Hysteretic hopfield neural network detection of multi-user CDMA signals

Abstract

Multi-user reception using Hopfield Neural Network (HNN) and Hysteretic Hopfield Neural Network (HHNN) for Multi-Carrier Code Division Multiple Access (MC-CDMA) is investigated. We have shown that by the appropriate choice of Hopfield Neural Network parameters from the communication system, Hopfield Neural Network can collectively resolve multiple-access interference in the system. The "hysteretic" Hopfield Neural Network detector performance is investigated and compared in various asynchronous, Gaussian and multipath channel conditions with respect to varying near-far conditions expressed as E<inf>i</inf>/E<inf>1</inf>. The investigations have shown that the application of hysteresis provides some performance improvement in terms of BER. The BER improvement provided by hysteresis is more noticable at higher desired signal to noise ratios E<inf>1</inf>/N<inf>o</inf>. © 2005 IEEE.