Powerful digital signal processors and CMOS RF chips are the key enablers
The Rebirth of Radio
By Michael J. Riezenman,Senior Engineering Editor

In the beginning, automobiles were known as horseless carriages, reflecting the feeling that horses were the natural way to propel vehicles. Similarly, radio was known as wireless. In the transportation arena, engines and motors have long since supplanted the horse, and the term horseless carriage has gone the way of the horse-drawn vehicle. But wiring of one kind or another is still regarded as the "natural" way to connect nodes in a communications network. To most of us, wireless is an inferior, niche technology appropriate only for mobile applications. Even television, which was originally popularized as a wireless service, is now broadcast largely over cable.

Today, that mindset is under attack. Irwin Mark Jacobs, the chairman and chief executive officer of Qualcomm Inc., San Diego, Calif., admittedly not a disinterested party, predicted recently that by 2005, most people will be using wireless phones for both voice communications and Internet access.

He may be optimistic, but he is right about the direction in which things are moving. IEEE 802.11, the wireless local-area network standard, is one of the hottest product areas at the moment; Bluetooth threatens to become even hotter; Nokia recently unveiled its self-configuring rooftop RF system for Internet access; fixed-wireless links are gaining increasing attention as candidate solutions to the Internet's last-mile problem; and cell phones need hardly be mentioned.

Why the sudden interest in radio? It can't compare with wired optical networks in capacity. It leaves users more vulnerable to eavesdropping than wired networks do. There is the problem of all those unguided waves interfering with one another. And cell phones with all their dropped calls and coverage gaps don't seem to be exactly a ringing endorsement of the technology.

The reason isn't hard to find: advances in microelectronics have made it possible to build complex wireless systems at low enough cost to make them economically viable. These advances, and others on the way, are not only allowing us to realize the well-known advantages of wireless communication [see "Why Wireless?"], they may also turn out to be the best way to access the Internet in many situations.

Besides, some of the cited drawbacks of wireless may not be as serious as they sound or they may be susceptible to treatment by further advances in semiconductor technology. Optical fiber indeed has much greater capacity than radio. But so what? Not every application needs immense transmission speed. Fiber makes sense at the core of a telecommunications network. Wireless is being touted for the periphery--or for small, self-contained networks--where blazing speed is less important than such factors as mobility, cost, and provisioning speed.

Today's cell phones certainly leave a lot to be desired. But is that because of any fundamental problem with the technology or is it because cellular service providers are signing up customers faster than they are building networks to serve them? If the latter, as appears to be the case, then time may well fix the problem.

As for eavesdropping and interference, advanced digital signal processors (DSPs) are already dealing with those problems and promise to do even more. DSPs, after all, are at the heart of every digital cell phone, compressing and encoding voice streams, and making them almost immune to eavesdropping in the process. They are only beginning to be used in smart antennas, which will mitigate the interference problem and boost system capacity [see The Myth, the Law, and the Spectrum].

Chips to the rescue
Take IEEE 802.11a, the high-speed wireless local-area network standard. This 54-Mb/s improvement to 802.11b (11 Mb/s) uses an exotic orthogonal frequency-division multiplexing scheme and works in the new Unlicensed National Information Infrastructure (UNII) band, where 300 MHz of bandwidth is available in two pieces: 200 MHz from 5.1 GHz to 5.3 GHz, and 100 MHz at about 5.7 GHz. With its 54-Mb/s data rate, 802.11a is comfortably above the 22 Mb/s needed for a home network capable of simultaneously connecting two computers to the Internet or handling multiple audio and video streams such as might be generated by CDs, camcorders, and so on. So why hasn't this attractive technology been adopted until now?

Adoption of 802.11a has been delayed by the high cost of implementing it--specifically, by the need to fabricate the radio-frequency (RF) portion of the circuitry in a compound semiconductor material like GaAs or SiGe. That is not only expensive, it also is incompatible with CMOS--the material of choice for the modem portion--with which the RF circuitry will one day be integrated. But no more.

