Imagine Wi-Fi on steroids. Imagine a wireless data technology that covers ranges up to 30 miles instead of a couple of hundred feet. Imagine data rates many times higher than for Wi-Fi's 802.11b technology—even higher than for the newer 802.11a and 802.11g. Throw in security features to protect your sensitive data. Add low latency and guaranteed quality of service—lacking in Wi-Fi—for uninterrupted transmission of time-critical voice and video data. Allow access by hundreds or even thousands of wireless users—unlike Wi-Fi, which can connect only tens of users.

What you've just imagined is an implementation of IEEE standard 802.16a, which describes a technology that's increasingly referred to as WiMAX (World Interoperability for Microwave Access). 802.16a, or WiMAX, is not a fantasy, but a coming reality. Ratified just a year ago, the 802.16a standard will soon make possible "last mile" broadband wireless access in areas that are too remote or too difficult or too expensive to reach with wire or fiber. It will also serve as Wi-Fi backhaul, enabling quick and easy connection of Wi-Fi hot spots to the Internet when there's no convenient access to a wireline (Figure 1). Eventually, it will enable notebook computers and PDAs to connect directly to metropolitan-area networks (MANs) that provide geographically continuous wireless coverage.

Figure 1:  802.16 and 802.16a used for broadband wireless access and 802.11 backhaul

And 802.16a has strong commercial backing to go along with its technical capabilities. The WiMAX Forum, a nonprofit group that promotes 802.16a technology, has as its goal the certification of interoperable 802.16a products, regardless of vendor. In that regard, they're taking a page from the playbook of the Wi-Fi Alliance, which helped popularize and commercialize 802.11 technology. Founded last year by chipmaker Intel, wireless service provider Nokia, and several wireless equipment makers, the WiMAX Forum now includes almost 70 member companies. Several of them expect to deliver WiMAX certified products later this year.

The big commercial push began in January 2003, when 802.16a became a much altered extension of 802.16 (Table 1). 802.16a adds non-line-of-sight operation to 802.16, which was intended basically as a high-speed tower-to-tower communications link. Non-line-of-sight, in turn, enables access by many users concurrently. In addition, 802.16a offers numerous tradeoffs in parameters like range, data rate, and number of concurrent users. That gives implementers of the technology enormous flexibility in setting up systems for their particular customers, operating conditions, and business models.

802.16a also provides useful tradeoffs between channel width and number of users. Unlike the fixed-width, 20 MHz channels of Wi-Fi, WiMAX channels can vary in width from 1.5 to 20 MHz. Wireless operators with few subscribers in an area can start with a narrow channel and then add channels or use a wider channel as they acquire additional customers. Similarly, they can use a narrow channel to provide broadband wireless access to a few rural customers and a wide channel to provide the equivalent of a T1 connection to multiple business customers simultaneously.

Flexible channel widths offer other advantages, too, among them the ability to meet requirements of regulatory agencies in different countries. For example, a wireless operator in Europe with 14 MHz of bandwidth in the 3.5 GHz band might want equipment that provides 7 or 3.5 MHz channel bandwidths, which 802.16a allows. Also, in licensed bands (802.16a has spectrum in both licensed and unlicensed bands), flexible channel widths prevent waste of purchased bandwidth. An operator that has paid for 14 MHz of bandwidth will not want a system that requires channel widths of, say, 6 MHz, which would waste 2 MHz of spectrum.

A big plus for 802.16a is that it offers guaranteed quality of service (QoS) for delay-sensitive applications like voice and video. Its MAC (media-access control) layer, by using a grant-request access mechanism, prevents the data collisions and subsequent wasteful retransmissions that occur with 802.11's contention-based access scheme. With QoS, 802.16a enables wireless service providers to provide different levels of service to different customers according to their contracted service-level agreements. For example, it can guarantee high bandwidth to business customers or low latency for voice and video applications, while providing only best-effort and lower-cost service to residential Internet surfers.

An 802.16a feature often stressed by the WiMAX Forum is capacity—enough bandwidth, the group claims, to "simultaneously support more than 60 businesses with T1-type connectivity and hundreds of homes with DSL-type connectivity using a single sector of a base station." There's no technology magic that creates that kind of capacity; it doesn't result from a huge increase in data rate. After all, the aggregate data rate in a communications channel depends simply on the bandwidth of the channel (Hz) and the efficiency of the modulation method used (bits/second/Hz). The modulation methods of 802.16a are more spectrally efficient than some others, but not by all that much. Compared to 802.11, for example, the difference is maybe a factor of two, not tens or hundreds.

