Existing wireless cellular networks were designed to transmit circuit-based voice signals across the air. This task is a very different undertaking than transmitting data or packet-switched information. Third-generation (3G) operators and vendors must now face this challenge as they try to shoehorn the data-intensive Internet into legacy wireless networks.

Over the last 18 months, wireless service providers and their vendors have promised mass-market customers and the enterprise and financial markets a wireless data experience that would surpass all others. This experience would include always-on broadband connectivity, live video to the phone, and lots of "killer applications" for which the customer would gladly pay.

People are rightly beginning to question these beliefs. Another reality is that the marketing hype came before the demand. The broadband experience just isn't there. It's too costly and too big a pill for the enterprise to swallow. As a result, there's great skepticism about the viability of the wireless data market. It would seem that the wireless industry has only itself to blame.

A year ago, the market experienced the beginning of a trend in which 3G license holders became increasingly aware of the reality of what 3G can and cannot deliver. Telefonica (www.telefonica.es) and Sonera (www.sonera.com)—two of the largest international wireless carriers in Europe—wrote down the investments that were made in their European 3G licenses. Other carriers have since followed this trend. Now, those carriers are re-examining the potential returns on building a 3G network versus going in a new direction. The reality has set in: 3G is great for increased voice capacity and a little bit of data. But it's not so great for enterprise-grade broadband data. With that type of data, the combination of average—not "burst"—downlink and uplink speeds of 500 Kbps to 1 Mbps (with latency below 100 ms) are entry-level requirements. These performance levels comprise the minimum that is needed to unwire all office-productivity applications.

According to a Spring 2003 report by Deutsche Bank Securities, the carriers that are attempting to roll out 3G networks have faced many challenges. This includes the high cost of deployment, technology delays, and a lack of demand. Although it was once thought to be the future of broadband wireless data, the deployment of 3G technologies worldwide is encountering many problems. Most of these challenges stem from the simple fact that 3G doesn't yet deliver on the experience or the low cost that are needed for mass-market or enterprise adoption.

Jason Chapman, a Gartner Group European Wireless analyst (www.gartner.com), was quoted in an interview with RCR Wireless News a few months back. He stated, "Consumers in North America may never take up 3G because it will never meet or exceed their PC experience. It's up to the mobile carriers to set expectations and to avoid over-complicated data services."

Does this mean that the idea of true broadband wireless data is dead? No, but these announcements are a clear signal that something must change before mobile broadband data becomes a reality. In the Yankee Group's (www.yankee
group.com) January 2003 Corporate Wireless Survey, findings showed that 60% of major U.S. corporations would prefer to wait for a WWAN technology that meets their requirements rather than deploy an inadequate solution.

Mass-market and enterprise customers are migrating from remote-access dial-up and wired broadband to wireless broadband data. As a result, providers must make wired data succeed as wireless data. They need to make it simple and easy to use at a price point that doesn't stifle the update of such services. It's time to consider a fundamental shift in air-interface technology.

As mentioned previously, cellular and PCS communications systems have historically been designed for voice traffic. The patterns associated with voice communications are well known, as they've been observed since the invention and widespread use of the telephone. Voice can be characterized as relatively predictable. Each party talks roughly half the time in an interactive manner. The statistics of call duration and time of day are well understood. Traffic engineers can therefore use a standard methodology to estimate the amount of capacity that's needed in a communications system.

The wireline telephone network has been engineered in a hierarchical fashion. Using large circuit switches, it efficiently connects one voice user to another. The physical circuit over which a call is made is held open for the entire duration of a call (hence the term circuit switching). Voice has similar characteristics in wireline and mobile settings. Existing cellular-telephone systems have therefore been designed in a similar way. They've been optimized to efficiently provide voice service.

Data traffic differs from voice in at least three important ways. First, data traffic is much more unpredictable than voice traffic. Data is characterized as "bursty." In other words, the time when the traffic arrives varies significantly. The same is true for the rate at which it arrives and the number of bits in the messages.

The second way that data traffic differs from voice is in terms of reliability. Voice is very robust. With a high bit error rate approaching 1%, voice is capable of being understood even in a noisy environment. In contrast, data applications require extremely reliable delivery. There's virtually no tolerance for bit errors. When bit errors do occur on wireless links, fast and efficient recovery schemes must be implemented to get the correct data bits to the application. A combination of forward error correction (FEC) and fast acknowledgements (ARQ) satisfies this need. The powerful FEC techniques are employed to dramatically reduce the bit error rate (BER). ARQ is used to guarantee reliable delivery.

Finally, data traffic encompasses a much different and wider range of services than voice. Different types of services have different requirements along several dimensions. A data service can be characterized by its importance or priority. This aspect is determined by the quality of service (QoS) that's required. QoS can be measured by the amount of delay that a user is willing to tolerate and the reliability that's required. A service provider may offer differing service rates or classes of service accordingly. Premium-service users may be given priority over best-effort users, whose traffic is sent if there's capacity available at the time.

For a data service, the data rate that's required and granted to a user is another dimension. A user may have a service-level agreement (SLA), which guarantees a certain minimum rate. It also allows a maximum or average rate over some period of time. A final aspect of a data service is latency or response time. This aspect determines the degree of interactivity that can be achieved, which is a measure of how quickly channel resources can be assigned at the user's request.

In addition, wireless communications have traditionally posed a difficult performance challenge for TCP/IP protocols. TCP was designed and optimized around reliable wireline links. There, bit and packet error rates were substantially lower than what is typically achievable in wireless networks. When TCP encounters dropped or lost packets, it assumes that there's congestion on the link. It doesn't consider the link itself as unreliable. Congestion is handled by reducing the information rate at which the sender is allowed to transmit. By interpreting the unavoidable errors that occur in a wireless environment as congestion, the effective data rate that's seen by the end user is reduced (FIG. 1).

These issues demonstrate that data communications have a much wider range of requirements and characteristics than voice-only network systems. This variability prohibits data from being efficiently carried over hierarchical networks, whether they're wireline or wireless. To overcome these inherent differences, orthogonal frequency division multiplexing (OFDM) offers a different approach to system design. It combines modulation and multiple access schemes, which segment a communications channel into sharable sections.

OFDM segments the channel according to frequency. It divides the spectrum into a number of equally spaced tones. It then carries a portion of a user's information on each tone. A tone can be thought of as a frequency—much in the same way that each key on a piano represents a unique frequency (FIG. 2).

OFDM can be viewed as a form of frequency division multiplexing (FDM). Yet it has a special property: Each tone is orthogonal with every other tone. Typically, OFDM requires that there be frequency guard bands between the frequencies. That way, they don't interfere with each other. Their orthogonality actually prevents interference by each other. OFDM therefore allows the spectrum of each tone to overlap. This approach reduces the overall amount of spectrum that's required.