Weight optimum for weighted polynomial expansion multiuser detector

Abstract

Weighted polynomial expansion (WPE) detector can be used to approximate linear multiuser detectors to avoid the computation of inverse of matrix. For finite stage WPE detector, weight vector optimum problem is very important to efficiently approximate linear multiuser detectors. In this paper, the optimal weight vector for the finite stage WPE detector is derived based on mean square error (MSE) criteria. And a low complexity adaptive WPE detector based on least square (LS) criteria is presented to simplify the computation of weight vector, because the direct computation of the optimal weight vector is complex in practical implementations. Simulation results show that the adaptive WPE detector can nearly achieve the performance of the exact optimal weight vector WPE detector with greatly reduced complexity.