On Sept. 1, Philadelphia announced that it is considering deploying a citywide 802.11 network following a trend set by Spokane, Wash., Corpus Christi, Texas, and Cleveland. Virginia Tech completed deploying its own campuswide 802.11 network this summer. It's expected that your local coffee shop will provide 802.11 service either independently or through a partnership with a telecommunications company, as is the case with Starbucks and T-Mobile.

While business models vary — some providers offer free service, some use a subscription-based service while others follow a pay-as-you-go model — offering 802.11 service is a way of positively differentiating a business or municipality. Clearly, there is a large market demand for high-speed data, if not for mobile high-speed data, in virtually every corner of America.

We believe that high-speed wireless data access will be nearly ubiquitous.

It does not appear practical to cover the entire country or even entire cities with WLAN hotspots, however. In a wide-area network, Wi-Fi's footprint is too small to justify the deployment and maintenance costs for the most prevalent business model — free network access in exchange for patronage. Here's where WiMax — 802.16 — hopes to gain a foothold. Providing greater coverage areas and higher bandwidths, WiMax is envisioned as a direct competitor to cable modems and digital subscriber line in providing last-mile data access. Depending on the flavor of WiMax, 802.16 will provide fixed broadband access (802.16 a), indoor fixed access (802.16d) or mobile-data access (802.16e).

While all of this may sound like another wireless standard that was never as widely deployed as promised (LMDS), WiMax differs in one key aspect: It is championed by a company with enough existing market share to make WiMax a reality. Intel has announced that WiMax will be incorporated into its laptop Centrino processors by 2006. Thus WiMax will have an installed customer base from its inception.

However, Wi-Fi and WiMax are not the only standards betting on providing wireless data access. For mobile data, Verizon has rolled out a 1xEVDO system and 1xEVDV is on the way. Another competitor in the mobile-data market could be 802.20. And then there's 802.11n, which will enhance the data rates of 802.11-like networks.

Outside of the data-centric standards, the wireless voice market includes WCDMA (UMTS), CDMA-2000, GPRS, GSM and still includes AMPS. And in this list we're ignoring the various satellite, telematic (Bluetooth, UWB, RFID), point-to-point, public-safety and regional standards. It seems that every week a new wireless standard is proposed and every week another legacy standard doesn't quite disappear.

Why the proliferation of standards? Simply put, different wireless protocols are better-suited for different applications and different environments, and no single protocol can optimally provide all of the desired services in all the desired environments. There is no reason to believe that this trend will end in the future as new applications will continue to emerge.

Thus rather than converging, we believe that the number of wireless standards will continue to grow.

Traditionally, the introduction of a new standard was accompanied by a fresh spectrum allocation. Much of the best spectrum is already taken, however. The higher-frequency bands (>3 GHz) are generally available, but even when allocated, there's not been much enthusiasm for using this spectrum. To a large part, this lack of enthusiasm is a result of the higher equipment costs and greater signal attenuation inherent in high-frequency band operation. Even within successful protocols, a higher-frequency band can significantly stunt deployment. The technically inferior 2.4-GHz ISM band based on the 802.11b protocol has enjoyed much wider deployment than the technically superior 5-GHz band-based 802.11a standard.

Noting the success of the unlicensed 2.4-GHz ISM band and the lack of enthusiasm for deploying at higher frequencies, the FCC is exploring ways of facilitating similar unlicensed operation in lower-frequency bands that are currently assigned to legacy licensed systems, particularly the public-safety bands and the TV UHF band. To make this scheme work, however, these new unlicensed systems will need to avoid interfering with the legacy systems. Frequently, this will require the unlicensed systems to adapt their waveforms around the legacy systems, perhaps in real-time and perhaps in response to other unlicensed systems. The lone exception to this is ultrawideband (UWB), which would operate at such a low power level that it would be lost in the noise floor of the legacy systems. In light of the growth of unlicensed operation, some mechanism, perhaps a radio etiquette, needs to be developed to coordinate these adaptations and share spectrum information between different networks.

And it isn't just unlicensed systems that are expected to adapt their waveforms. In an attempt to squeeze out every possible bit per second/Hz, many newer standards provide mechanisms for adapting their waveforms to changing channel and interference conditions. 1xEVDO includes mechanisms for performing dynamic scheduling and for adapting modulation and coding in response to changing channel conditions. Similarly, HSDPA and GPRS provide mechanisms for performing code rate and modulation adaptation. It isn't too much of a leap to conclude that this adaptation could begin to be adjusted based on the "experience" of the radio system, initiating the start of cognition for radios.