But 802.16a does allow many more concurrent users than 802.11. As noted by Aditya Agrawal, senior marketing manager at Fujitsu Microelectronics America and a WiMAX vice president, not all of a network's users are active at the same time. So, Agrawal says, if the network can manage the users well, they can all gain the access and the bandwidth promised to them by their service level agreements. The key, he says, is 802.16a's MAC layer.

802.16a also keeps more users connected by virtue of its flexible channel widths and its adaptive modulation. Because it can have channels narrower than Wi-Fi's fixed 20-MHz channels, it can target lower-data-rate subscribers without wasting bandwidth. And when subscribers encounter noisy conditions or low signal strength, adaptive modulation can keep them connected when they might otherwise be dropped.

WiMAX doesn't yet provide Wi-Fi's portability, however. Initial 802.16a customer premise units (CPEs) will be comparable in size to a satellite TV hookup and will cost several hundred dollars. The signal from the WiMAX CPE will connect either to an Ethernet LAN or directly to a computer. Notebook computers can benefit from WiMAX's presence when it's used for 802.11 backhaul, but their connections will still be through Wi-Fi's 802.11 technology. 802.16a modems will eventually be available for notebooks and PDAs, but not right away.

WiMAX does hold out an attractive promise for the future, however—mobility. Proposed IEEE standard 802.16e will add nomadic operation within metropolitan area networks (MANs) consisting of cells with a typical radius of 2 to 5 km. Such MANs are several years off, however, and planning for 802.16e itself is still in the early stages. Also, as now envisioned, 802.16e will not enable mobile users to move as rapidly as they might always want—for example, on a high-speed train. Current thinking limits mobility to "vehicular speeds" or less, with data rates while a user is mobile topping out at around 5 Mbps.

Another future possibility for WiMAX technology is the mesh network. In a mesh network, subscriber stations can communicate not just with base stations, but also with other subscriber stations. One advantage of this is the ability to operate even when a large obstacle, such as a mountain or hill, blocks direct access by a subscriber to a base station. Blocked subscribers can get to the base station indirectly by going through other subscriber units. Even a small amount of meshing can greatly improve a base station's coverage, but it depends on having enough subscribers in place. "It's a chicken-and-egg problem," says Fujitsu's Agrawal. "A mesh network is not effective unless and until there's a lot of deployment there."

But could WiMAX systems eventually replace their Wi-Fi counterparts? Possibly. Accessing WiMAX via Wi-Fi would entail an unnecessary step, and Wi-Fi, as it now exists, lacks many WiMAX capabilities, such as long range and the ability to handle voice and video applications (Table 3). Wi-Fi is addressing the low-latency requirements necessary for voice and video with proposed standard 802.11e, but current efforts are headed only toward improving latency with prioritization, not toward a guaranteed QoS. "It still doesn't solve the contention problem," says Fujitsu's Agrawal, "and that's the core of 802.11." Needless to say, though, Wi-Fi hot spots will need to remain in place unless and until there are WiMAX-based networks to replace them.

Down the road, competition could emerge in another area—mobility. The mobility extension to 802.16, to be covered in proposed standard 802.16e, is still in its very early stages, but it's nevertheless farther along than the proposed standard 802.20, which is addressing mobility for wide-area networks, such as 3G for the cellular industry. Officially, the WiMAX Forum denies any competition: "The purpose of 802.16e is to add limited mobility to the current standard which supports only fixed operation," a WiMAX white paper says. "IEEE 802.16e is not intended to compete with 3G or other truly mobile efforts." Some observers note, however, that the goals of the two proposed standards aren't dramatically different, and some do see a possible competition. As Proxim's Orr notes, "They both are approaching the solution from different motivators, and they may either coexist, or there may be a decision made by the marketplace when they're available."

In the short term, the goal of the WiMAX Forum and its members is simply to get 802.16a technology deployed. To that end, they're banking on a pinned-down standard 802.16a, commodity silicon by suppliers like Intel and Fujitsu, and WiMAX certifications of compliance and interoperability for 802.16a equipment. Those conditions, they say, will spur 802.16a activity just as efforts by the Wi-Fi Alliance spurred deployment of 802.11 technology.

And 802.16a products are on the way. Both Intel and Fujitsu expect to have silicon within the next few months, and several suppliers of wireless equipment have stated their intentions to provide WiMAX base stations and subscriber stations. So, if all goes as planned, WiMAX should soon be on its way to being as well known and as ubiquitous as Wi-Fi. Until then, the future of wireless is in the air.