Less & Moore

So far, Moore's law has proven pretty accurate in forecasting how computing power would improve, and wireless has benefited with small, feature-packed handsets and automated, end-to-end network-management solutions.

Moore's law doesn't apply to radio spectrum, however, and that's a problem because strong subscriber growth and third generation (3G) means shoehorning more users and services into a finite amount of space. One example of how to do more with less is what might be called the really IN. Currently, network optimization is expensive and time-consuming because it's done by hand by experts. One way to streamline that process and cut overhead is with a network that can optimize itself. The concept is simple: Modern wireless phones already measure signal strength as the user moves around, so why can't the network use that information to tweak itself?

"In an ideal world, the network would know where the cell phones are, know what kind of signals they're receiving, map that out in the function of time of day and location, and adjust itself to give the kind of seamless coverage you'd really like to have," said Richard Howard, Bell Labs/Lucent Technologies wireless-research lab director. "You're going to get that only when you have a network that pays attention to these details and can do that itself."

The network would pre-optimize itself after construction by using propagation and traffic simulators and continue to tweak itself after roll-out by adapting as the topography, traffic and propagation dictate. A self-optimizing network would require, for example, base stations that could tip their antennas up and down and change power as needed. That's already possible, to some degree, with actuators, but declining hardware prices and increasing software sophistication are making it rapidly feasible.

As Howard put it, "Something that was difficult last year is conceivable this year and trivial next year."

IP: ONE FOR ALL? Changing one part of the network changes the rest of the network.

"It's really like looking at an onion," said Stephen Blust, BellSouth Cellular senior manager of strategic technologies. "If you've got technology at the center, and it evolves, you've got to evolve all these other layers that go with it: architectures, management, engineering, design, operations. Each one of those is a marketplace unto itself, but all are intertwined."

Case in point: IP-based wireless networks. Nortel's purchase of Bay Networks earlier this year indicates it's no longer a question of if but when.

"Our vision is to be able to deliver IP-centric networks for wireless operators in the next 2 to 5 years," said Mehmet Unsoy, Nortel director of wireless Internet business development. "By 2003 to 2005, circuit-switched networks will be passe."

At least initially, the IP network will stop at the base station and won't be used over the airlink. Even so, a packet-based, voice-over-IP (VoIP) network is more efficient than a circuit-switched network. IP telephony also should streamline interconnection with IP-based landline and computer networks and make wireless more data-friendly. Another attraction is cutting capital costs and network-management overhead by as much as 70%, according to some estimates, because when voice, data, wireless and landline all share a common protocol, carriers should be able to use a single network-management system instead of two or three, each with its own staff.

"There are a lot of compelling reasons to the service provider to offer IP services," said Chuck Brown, Intel wireless-marketing manager. "They're really one of the driving forces behind it."

Another motivation is the ability to offer self-provisioning, which allows carriers to reduce personnel overhead because subscribers can activate services and check their account status via their handsets.

"That type of ability is immediately providing benefit to the service provider," he said.

Where IP falls short is in quality of service (QoS). It doesn't guarantee that packets make it to their destination, so realizing the dream of IP-based networks means first finding a way to ensure capacity and redundancy. One of the reasons Nortel bought Bay was for its QoS experience.

"Their switches run stock exchanges," Unsoy said. "They have mission-critical products that we believe will meet the telco and carrier requirements in the IP paradigm."

Meanwhile, standards committees are developing new versions of IP with improved service and performance guarantees.

"There is technical work going on to make sure that IP delivers carrier-grade quality of service without hampering the bandwidth-on-demand characteristics of IP infrastructure," Unsoy said. "We believe solutions will be found within the next year to deliver quality VoIP so that carriers can offer commercial VoIP by the end of next year, at the latest."

MULTIPATH: FRIEND OR FOE? Although IP will increase network capacity, the airlink remains a major hurdle to high-bandwidth data services. Ironically, the solution might lie with wireless' arch enemy: multipath. Researchers used to be concerned primarily with mitigating multipath, but now the trend is more toward exploiting it to increase capacity.

One example is Bell Labs Layered Space-Time (BLAST), which uses upward of eight antennas to receive multipath signals from a single source. BLAST starts by dividing a single data stream into multiple substreams, which then are transmitted in parallel over a single frequency band via an array of antennas. Once in the air, multipath scatters the sub-streams, which the receiver, using a similar antenna array and signal processing, pieces back together into the original stream. In early tests, BLAST achieved 50bp/s per Hz, better than the current 1bp/s or less and enough for low-bit-rate video over a single AMPS channel.

At least in its current form, BLAST appears better-suited for fixed wireless applications, but initial tests suggest a handset version is possible. Howard said they found the antennas could be closer together than originally thought without significantly decreasing the overall data rate. Beamforming at the base station also could reduce the number of antennas required at the handset.

The Virginia Tech Antenna Laboratory is testing its own multipath antenna, one designed specifically for handset applications.

"The handset can virtually synthesize an antenna pattern, one that could block multipath and one that could also take advantage of it by putting beams in the direction of the multipath and adjusting for the phases," said Jeffrey Reed, an associate professor and assistant director of the university's wireless-research center.

Smart antennas are appealing because they mitigate multipath. Handsets and base stations are relatively cheap compared to spectrum, which likely will grow even more scarce with 3G, so it's justifiable to increase their costs if they'll wring every bit of bandwidth out of the airwaves. Smart antennas also are a throwback to diversity antennas, which quickly became unnecessary as the wireless industry gained experience. But with the need to increase capacity, they might be an idea whose time has come again -- for now.

"Let's say in the next eight to 10 years you see those kinds of functions back in a handset because we need them to do wideband 3G services," said BellSouth's Blust. "Maybe 10 years further down the road you find we've sufficiently improved the way we do things and take them back out again. So it's kind of an interesting cycle."

MOORE TO COME One indicator of what technology might lie just around the corner is the military, whose battlefield communications networks are modified for commercial applications. The Defense Advanced Research Projects Agency is funding research into "multi-hop," ad-hoc networks, where each handset uses other nearby handsets as repeaters. One nettle-some hurdle is increased battery drain on the handset as it acts as a repeater, but multi-hop still might be a viable alternative to traditional cellular networks in areas where deploying infrastructure isn't practical.

"You get a critical mass of these things out there, and you have a fully connected network," Reed said. "It's a concept worth watching now. It may evolve in the next 10 years for commercial application."

Direct Wireless is taking a similar approach with its network of handsets that communicate directly with one another over secure, full-duplex connections. Repeaters would provide PSTN access via microwave, satellite or landline. Eliminating base stations, switches and other network infrastructure should make it a cost-effective solution for marginally profitable markets, such as rural North America and the Third World. Direct Wireless' first installation is slated for June 1999 in one of the undisclosed Western U.S. markets where the company is buying PCS licenses.

"We see companies that have made a (PCS) choice wanting our technology for their more sparsely settled areas where they can't afford the infrastructure of a traditional distribution system," said Frank Neukomm, Direct Wireless president. For example, a carrier might offer dual-mode phones that can switch between the Direct Wireless system and cellular networks as coverage allows. For start-ups, the appeal could be rapid deployment and a return on the license investment.

"You can set up these systems literally overnight," Neukomm said. "You just come with a box of phones, hand them out, and you've got a system on the air. "