针对非高斯α稳定分布噪声环境下的自适应滤波问题,提出一种新的基于梯度加权平均的变步长归一化最小平均p范数(VSS-NLMP)算法。该算法基于梯度矢量能够跟踪自适应过程的均方权值偏差(MSD)这一特点,通过对梯度矢量的平滑减小梯度噪声的影响,然后利用平滑梯度矢量的欧氏范数和系统误差的分数低阶矩控制步长的变化。文中给出了新算法的迭代过程,然后对其收敛性进行简要分析,仿真结果表明,本文新算法较现有变步长NLMP算法具有更快的收敛速度和更高的稳态精度。
For the problem of adaptive filtering in non-Gaussian alpha stable distribution noise environment, a variable step-size normalized least mean p norm (VSS-NLMP) algorithm with gradient-based weighted average is proposed. The Euclidean norm of the smoothed gradient vector, which can track the variation of the mean square deviation ( MSD ) at iteration, and fractional low order moments of the system error are used to update the step-size parameter in recursion. The weighted average of the gradient vector reduces the noise effectively. The update and convergence of the proposed algorithm are formulated in this paper. The simulation results indicate that the proposed algorithm has better performance compared to the existing VSS-NLMP algorithms.