RBF network based receiver design for multiuser detection in SDMA-OFDM system

Abstract

In this paper, the computationally efficient Radial Basis Function (RBF) Neural Network (NN) model is suggested for multiuser detection in the context of Space Division Multiple Access - Orthogonal Frequency Division Multiplexing (SDMA-OFDM) system to achieve a significant low computational complexity than the optimal ML detector. The classical Minimum Mean Square Error (MMSE) and Maximum Likelihood (ML) Multiuser Detection (MUD) techniques are suffer with poor performance and high complexity respectively. Although, the optimization techniques aided ML detection techniques are less complex compared to ML detector, but these techniques are still complex as those require an additional channel estimation block. Unlike these existing techniques, the suggested blind multiuser detection using RBF NN structure performs better in terms of BER performance with low complexity. In addition to that, as GA-ML detector the RBF aided MUD also have capability of detecting users in over load scenario, where number of users are more than that of number of receiving antennas.