The new approach is detailed in a paper, "Co-Channel Interference Reduction in Dynamic-TDD Fixed Wireless Applications Using Time Slot Allocation Algorithms," published in the October issue of the IEEE Transactions on Communications. The authors are Wuncheol Jeong, doctoral candidate in electrical engineering, and Kavehrad.
Kavehrad explains that, currently, high speed Internet access capable of carrying MP3 files, video, or teleconferencing is available primarily over wired networks. However, wireless local loops are being introduced as broadband alternatives in some test markets. These new wireless networks are facing serious obstacles in competing for bandwidth; sometimes, having to share bands with cordless phones or even microwave ovens. Even when the wireless providers use licensed bands, they face the prospect of many customers simultaneously uplinking and downlinking information across the net, creating co-channel interference.
The wireless local loops work much like cell phones via a base station that sends the radio signals carrying the Internet connection out to any customer whose residence or business is equipped with an appropriate antenna. Unlike cell phone usage, however, the two-directional uplink and downlink traffic between the customer and the Internet provider's base station is more asymmetrical with very little use during sleeping hours and lots of use when kids come home from school and download music or play games, for example. Kavehrad says, "The nature of multimedia traffic is not static in uplink and downlink directions, as with voice telephony, and the bandwidth is more biased toward downlink transmissions."
Wireless local loops need both software and hardware that enables the network to respond to the changes in traffic while also making sure that every hertz in the available spectrum is used as efficiently as possible. In addition, the system must contend with the fact that some incoming interfering signals are stronger than others.
The solution developed by the Penn State engineers is software that allows the subscriber signal whose direction of arrival is subject to a lesser number of strong interferers to be processed ahead of the ones experiencing the most interference. In other words, the new strategy is a scheme that allows avoiding strong co-channel interference by sequencing the processing of the signals according to the amount of interference they are experiencing. Since the amount of interference any subscriber's signal experiences varies microsecond by microsecond, no subscriber has to wait very long for a turn.
Kavehrad adds, "The usual techniques employed to suppress interference use adaptive spatial filters which require expensive RF components and a large number of computations to queue the subscribers' signals. However, with our approach, we need only a simple, cost effective spatial filter and relatively fewer computations. Our simulations show that the performance of the new approach and the traditional technique are comparable. Thus, our strategy shows a practical compromise between complexity and cost, while achieving the desired signal quality."
The research was supported by Penn State's CICTR and a grant from the National Science Foundation.