wireless networks with cellular data rates of 20 Mbits/second and beyond. Many of those technologies were addressed by technical papers at the Radio & Wireless Conference (Rawcon) held in Denver Sept. 10-13.

Rawcon keynoter Al Javed, chief technology officer for access networks at Nortel (Brampton, Ontario), predicted that more than 1 billion users will be accessing the Internet wirelessly by 2004.

AT&T demonstrated in New York an asymmetric network it called 4G Access that combines existing Enhanced Data Rates for GSM Evolution (Edge) technology for an uplink with wideband orthogonal frequency-division multiplexing (OFDM) for the downlink. The goal: to speed downloading of packet data, particularly for streaming audio and video.

"There is a developing industry consensus on focusing on improved downlink performance for 4G," said Michael Bamburak, vice president of the technology development group at AT&T Wireless. AT&T believes multicarrier OFDM will provide wireless downlink access at up to 10 Mbits/s for stationary systems and more than 384 kbits/s at 800-kHz bandwidth in a high-mobility environment.

Deploying the 800-kHz OFDM 4G network will require wideband software radios and advances in digital signal processors. DSP cores might need built-in channel-coding cores, and the ability to handle multipath fading and fast Fourier transforms to extract the multiple tones from the OFDM signal, said the director of the 4G program, Nelson Sollenberger, division manager for Wireless System Research at AT&T Labs (Red Bank, N.J.).

AT&T has already begun a two-phase upgrade of its wireless network on the way to 4G Access. The first phase will involve deployment of software at cellular basestations. A second phase, likely to occur in approximately two years, will involve aggressive hardware deployment of smart antenna technology being developed by AT&T and its OEM partners.

Cellular data surge

Despite the fact that 4G networks are still several years out, testing and planning are necessary now if carriers are to meet demand for what researchers call an intense hunger for wireless high-speed data services. Forecasts suggest that by 2005, 50 percent of cellular subscribers will be data-capable and handsets will surpass PCs as Internet access devices.

"Messaging will be the primary driver of wireless-data adoption over the next few years," said Dave Jackson, senior wireless analyst for Cahners In-Stat (Scottsdale, Ariz.). "We estimate that the number of wireless messages sent per month will balloon from 3 billion in December 1999 to 244 billion by December 2004."

Mobile commerce and location-based services will follow messaging as growth vehicles, Jackson added. However, In-Stat warns that the success of wireless data will depend on building out a new infrastructure that can handle location-based services, high levels of security, new payment options and detailed billing. "After that, the march toward wireless data will become a stampede," said Jackson.

Nortel's Javed detailed a feature list for Internet protocol-based 4G networks with data rates up to 20 Mbits/s. Such nets should sport aggregate data throughput per cell of 100 Mbits/s (forward); a spectral efficiency (bits/s per hertz per cell per carrier) of 20 forward, six reverse; and a dormant-to-active transition time of 0.1 second. These figures up the ante from 3G Universal Mobile Telecommunications System's (UMTS) specs of 3.8-Mbit/s throughput, spectral efficiency of 0.8 and transition time of 2 seconds.

Technologies needed to hit those specs, Javed said, include wideband receivers, multibeam antenna systems and space-time coding, followed by better power amplifiers, advanced modems, RF transceivers and multi-user detection. Papers at Rawcon outlined advances in several of these areas.

Designers lack a formal methodology to implement software radios on reconfigurable platforms. Researchers at the Mobile and Portable Radio Research Group at Virginia Tech sought to solve that problem by proposing in one Rawcon paper a Layered Radio Architecture using stream-based computing, which allows over-the-air updates and software validation.

"The design is based on an FPGA that was developed in-house called Stallion, which supports fast run-time reconfiguration," said Srikathyayani Srikanteswara, one of the researchers. "The stream-based approach gives a uniform, modular structure to the processing modules and defines the protocol for interaction between various modules, while the layered architecture makes it possible to incorporate all of the features of a software radio while minimizing complexity."

Power amplifiers, meanwhile, could surge to 70 percent of the cost of 4G basestations without innovative design and channelizer power savings, Javed said. Techniques such as digital predistortion would compensate for nonlinearities, a big power dissipater and contributor to poor system performance.

"Digital predistortion is a result of accurate modeling of LDMOS power devices and could replace other compensation techniques such as RF feed-forward for greater accuracy," Javed said. DSP-based algorithms could also be used, but the overall goal is to use digital correction techniques to lower cost and improve efficiency.

Researchers at Stanford University described at Rawcon a method of maintaining continuous regulation of oscillator phase alignment for the two decoupled feedback loops in Cartesian feedback. Cartesian feedback is a linearization technique that holds much promise if stability can be maintained. But instability and phase error can occur over time due to temperature and process changes if the system isn't properly adjusted.

The Stanford team outlined a way to regulate the local-oscillator phase alignment — without the use of DSP techniques. "Poor phase regulation is one of the major barriers to the use of Cartesian feedback," said researcher Joel Dawson. "Our system could greatly reduce that barrier and allow it to operate — maintenance-free — in a hostile environment."