DS/CDMA Multiuser Detection with Evolutionary Algorithms

Fernando Ciriaco (State University of Londrina - UEL, Brazil)

Taufik Abrão (State University of Londrina - UEL, Brazil)

Paul Jean E. Jeszensky (Polytechnic School of São Paulo University - EPUSP, Brazil)

Abstract: This work analyses two heuristic algorithms based on the genetic evolution theory applied to direct sequence code division multiple access (DS/CDMA) com­ munication systems. For different phases of an evolutionary algorithm new biological processes are analyzed, specially adapted to the multiuser detection (MuD) problem in multipath fading channels. Monte Carlo simulation results show that the detection based on evolutionary heuristic algorithms is a viable option when compared with the optimum solution (ML ­ maximum likelihood), even for hostile channel conditions and severe system operation. Additionally, a comparative table is presented considering the relation between bit error rate (BER) and complexity as the main analyzed figure of merit. Each algorithm complexity is determined and compared with others based on the required number of computational operations to reach de optimum performance and also the spent computational time.

Keywords: code division multiple access, evolutionary computation, genetic algorithm, multiuser detection

Categories: D.2.2, J.2