基于模拟退火并行进化规划的MIMO-CDMA多用户检测器

Multiuser detector based on simulated annealing parallel evolutionary programming for MIMO-CDMA systems

为解决最佳多用户检测计算复杂度高的难题,将多种群并行进化规划和模拟退火两种思想有机地融合起来,提出一种新的模拟退火并行进化规划算法,并应用到MIMO-CDMA系统多用户检测问题求解中。在新算法中,不同的子群并行进行进化过程,利用模拟退火的局部寻优能力,避免单个种群在进化过程中出现的早熟现象,从而加快整个算法的收敛速度。实验结果证明,新的多用户检测器抗远近效应和抗多址干扰的能力都优于传统检测器和单种群的模拟退火进化规划多用户检测器,并且在远近效应和多址干扰存在的条件下,该检测器的收敛速度比单种群模拟退火进化规划检测器提高了约108%和100%。

To resolve high complexity of optimum multiuser detection, Multigroup parallel Evolutionary Programming and simulated annealing are well combined in this paper. The paper apply the hill climbing performance of simulated annealing and the global optimum performance of evolutionary programming, A novel algorithm based on simulated annealing parallel evolutionary programming is proposed and applied to address the MIMO-CDMA muhiuser detection problem. In the algorithm, evolutions of subgroups are parallely performed among subgroups, and the local optimum of simulated annealing is utilized, so this algorithm avoids premature convergence of alone group evolution- ary process and accelerates the convergence speed of the algorithm. Simulation results show that the new muhiuser detector is superior to the conventional detector and the alone group simulated annealing evolutionary programming muhiuser detector in the aspects of near-far effect and multiple-access interference, and the convergence speed in- creased by about 108% （ near-far effect ） and 100% （ multiple-access interference） than the alone group simulated annealing evolutionary programming detector.