基于复值多态连续Hopfield网络的MPSK信号盲检测  

Blind MPSK signal detection based on complex-valued multistate continuous Hopfield neural networks

作　　者：钱建平[1] 张志涌[1] 阮秀凯[2] 

机构地区：[1]南京邮电大学自动化学院,江苏南京210003 [2]南京邮电大学通信与信息工程学院,江苏南京210003

出　　处：《重庆邮电大学学报（自然科学版）》2012年第3期292-296,共5页Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)

基　　金：国家自然科学基金(60772060)~~

摘　　要：为实现多进制相移键控(M-ary quaternary phase shift keying,MPSK)信号的直接盲检测,给出一种直接盲检测发送信号的性能优化函数,进而提出一种新的复值多态连续激励函数形式来提高网络性能,并设计出一个新的能量函数来分析网络的稳定性,以求利用复值Hopfield神经网络来解决直接盲检测发送信号性能函数的优化问题。计算机仿真结果表明,与子空间算法等经典算法相比,新方法具有较好的收敛性,能有效实现相移键控信号盲检测。An optimization performance function is constructed to detect M-ary phase shift keying（MPSK） signals blindly and directly.To use complex-valued Hopfield neural network（CVHNN） for solving above optimization performance function,a novel complex-valued multistate continuous activation function is given to improve network performance,and a new energy function is proposed to prove the stability of CVHNN with the given activation function.Computer simulation results show that the proposed method has faster convergence for the blind detection of MPSK signals than classical algorithms such as subspace algorithm and so on.

关 键 词：复值Hopfield神经网络 多进制相移键控(MPSK) 盲检测 复Sigmoid函数 

分 类 号：TN911.23[电子电信—通信与信息系统]