基于正交小波变换的前馈神经网络盲均衡算法

Blind equalization algorithms based on Orthogonal Wavelet Transform and Feed-forward Neural Network

摘要:

针对前馈神经网络(FNN)盲均衡算法收敛速度慢、均方误差大的缺点，在分析FNN盲均衡算法和正交小波变换(OWT)理论的基础上，提出了基于正交小波变换的FNN盲均衡算法。该算法利用正交小波变换良好的去相关性，对FNN均衡器输入信号进行预处理后，降低了输入信号的自相关性，从而加快了收敛速度和减小了均方误差。水声信道盲均衡的仿真结果表明，该算法在收敛速度与均方误差方面的性能比FNN盲均衡算法优越。

Abstract:

In order to overcome the slow convergence rate and bigger mean square error of Feed-forward Neural Network(FNN) blind equalization algorithm，a FNN blind equalization algorithm based on Orthogonal Wavelet Transform(OWT) was proposed. In the proposed algorithm, orthogonal wavelet transform was prosecuted on input signal of FNN equalizer to reduce the correlation of the input signal by using the de-correlation ability of wavelet transform. Accordingly, the proposed algorithm could improve the convergence rate and reduce the mean square error. The simulation results of underwater acoustic channels showed that the proposed algorithm outperformed FNN blind equalization algorithm in the convergence rate and mean square error.