Despite all the talk about broadband, only about7 percent, or 7.5 million, of U.S. households subscribed to high-speed Internet access services as of last June, according to a February report issued by the U.S. Federal Communications Commission (FCC). And the situation isn't much different elsewhere. Only a few countries with much smaller populations report somewhat higher percentages.

Lack of need is, of course, one reason for the unimpressive numbers. Outside of playing interactive games, which is hardly a universal activity, no broadband "killer app" has yet emerged. Another reason is difficulty in getting service. For a variety of reasons, many would-be subscribers have been unable to get cable or digital subscriber line (DSL) service. For them, a fairly new type of technology known as non-line-of-sight (NLOS) wireless may be just what they need.

At first, NLOS wireless may not sound like a big deal. After all, ordinary radios and cellphones are non-line-of-sight devices. But they don't carry broadband data [see , "What is Broadband?" What makes the latest generation of NLOS wireless technology worth talking about and having is that it delivers data at high rates over substantial distances. Moreover, most implementations do so without the need for a visit from an installation technician.

This last point is crucial. Previous attempts to provide wireless Internet access to the home failed in large part because their installation costs were too high. For example, multichannel, multipoint distribution service (MMDS), which operates at 2.5 GHz, delivers high data rates over tens of kilometers. But first-generation systems required a directional antenna to be installed on the subscriber's premises within sight of a base station antenna.

Since service providers often find out that a potential location has no unimpeded view of the base station only when they send an installer to the site, the average successful installation takes more than one visit. With "truck rolls," as technician visits are known, costing anywhere from US $300 to $800, it could take more than a year's worth of subscriber revenue just to break even on the installation.

Worse, line-of-sight (LOS) systems do not scale well. When a base station's capacity reaches saturation, the usual procedure is to divide the coverage area into two pieces, use the existing antenna to serve half of it, and put up a new antenna to serve the other half. With directional antennas, that means that the antennas at roughly half the existing subscriber locations will have to be re-aimed at the new base station site, necessitating truck rolls that bring in no new revenue.

That's why communications companies—from household names like AT and T, Sprint, and WorldCom to newcomers like Winstar and Teligent, all of which rolled out broadband LOS wireless Internet access services, principally in the MMDS band—have pretty much given up on LOS wireless, at least insofar as price-sensitive residential customers are concerned. Last year, AT and T stopped service to 47 000 customers, Sprint and WorldCom slowed or halted their development, and both Teligent and Winstar filed for bankruptcy protection.

Fixing" wireless

Enter NLOS. A number of technologists and investors believe that this relative newcomer can overcome the problems faced by existing line-of-sight wireless services. Briefly, the challenge they have to meet is to establish communication links with signal-to-noise ratios high enough to support broadband communications with easily installed, preferably indoor, antennas.

That goal may be achieved in several ways. Local-area networks, like those based on the popular IEEE 802.11b standard, do it by limiting the distance between transmitter and receiver. Cellphones operate over longer distances, but offer no broadband connectivity. LOS systems rely on a high-power transmitter at the base station, an unimpeded line of sight between transmitter and customer, and a highly directional outdoor antenna at the customer premises, all of which add up to a technology too expensive for the residential market.

NLOS attacks the problem with smart antennas, advanced modulation techniques, and, in some cases, a mesh architecture in which nodes—the individual routers on the customer's premises—are connected by multiple links [see ,

 
The mesh architecture helps keep signal strength up by replacing single, long radio links with multiple short ones.

Dave Beyer of Nokia's Wireless Routing Group smiles from behind an array of decorated wireless routers, which when mounted on subscribers' buildings will configure themselves into a mesh network.

Whereas LOS base stations use omnidirectional or sectorized antennas that spew energy over large areas, non-mesh NLOS systems (those built around a central tower) fit their towers with small antenna arrays that direct the energy where it is needed. The advanced modulation techniques like orthogonal frequency-division multiplexing (OFDM) use the available radio spectrum with great efficiency, maximizing the number of bits per second they transmit per hertz of spectrum bandwidth. OFDM does that by sending data over multiple carriers within a frequency band.

Players in the NLOS field include equipment manufacturers like Nokia Corp. (Espoo, Finland) and Navini Networks Inc. (Richardson, Texas); companies like Iospan Wireless Inc. (San Jose, Calif.), which provide transmitter and receiver designs and chips; and Internet service providers (ISPs) like Vista Broadband Networks Inc. (Petaluma, Calif.) and T-Speed (Dallas), which sell wireless access service to customers.

