Plenty of wireless last-mile plays target Internet access, but SOMA Networks Inc. is trying something different. The startup is marketing not to end users, but to carriers, offering a setup that can provide multiple services.

The idea is to use a wireless network to hit the pockets of population untouched by DSL, or to reach the rural zones where DSL isn't practical. It's hardly a new idea. Services such as Ricochet, reanimated by Aerie Networks Inc., are delivering wireless Internet access in a similar fashion. Satellite broadband is another alternative, one that hasn't attracted large subscriber numbers but could be bolstered as RBOCs team up with satellite TV providers. (See Ricochet Rides Again and Bells Embrace the Dishes.)

SOMA's plan is closer to the satellite-RBOC idea, in that the company wants to deliver the triple-play connectivity of voice, video, and data, and do so using fixed wireless (meaning the customer doesn't move around the way a cell-phone user does). Its target customers are the RBOCs and backbone carriers, however, with its first announced customer being Japanese backbone provider NTT Communications Corp. Those customer, in turn, would use SOMA's equipment to offer broadband, particularly to businesses.

SOMA hopes to stand out by packing the connection with intelligence, letting the customer-premises terminal make adjustments for the unpredictable quality of wireless connections. Assuming SOMA can do it right, it would provide something akin to a DSL connection, but with far fewer operational hassles. The connection would be able to mend itself if the signal deteriorated, and the customer terminals could be upgraded and maintained remotely from the carrier's site.

"If you look at the way DSL is built or a cellular network is built, you're putting more on top of an existing mesh," says Greg Caltabiano, chief operating officer. "Even if the intelligence is there, there isn't a way to change the protocol."

The five-year-old startup has moved quietly so far, having raised more than $110 million mostly from individuals such as Compaq Computer Corp. founder Ben Rosen and former U.S. Senator Bill Bradley. SOMA recently landed Bank of America Corp. as well (see SOMA Seals $10M Financing).

SOMA's air interface is based on wideband CDMA (W-CDMA), but the startup had to add proprietary extensions to accommodate its antenna technology -- meaning SOMA's customers have to buy both the base stations and the customer terminals. The payoff is that its terminals aren't susceptible to multipath interference, and they can seek out the strongest signal path by continually listening in all directions.

The terminals pack the software to automatically assign priority to certain traffic types, such as voice, and they can also adjust to changing signal conditions. If coverage becomes weak, the connection will boost its level of forward error correction to compensate. "It's constantly squeezing as much out of the channel as it can," Caltabiano says. "If it sees that it's voice coming in, it will change to a low-latency channel and handle every packet carefully."

SOMA base stations can serve more than 1,000 users at 12 Mbit/s per user. On the customer side, SOMA's box plugs into whatever network infrastructure is set up: Ethernet, 802.11, or plain old telephone service. Given the amount of variation possible on the customer side, SOMA hopes to get consumer electronics firms to build the terminals.

Prices for the equipment vary depending on configuration, but a high-volume deployment often runs between $500 and $1,000 per subscriber. What the service provider avoids, Caltabiano asserts, are the heavy operating costs associated with setting up and maintaining a DSL connection.

— Craig Matsumoto, Senior Editor, Light Reading