论文部分内容阅读
为了改善Sigmoid函数变步长LMS算法(SVS-LMS)在高斯噪声和冲激噪声干扰下的性能,改变了步长调整自变量,原SVS-LMS算法以瞬间误差功率控制步长更新,新算法以误差的自相关时间均值估计调节步长,抑制了噪声干扰;使用Hassab和Boucher(HB)加权进一步平滑了因噪声干扰导致的自适应滤波器权系数伪峰、利用归一化处理方法获得了更大的输入信号动态范围.自适应时延估计仿真实验结果表明,在高斯噪声和冲激噪声干扰下,相比于固定参数下的SVS-LMS算法,新算法及其HB加权能获得更好的时变时延跟踪均方误差性能.
In order to improve the performance of Sigmoid function variable length LMS algorithm (SVS-LMS) under the interference of Gaussian noise and impulsive noise, the step size adjustment independent variable is changed. The original SVS-LMS algorithm is updated with the instantaneous error power control step size. The step-length is estimated by the mean value of the error autocorrelation time, and the noise interference is suppressed. Hassab and Boucher (HB) weighting further smoothes the pseudo-peak of the weighting coefficient of the adaptive filter caused by the noise interference. The normalized method is used to obtain the And the dynamic range of the input signal.Adaptive delay estimation simulation results show that the new algorithm and its HB weighted better than the SVS-LMS algorithm with fixed parameters under the interference of Gaussian and impulsive noise The time-varying delay tracking mean square error performance.