基于进化规划和神经网络的多用户检测器

DETECTOR BASED ON EVOLUTIONARY PROGRAMMING AND NEURAL NETWORK FOR MULTIUSER

利用两种方法的优点,并将它们应用于CDMA多用户检测中,提出—种基于并行遗传算法和Hopfield神经网络的多用户检测器.该检测器中,进化规划首先给神经网络提供较好的初始值,然后用神经网络实时快速地获得最优解.计算机模拟结果表明:本文所提出的检测器在误码率和"抗远近"效应方面均具有良好的性能.

The advantages of detection; evolutionary programming (EP) and Hopfield neural network (HNN) are applied to multiuser detection in CDMA communication. A multiuser detector based on EP and HNN is proposed. In this detector, EP provides the initial value to HNN, and then the HNN gets the optimal solution in real time and quickly. Computer simulation results show that the method has powerful properties both in bit error rate and near-far resistance.