No abstract

No abstract

In this paper, we present fundemental linear multiuser detection (MUD) techniques and compare them with the technique based on neural network (NN) in multicarrier code division multiple access (MC-CDMA) systems. In a MC-CDMA system, increasing with the number of users, receiver's bit error rate (BER) performance goes up. Also, the system' s performance is effected by the power level differences among the users. Simulation results demostrate the higher performance of NN receiver compared to conventional receiver (matched filter) for MC-CDMA. And also, performance results show that the NN structure, gives nearer results comparing to decorrelator and MMSE receivers. Simulations implemented in MATLAB program and performances are examined for synchronous communication and AWGN channels. The Levenberg-Marquardt algorithm is used as the learning algorithm for NN.