Often the simplest questions asked are the hardest ones to answer. For example, ask any group of technology-savvy colleagues if third-generation, or simply 3G, networks exist today. You'll get three different answers: yes, no, and maybe. Why is there this disparity? The definition of 3G has become a moving target. The response therefore depends upon the market segment questioned—whether it's telecom or datacom—and the country in which it's asked.

Why should anyone be concerned about the fate of 3G? Because it may shape the fast-growing 802.11b wireless local-area-network (WLAN) market—and be shaped by it. The struggles of these two powerful, yet quite different wireless networks—the much-heralded 3G and the dark horse known as Wi-Fi—will determine the direction of wireless technology for years to come. As a result, design engineers, market investors, chip manufacturers, and infrastructure vendors are closely watching this evolving struggle.

A ROSE BY ANY OTHER NAME
3G wireless technology is the next evolutionary step in telecom cellular networks. It promises increased bandwidth for the packet-based transmission of digital voice, data, and video communication. The Universal Telecom-munication Union (UMT) has issued guidelines for 3G that specify transmission speeds of 144 Kbps inside a moving vehicle and 2 Mbps at a fixed location.

As an evolving technology, 3G has several variations: Wideband Code Division Multiple Access (W-CDMA), CDMA-2000, and EDGE. Japan's NTT Docomo and Manx Telecom, located in the United Kingdom's Isle of Mann, have deployed W-CDMA networks with limited success.

One of the chief technical problems with the deployment of W-CDMA has been the lack of system testing, notes Dave Whipple, Agilent Technologies' (www.agilent.com) Wireless Industry Technologist. As he states, "It is not clear that lab testing has been done for W-CDMA." Whipple explains that more cooperation is needed between network-equipment vendors and terminal devices to ensure interoperability between the whole system.

Another component of 3G implementation is CDMA-2000. This variation has been deployed by Verizon and Sprint PCS in the U.S., along with several different carriers in Korea.

The third main 3G component, known as EDGE, serves as a follow-on to GPRS and an upgrade to GSM. Its networks are being deployed in the U.S. by AT&T Wireless, Cingular, and Voicestream (now called "T-Mobile"). EDGE allows any user three times the data rate over GPRS. It requires a substantial amount of network reuse, however.

System complexity in both the handset (terminal) and networks (infrastructure) are challenging even the best wireless designers. Many 3G handsets, for example, require two processors: one for the digital baseband and another for applications. Both processors must operate at increased throughput rates, while maintaining a total power consumption that's equivalent to 2G handsets.

These are conflicting requirements. Higher throughput rates typically mean higher processor clock speeds. This, in turn, translates into higher power requirements. Balancing throughput and power rates adds even more pressure to the standard handset design criteria of reduced component size, time-to-market, availability, and cost.

Eva Skoglund, OSE Systems' (www.ose.com) Product Manager of Wireless Communications, agrees that system complexity is a hot issue. She says that designers must decide how many CPUs and DSPs are needed while partitioning the functionality for each processor. "Handset designs must include a very complex baseband chip—Bluetooth chips and high-end parts that must handle multimedia, high-bandwidth data, and Internet connections." This task is certainly not an easy one.

Dual processors also create challenges in the design of subsystem components, like memory. Baseband and application cores, for example, perform different computing tasks. These tasks lead to both volatile and nonvolatile memory requirements, observes Mario Fazio, Micron's (www.micron.com) Director of Strategic Marketing for Wireless Products (see figure). Fortunately, new Flash devices, such as Micron's V-SyncFlash memory, can reside on the synchronous SDRAM bus instead of a separate Flash bus.

Even with dual processors, however, the computing power and throughput may not be enough. Using the traditional approach of a DSP and ASIC combination, how does a designer provide the high capacity needed to handle complex algorithms? Add to this challenge the ever-changing standards specification, and the result is a designer's nightmare.

"It takes too long to get it right using traditional ASIC technologies," observes Bob Plunkett, Director of Product Management for QuickSilver (www.qstech.com). "By the time you have one version figured out, the standards bodies have moved onto the next version." The result, Bob argues, is that only companies with very deep pockets have the resources to develop 3G—and even then they struggle. Quicksilver's adaptive computing machine (ACM) and SilverC languages make it possible to develop new processing technologies that keep pace with evolving standards and design complexities at a reasonable cost.

The 3G base-station-infrastructure side faces similar problems. Here too, traditional DSP+ASIC combinations are struggling to keep pace with the processing power needed in 3G. In order for equipment manufactures to build profitable businesses, they must use new techniques to drive down the cost-per-channel across the network, explains Ravi Subramanian, President of Morphics Technology (www.morphics.com). Morphics has developed a new class of processor, called the wireless signal processor (WSP), that drastically improves the traditional ASIC+DSP hardware-software partitioning used in 3G baseband processing.