一种基于多项式函数的VSSNLMS算法

为了改善现有变步长NLMS自适应算法存在的步长调节参数较多、不易调节、对噪声较为敏感的缺点,综合已有的变步长算法提出一种基于多项式函数的变步长归一化LMS(VSSNLMS)算法。理论分析和仿真结果表明,新算法在低信噪比条件下具有较快的收敛速度和较小的稳态误差,而且参数调节简单,适用于不同的输入信号带宽。新算法相比其他同类算法实时计算量小,具有很好的工程实用价值。

A Variable Step Size Normalized LMS Algorithm Based on Polynomial Function

The existing variable step size normalized LMS algorithm is quite sensitive to noise and its step size is relatively difficult to adjust. In order to overcome the above shortcomings, a new variable step size normalized LMS algorithm based on polynomial function is proposed. The theoretical analysis and simulation results demonstrate that the proposed algorithm with high convergence speed, low steady error and fixed parameter is suitable to different signal bandwidths in low SNR condition. The new algorithm is superior to other variable step size algorithms on real-time computation and engineering implementation.