October 2002

Clear Connection

"Multiuser detection" could greatly expand cell towers' capacity-increasing their range, their data transmission rates or the number of users they can handle.

By Larry Hardesty

Anyone who's suffered broken cell-phone connections in urban areas already choked with voice traffic might doubt the imminence of the wireless Web. But researchers predict that a new technology called "multiuser detection" could greatly expand cell towers' capacity-increasing their range, their data transmission rates or the number of users they can handle.

Multiuser detection disentangles cell phone signals that interfere with each other, and "everybody who has an interest in cell phone systems is working on the implementation," says Peter Okrah, a researcher with wireless-equipment supplier Motorola. Ericsson, Nokia, Lucent Technologies, Intel and Microsoft have all investigated the technology, though the first to market may be smaller competitors such as industry supplier Ascom of Berne, Switzerland, or Mercury Computer Systems of Chelmsford, MA.

The central problem of cellular telephony is how to let users access the same tower without breaking up each other's calls. The most efficient solution is to assign a unique "code," or combination of radio frequencies, to each phone; a tower can key onto any one code and screen the others. In practice, however, buildings and other obstacles knock signals out of sync and cause them to interfere with each other.

The solution: subtraction. Since a tower knows the codes for all its current calls, it knows what the interference looks like and could remove it from a distorted signal. But that requires a lot of processing power-or some mathematical shortcuts.

Proposed remedies use both. Removing interfering phone signals one at a time would cause delays, but Mercury's computing platform, which deploys many microprocessors in parallel, can remove multiple signals at once. In simulations, it has increased a cell tower's capacity by 50 percent. Ascom's algorithm makes a series of guesses about the contents of a signal, using fractions-rather than ones or zeroes-to represent ambiguous data. Promising guesses are refined further, and only after several passes does the algorithm settle on a single interpretation.

Ascom is a year or two from a hardware implementation of its algorithm, but Mercury hopes to test its system with a telecommunications firm this year. Some insiders consider that hope optimistic: "Will there be applications for multiuser detection? Sure," says Paul Whalen, chief executive officer of Colchester, VT-based telecommunications training company Hill Associates. But he adds that right now, the technology "is in the domain of the researcher, and the apparent complexity of the solution may scare a lot of people." Nonetheless, Mercury vice president Barry Isenstein says the technology is ready to go: "At this point," he says, "we are not the limiting factor." Soon, the same could be true of cell towers.