Particle Swarm Optimization Assisted Multiuser Detector for M-QAM DS/CDMA Systems

This paper analyzes the particle swarm optimization multiuser detector (PSO-MuD) under high-order modulation schemes, (particularly for M-QAM), in DS/CDMA systems single-input-single-output (SISO) multipath channels. In order to avoid the computation of complex-valued variables in high-order squared modulation, the optimization problem is reformulated as a real-valued problem. Considering previous results on literature for low-order modulation formats (usually binary/quadrature phase shift keying - BPSK/QPSK), a performance x complexity trade-off comparison is carried out between PSO-MuD and local search multiuser detector (LS-MuD). Performance is evaluated by the symbol error rate (SER), and complexity by necessary number of cost function calculations for convergence. If the background for BPSK heuristic multiuser detection (HEUR-MUD) problem [1] indicates that the 1-opt local search method is enough to achieve excellent performance x complexity trade-offs, our Monte-Carlo simulation results and analysis show herein indicate the LS-MUD presents a lack of search diversity under high-order modulation formats, while the PSO-MuD is efficient to solve the MUD problem for high-order modulation schemes.