An Improved Multi-Stage Detection Technique for CDMA Communications Systems

H.-L. Hung (Taiwan)


Genetic-algorithm, Multiuser detection, Fuzzy Inference System, partial parallel interference cancellation (PPIC)


This paper introduces a new application of the Genetic-algorithm (GA) to the partial parallel interference cancellation (PPIC) in a direct-sequence code-division multiple-access (DS/CDMA) communication system over Rayleigh fading channels. Both the near-far problem and multiuser access interference (MAI) are the main factors to deteriorate the performance of DS/CDMA cellular systems. Parallel interference cancellation (PIC) is an effective method to cancel the MAI at light system load [1]; nevertheless, it suffers significant degradations in performance at heavy system load. PIC (CW-PPIC) proposed in [2] over fading channels. However, the control rules and membership functions in FB-PPIC are obtained by try and error methods [3]. Therefore, the performance of the FB-PPIC cannot approach the optimization for the PPIC schemes. The optimal membership functions and control rules can be achieved by the Taguchi-MHGA [4]. To find the optimal fuzzy inference system and improve the interference cancellation effect, the GA will be adopted in the FB-PPIC in this paper. The GA is uses to search for the optimal fuzzy centers of the fuzzy regions, the optimal parameters (i.e., the peak points and the left and right anchoring points) of the membership function for each antecedent fuzzy set, the optimal number of input membership function. Therefore, the error propagation due to unreliable estimates of the MAI would be minimized. Simulation results show that the proposed GA FB-PPIC scheme is superior to the FB-PPIC [3] in over fading channels.

Important Links:

Go Back