How NLOS systems work

The most important technology in a point-to-multipoint (non-mesh) NLOS system is its smart base-station antenna. Instead of a single omnidirectional or sectorized antenna, these systems use an array of radiating elements. Each element is fed a version of the signal to be transmitted that differs from the others only in its amplitude and phase (time delay). The signals radiated by the array elements combine with each other in space to form one or more beams of carefully calibrated strength propagating in specific directions. The directions are so chosen that the beams—after bouncing off assorted objects in the environment, like mountains, buildings, motor vehicles, and even aircraft—all reach the location of the intended subscriber at the same time and in phase with one another [see ,

Illustrations: Steve Stankiewicz

 
When the beams combine constructively, the result is a strong signal at the receiver, which can therefore use an indoor antenna.

Sounds good, but how do they do it? The answer is by first monitoring signals received from the subscriber unit to determine the characteristics of the environment and then by generating a complementary signal. For example, if the subscriber unit has a simple omnidirectional whip antenna, the signal it transmits will, in general, undergo multipath distortion—that is, it will take multiple paths to the base station, bouncing off various objects, being attenuated to various degrees, and undergoing various delays, depending on the different path lengths.

Say the base station receives two signals, one from the north and, 2 us later, one from the east that is 8 dB weaker. Then the base station transmitter will format its signal into two beams, first a strong one to the east, and 2 us later, one 8 dB weaker to the north. Of course, since the environment is constantly changing, the base station must keep monitoring subscriber transmissions, analyzing them, and updating its picture of the environment. Small wonder, then, that this sort of technology could not even be considered for commercial applications until cheap and powerful digital signal processors became available.

Taking advantage of those processors, Navini Networks uses the technology in its Ripwave product line, versions of which operate in both the licensed and unlicensed bands in the vicinity of 2.5 GHz. According to Sai Subramanian, director of marketing and product line management, the base station antenna has eight elements, but is not very large because all eight elements are within a wavelength of each other, and at 2.5 GHz, a wavelength is just 120 mm.

Although Navini Networks' subscriber premises equipment has two antennas, they don't work together as a phased array. Rather, they provide spatial diversity: it is less likely that two antennas will simultaneously find themselves in a dead spot than it is for one antenna. Navini has several U.S. trials under way.

In another approach, Iospan Wireless uses two transmit antennas at the base station and three receivers at both ends of its links. Iospan's multiple antenna technology, which was developed by the company's founder, Arogyaswami Paulraj, professor and head of the Smart Antenna Research Group at Stanford University, is known as MIMO, for multiple-input, multiple-output.

Referring to multipath distortion, Asif Naseem, vice president of business operations, marketing, and business development for Iospan, says, "Multipath is our friend. In the best conditions, we get six separate data streams out of the frequency chunk [there are six paths between three receivers and two transmitters] and realize multiples on the user data rate. In our tests from our base station on the roof to a customer over 1.5 km away, we are measuring over 13 Mb/s downstream and 6 Mb/s up. At 6 km we get over 6 Mb/s down and 4 Mb/s up. This is usable capacity." Iospan's multiple antenna enhancements of the OFDM modulation technique are being standardized in the IEEE 802.16 Working Group on Broadband Wireless Access Standards. The company is aiming the price of customer premises equipment at less than $500, and has begun trial deployments with partners in the United States and internationally. (Iospan will rely on others to manufacture, market, and install its systems; it simply provides ASICs, software, and reference designs.)

The mesh we're in

Smart antennas aren't the only way to sustain signal strength over long distances. Another way is to break those long distances into a series of shorter hops, with the signal boosted every time it is relayed from one node to another. That's the key idea behind the mesh approach. Each customer's equipment acts both as a means of connecting that customer to the network and as a node through which communications traffic from other customers in the vicinity is boosted and rebroadcast on its way to and from the facilities of the service provider.

So far, at least three California companies are promoting their use of the mesh network approach: Nokia's Wireless Routing Group (Mountain View), SkyPilot Network (Belmont), and CoWave Networks (Fremont). SkyPilot, still in quiet start-up mode, is making equipment and providing service and seems to be focused on the residential and small business market using 802.11 standard technologies. In its plans, CoWave talks primarily about providing equipment that uses licensed bands for these markets.

Farther along is Nokia, which has been shipping its R240 Wireless Router for about a year now, with over 60 wireless ISPs deploying networks around the country. Nokia also operates a test network near its offices in Mountain View to test new network software releases. Its technology has been deployed by Vista Broadband Networks (Petaluma, Calif.) to more than a hundred customers just beyond the wine country north of San Francisco after a test of more than two years. Vista offers 384-kb/s symmetric service for $45 per month and 1-Mb/s symmetric service for $55 per month. Installation today costs $200.

