使用变步长频域LMS算法的自适应Volterra均衡器

Volterra均衡器能够有效地克服卫星信道的非线性失真,但由于非线性均衡器的输入矩阵特征值扩展严重,使得自适应过程收敛缓慢。为克服这个缺点,提出应用变步长的频域LMS算法对Volterra均衡器的权值系数进行自适应更新。算法利用正交变换降低输入序列的相关性,同时动态地调整迭代步长提高均衡器的收敛速度。仿真结果表明与时域算法相比,均衡器的收敛速度提高了25倍左右;均衡器收敛后纠了信号的幅度和相位失真。

Adaptive Volterra equalizer based on variable step-size frequency domain LMS

Volterra equalizer is an effective technique to compensate the nonlinearity for satellite channels.However,the eigenvalue spreads of input auto-correlative matrix make the equalizer converge slowly.To counteract this problem,variable step-size frequency domain LMS(FDLMS) algorithm is applied to the Volterra equalizer.The algorithm can suppress the correlation of the inputs and adjust the step-size dynamically,so as to raise the convergence rate.Simulation results indicate that the convergence rate is 25 times as time domain algorithm.The converged equalizer also corrects signal amplitude and phase distortions.