摘要
由单层神经自适应滤波器和Chebyshev正交多项式滤波器级联构成一个均衡器,基于此均衡器结构,提出一种基于分数低阶统计量的自适应均衡算法(称为NC-NLMP算法)。仿真结果表明,对线性信道和非线性信道,NC-NLMP算法的均衡性能均优于传统的FIR横向滤波器的自适应均衡算法和神经网络均衡算法;对于非线性信道,NC-NLMP算法的均衡性能远优于后两者;NC-NLMP算法对高斯噪声和稳定分布脉冲噪声都具有很好的均衡性能,对背景噪声韧性强。
An adaptive equalizer algorithm(named as NC-NLMP algorithm) is proposed based on the nonlinear equalizer structure which consists of a single layer neural adaptive filter and a Chebyshev orthogonal polynomials filter.The simulation results indicate that the performance of the NC-NLMP algorithm is much better than that of FIR transverse equalizer algorithm and neural network equalizer algorithm,no matter what the channel is linear or nonlinear.Especially for nonlinear channel,the predominance of the NC-NLMP algorithm is notable.And the NC-NLMP algorithm can achieve excellent performance,whether the background noise of channel is gauss or α-stable distribution pulse noise,that is to say,the NC-NLMP algorithm is robust to background noise.
出处
《电视技术》
北大核心
2009年第S1期15-18,共4页
Video Engineering
基金
浙江省自然科学基金项目(Y106839)