In the Mountain View network, about 80 customers within three or four blocks are participating, said Dave Beyer, head of Nokia Wireless Routing. Indeed, walking with Beyer around the test neighborhood, one of the authors (Schrick) could easily see many units mounted on buildings and houses [see ,

 
These are the wireless routers that make up most of the customer premises equipment needed for the system. These routers accept not only the signals destined for their locations but others as well, which they boost and rebroadcast.

Router density is crucial, because, according to Donald Cox, a professor who leads Stanford University's Wireless Communications Research Group, maintaining an adequate signal-to-noise ratio would require a service provider to install a transceiver base station every 50 meters, a proposition that would appear to be prohibitively expensive. After all, the main reason that Ricochet, an early wireless Internet access provider, for example, went out of business was the high cost of deploying its network. But unlike Ricochet, Nokia's mesh approach makes the customer equipment part of the network and eliminates the need to rent space for each and every node. Nokia's Beyer estimates that the price of the wireless router and associated equipment for the home or office is about $800 in small quantities.

In addition to individual subscriber's wireless routers, Nokia's system uses base stations called AirHeads, each of which serves an area called an AirHood. The subscriber units and AirHeads are roof-mounted, but have no need for careful antenna siting and aiming.

Still, some criticism of the mesh architecture is common, Beyer notes, because it can increase the number of hops that a packet must take in both directions. But clearly, keeping hop lengths short keeps signal strength up, which allows higher data rates. This, Beyer claims, makes the AirHood and other mesh architectures much faster overall than distribution from a central point.

One concern with mesh networks is that the extra hops add latency, which can be a problem when carrying real-time interactive traffic like voice. Although each hop does indeed add some latency, Beyer says that a quality-of-service software upgrade to be released later this yearwill make high-quality voice communications possible for subscribers within three hops of the nearest AirHead.

Right now Nokia is focusing its wireless routers and system on North America in the industrial, scientific, and medical (ISM) public band. Europe and Asia are also potential markets, but Beyer says that even though the 2.4-GHz band is also public in those markets, the limits on radiated power are much stricter there, at 100 mW of effective radiated power compared with 4 W in North America. This makes a huge difference in range and coverage. European and Asian authorities have been petitioned to increase these power limits, and if they do, Nokia will reexamine those markets.

Facing interference

The new wireless technologies face substantial technical and commercial challenges [see , "Grass-Roots Wireless Networks" Some of these broadband systems use regions of the spectrum that have, in the past two decades, been opened for public and private use. Equipment working in the public regions must coexist not only with similar equipment but also with completely different gear—microwave ovens, for example. The licensed bands, of course, do not face this uncertainty.

The well-publicized Bluetooth standard, now enshrined in part as IEEE Standard 802.15, lives in the ISM band, which is also where most 802.11 equipment works. If Bluetooth becomes prevalent and at the same time the use of 2.4-GHz cordless phones continues to grow, that band may become too crowded to support reliable service. Another consideration is that in the public spectrum, power must be kept low because of interference concerns. In the licensed parts, safety determines the permissible power levels, which results in higher limits.

Will they come?

What if you built a network and nobody came? The February 2002 FCC report also cited a survey from the Strategis Group (Washington, D.C.) that found that only 12 percent of on-line customers were willing to pay $40 per month for high-speed access, a number that rose to only 30 percent when the price was dropped to $25 per month.

Another survey, from Hart Research and The Winston Group (Washington, D.C.), quoted in a November 2001 Communications Daily article, found a similar mood, indicating that 36 percent of then-current dial-up users were interested in faster access, but not at current prices, which are well over $40 per month and rising. In other words, these surveys found no additional demand for high-speed at today's prices.

Many experts believe that compelling new services will be needed before demand grows enough to drive broadband access to most households [see "Third-Generation Wireless Must Wait For Services to Catch Up," pp. 14-16 Maha Achour, president and chief technical officer of Ulmtech (San Diego, Calif.) and a broadband wireless and fiber-optic communications expert, believes that what is needed to increase demand for broadband services is not games, as many claim, but learning, collaboration, and conferencing. "I think that distance learning and collaboration will be the killer applications—I strongly believe in on-line learning," she emphasizes.

Nevertheless, the numbers seem to indicate that we have a way to go before serious demand for high-speed Internet access develops among wireless users. Several more corporate flameouts may occur before we get the speedy, reliable, and safe services that new wireless network technologies promise.

Tekla S. Perry, Editor

By Clay McLachlan/Getty Images