频率分集多小波神经网络盲均衡算法  

Multiwavelet Transform Neural Network Blind Equalization Algorithm Based on Frequency Diversity

作　　者：纪娟娟[1] 张兰芳[1] 唐飞[1] 

机构地区：[1]安庆师范学院物理与电气工程学院,安徽安庆246133

出　　处：《安庆师范学院学报（自然科学版）》2014年第4期65-68,73,共5页Journal of Anqing Teachers College(Natural Science Edition)

基　　金：安庆师范学院青年科研基金项目(KJ201316;KJ201211)资助

摘　　要：在分析多小波神经网络盲均衡的基础上,提出了频率分集多小波神经网络盲均衡算法(FD-MWTFNN),该算法利用多小波的特性,加快算法的收敛速度,采用具有对称特性的误差函数减小算法的均方误差,为了进一步提高算法的性能,引入频率分集技术克服多径衰落的影响。水声信道的仿真结果表明,该算法具有收敛速度快,稳态误差小的特点。By analyzing multiwavelet transform neural network blind equalization algorithm , multiwavelet transform neural network blind equalization algorithm based on Frequency Diversity （ FD-MWT-FNN） is proposed.The proposed algorithm uses the characteristics of multiwavelet transform to improve convergence rate and uses error function with characteristics of odd symmetry to reduce mean square error, to further improve the performance of the algorithm frequency diversity is introduced to overcome the im-pact of multipath fading.Simulation tests with underwater acoustic channel indicate that the proposed algorithm has not only faster convergence rate but also less mean square error.

关 键 词：频率分集 多小波 神经网络 盲均衡 水声信道 

分 类 号：TN911.7[电子电信—通信与信息系统]