At Networld+Interop in Atlanta last September, two unrelated companies announced all-CMOS solutions to the 802.11a problem. Radiata, an Australian company with roots in Sydney's Macquarie University, and recently acquired by Cisco, unveiled a pair of chips that between them almost constitute the physical layer of an 802.11a transceiver. The R-M11a modem chip, which includes 20-MHz analog-to-digital and digital-to-analog converters, is built with a 0.25-µm CMOS process. Its companion R-RF5 chip, fabricated with a 0.18-µm process, is a complete 5-GHz radio, including all necessary amplifiers, mixers, and filters. All that's needed to complete the transceiver is a media access controller (MAC) and possibly a power amplifier, if the R-RF5's 0-dBm transmitter output is insufficient. The chip set will sell for US $35 in large quantities.

Atheros Communications, Sunnyvale, Calif., introduced an even more highly integrated chip set at N+I. Its two chips include both the MAC and a power amplifier. The company, which also has a tight academic connection (with Stanford University), priced its chip set at "below $35." For both companies, the quoted price for the chips translates into an end-user price for the complete transceiver of about $150.

By contrast with 802.11a, Bluetooth--the self-configuring short-range network pioneered by Sweden's Ericsson--aims a lot lower down the performance scale--around a megabit per second--but also has a much more stringent budget. Companies contemplating the addition of Bluetooth to their products tend to regard $5 as the acceptable cost of adding that capability. What that $5 buys is connectivity--the capability to participate in an ad hoc network anytime two or more Bluetooth-enabled devices get within about 30 meters of each other.

Most Bluetooth visions involve portable equipment with limited display capability, so its fairly low data rate is consistent with its probable missions--e-mailing, wireless keyboards, wireless headsets, smart home appliances, and the like. Wireless headsets for cellular phones may be the killer app here; their very low power is likely to ease the anxieties of cell phone users worried about brain damage from excessive exposure to RF power.

Like several other RF technologies--including IEEE 802.11b--Bluetooth operates in the 2.4-GHz ISM band using frequency-hopping spread-spectrum. If it gains anywhere near the popularity anticipated for it, problems may arise with other services operating in that band, including wireless LANs based on 802.11b. Therefore, in a sort of perverse way, Bluetooth--or rather its avoidance--may turn out to be a factor contributing to the success of 802.11a.

Yet another self-configuring network operating in the ISM band--at least for now--is the kind based on Nokia's RoofTop Wireless Router. All that's required to connect to a network based on this technology is to mount one of the wireless routers on a convenient rooftop; supply it with power; connect it to a computer's Ethernet port; and, of course, make financial arrangements with an Internet service provider with RoofTop equipment. It's not even necessary that the customer's unit have a direct line-of-sight link to the service provider. Just as long as it can see one other node in the network, all will be well. The unit will "talk" to similar units in the neighborhood and together they will configure themselves into an ad hoc mesh network. RoofTop routers self-configure and self-heal as new customers connect to the local system. New subscribers add redundancy to the network, and thereby strengthen the infrastructure, according to Nokia general manager Ari Leppä.

In describing the system, Leppä was at pains to explain that although the wireless routers run in the unlicensed 2.4-GHz band, there is no reason they cannot be moved to one or more other parts of the spectrum. The radio portion of the gear is separate from the router portion and can be independently re-engineered.

Nokia's system is illustrative of a new use of radio--wireless Internet access unrelated to mobility. The RoofTop Wireless Router makes possible data rates of up to 12 Mb/s (six 2-Mb/s channels), which compares favorably with cable TV and DSL access methods. Of course, since each user shares a cell with several others, the data rate achievable at any given time will vary with the number of users actively communicating in the cell.

More deterministic fixed-wireless systems are based on Local Multipoint Distribution Service (LMDS) technology, which uses directional antennas but has the corresponding drawback of requiring expert installation. These systems work in the 30-GHz region of the spectrum and provide data rates on the order of 10 Mb/s. As always, the name of the engineering game is tradeoff. Easy installation or high speed--you pays your money and you takes your choice.