DS-CDMA blind detection for frequency-selective multipath channels by neural networks

Abstract

Up to now, various detection algorithms have been offered and investigated for DS-CDMA systems in multipath conditions. Here, We intend to implement sub-optimum receivers based on Maximum-Ratio-Combining (MRC) via neural network structures in Downlink systems. We will demonstrate that our design based on Radial Base function (RBF) and Multi Layer Perceptron (MLP) outperform in comparison of conventional detectors such as Match-Filter, Decorrelator Detector (DD) and MMSE manner. We also propose a new method when receiver doesn't know sequence code in CDMA receiver system and illustrate that RBF is proper when number of users are low and MLP is prefer where the number of users increased.