量子遗传算法优化神经网络的MIMO-OFDM检测研究

Research of MIMO-OFDM detection based on Neural Network optimized by Quantum Genetic Algorithm

信号的最优检测在常规条件下是一NP难解问题,针对RBF(径向基函数)神经网络算法易陷入局部极值和简单遗传算法收敛速度慢的问题,提出一种新型智能算法并将其用于MIMO-OFDM系统信号检测中:该算法将量子计算、遗传算法与神经网络相结合,用量子遗传算法(QGA)优化神经网络初始值。由于QGA给RBF网络提供了较好的初始值,故能够使RBF网络快速收敛到最优解,避免了由初始值的随机选取而带来的检测误码。实验结果表明,该算法能够有效地提高系统的信号检测性能,降低误码率。

The optimal solution of signal detection is a Nondeterministic Polynomial（NP） problem.Aimed at the problems that Radial Basis Function（RBF） Neural Network is prone to the local optimum and simple genetic algorithm has the shortcoming of slow convergence,a new type of intelligent algorithm is proposed and applied into the MIMO-OFDM detection systems：it makes use of Quantum Genetic Algorithm（QGA） to optimize the initial data of RBF neural network.In this scheme,the output of detector by the QGA as the input of detector by neural network to avoid the bit-error rate for selecting initial data randomly and improve further the detection property.Simulation results show the proposed method is good for the improvement of the detection rate and reduction of bit-error rate.