We believe that waveforms will become increasingly adaptive in an attempt to improve spectrum efficiency.

To provide the expected ubiquitous data access in a market with numerous heterogeneous waveforms, service providers are bundling together several protocols. For instance, T-Mobile and Sprint are offering combined mobile- and fixed-data service by bundling their traditional multi-mode cellular service with 802.11 access. As the number of waveforms included on a single device increases, the efficiency of implementing an all-hardware radio where the various standards are "Velcroed" together decreases. The addition of an extra RF chain is never cheap, and chip yield rates decline with complexity. When this problem is viewed in light of the anticipated rise of adaptive spectrum utilization, it is seen that a high degree of radio flexibility will be required.

This indicates that there is an opportunity for software radios — radios whose physical-layer behavior is largely defined in software. By implementing the operation of the radio in software, adapting a waveform can be as simple as moving a pointer; adding a new waveform can be as easy as downloading new code to memory.

There are a couple of technical challenges that are limiting the development of software radio, however. The first and more serious challenge is the development of reconfigurable analog components, perhaps using microelectromechanical systems. Without reconfigurable analog components, the radio is restricted to a predetermined set of frequency bands, limiting the options for spectrum adaptation and waveform addition.

The second challenge limiting software radio is the creation of a standardized software architecture. While a software radio can be built without a standardized architecture, the lack of a standard stunts software radio's growth. To draw a parallel, the PC market didn't take off until the PC/ X86/DOS-Windows architecture became a de facto standard.

Unlike the proliferation of wireless standards, there is a possibility of convergence in software radio platforms. Currently, the Software Defined Radio Forum, the Joint Tactical Radio System (JTRS) program and the Object Management Group are actively working toward standardizing services and functionalities for their software communications architecture (SCA). Software radio architecture standardization remains an ongoing process, however, and commercial entities have been disinclined to adopt a framework that seems overly complex and inefficient. The introduction of SCA 3.0 Hardware Supplement may overcome many of these limitations.

In spite of these technical challenges, software radio has experienced a tremendous growth in acceptance over the past few years. As part of the JTRS program, General Dynamics Corp.'s DMR radio is being fielded by the Navy, and Thales has demonstrated functioning handheld military software radios. Commercially, both Vanu Inc. and AirNet Communications Corp. have successfully fielded Global System for Mobile Communications (GSM) basestations on software radio platforms.

Even though software radio is still in the process of development, the concept is already being extended to cognitive radio. A cognitive radio is a software radio that is aware of its environment and its capabilities, is able to independently alter its physical-layer behavior, and is capable of following complex adaptation strategies. While some rather interesting claims have been made about what cognitive radio will be able to do in the future using a well-developed artificial intelligence, the more immediate application of cognitive radio is autonomous adaptation of its spectrum usage.

The cognitive radio concept has already attracted a bit of attention. Most notably, the Federal Communications Commission has been exploring ways in which it can accommodate cognitive radio technology, holding a workshop on the subject in May 2003 and issuing a Notice of Proposed Rule Making in December 2003. Similarly, IEEE USA has issued a position paper that states that it views cognitive radio as a promising approach for the implementation of adaptive-spectrum utilization.

We believe that software radio in the short term and cognitive radio in the long term will be key enabling technologies to data access and spectrum adaptation.

In the future, wireless data access will be ubiquitous and spectrum utilization will be flexible and networks heterogeneous. Subscribers will be able to transparently connect to the Internet from home, from work, from their favorite restaurant, from their car, and from their phones, and maintain their connections as they move from place to place. Spectrum regulation will be less rigid and waveforms will adapt to the needs of the application and to the current spectrum situation.

To realize this scenario, the following technologies will be critical:

• Wireless local data loop technology through a combination of 802.11 and 802.16 networks;

• Lightly regulated software radio networks that permit spectrum sharing and adaptation;

• A radio etiquette that facilitates sharing spectrum information between dissimilar radio networks;

• Reconfigurable analog components that enhance the frequency agility of software radios; and

• A standardized software platform that supports waveform reconfiguration and waveform porting.

Some of these technologies, particularly the proliferation of wireless local data loop technology, will become a reality in the near future. However, the other technologies will remain topics of research.

Jeffrey H. Reed is a professor at Virginia Tech's Bradley Department of Electrical and Computer Engineering, Mobile and Portable Radio Research Group (Blacksburg, Va.), and James Neel is a research assistant for the same group.