一种变步长解相关NLMP快速算法

A Varible Step Size Decorrelation NLMP Fast Algorithm

提出了一种在α稳定分布信号条件下能快速稳定收敛的变步长、解相关NLMP(VSDNLMP)算法。根据输入信号的相关性原理,提出基于P阶分数相关的解相关方法,改进NLMP算法的更新方向向量,提高算法的收敛速度;同时采用变步长因子进一步提高NLMP算法性能。在高斯和非高斯α稳定分布有色信号环境下的仿真结果表明:在强相关信号条件下,VSDNLMP算法性能明显优于NLMP算法;VSDNLMP算法的收敛速度和稳态失调都较之NLMP算法有较大的改善;该文算法具有良好的鲁棒性。

This paper proposes a variable step size decorrelation normalized LMP algorithm(VSDNLMP),which has performance of fast and steady convergence under the α-stable distribution conditions.Based on the correlation theory,the fractional p-order decorrelation method is proposed in this paper to ameliorate the convergence speed.The performance of the decorrelation NLMP algorithm is improved further by the means of variable step size.The comparative simulations under Gaussian and non-Gaussian α-stable distribution ...