Performance Analysis Of Linear Multiuser Detectors And Neural Network Detector In Non- Gaussian Noise Channel

In multiuser detection techniques, the maximum likelihood multiuser detector provides the best performance in DS-CDMA systems with impulsive ambient noise. But the high complexity makes it impractical. And therefore, effective sup-optimal detectors in non-Gaussian noise are needed. In this paper, we present two-layer perceptron neural network with back propagation training algorithm as a multiuser detector of synchronous DS-CDMA system with nonGaussian ambient noise. We provide some simulation examples to analyze and compare the performance of the neural network detector with the decorrelator and MMSE (linear multiuser detectors) against MAI, Gaussian and non-Gaussian additive noise. Simulation results show that the performance of the examined detectors degrades in the presence of non-Gaussian noise than in AWGN. However, the neural network detector performs better than the linear multiuser detectors. Keywords—DS – CDMA, Multiuser detection, impulsive noise, Decorrelating detector, MMSE detector, neural network detector.