In this paper a switching multiuser detector for CDMA systems is proposed which dynamically switches between NN tuned SIC (Neural SIC) and matched filter operations depending upon the channel characteristics. The switching detection process significantly reduces the computational complexity while maintaining the quality of service requirements such as it error rate, throughput and delay constraint. The receiver processes the received signal matched filter and it performs certain estimation on output of the matched filter. The estimates are used to decide whether to perform NNSIC or not. The work has been carried out in a Synchronous environment with AWGN and Rayleigh fading channels. The neural tuning is done by means of back propagation technique with feedforward network. Comparisons of BER vs SNR (Signal-to–Noise Ratio), with number of users involved, near- far and switching cases with SIR (Signal-to-Interference Ratio) are also considered. The simulation results prove that the derived model has good performance than the existing and conventional receivers. Simulations were done in Matlab environment with different channel characteristic, channels and users scenario with neural, neural followed by SIC for analysis and justification.

Keywords: Multiuser Detection (MUD), Channel, CDMA, Neural SIC, Users, SNR.