基于MWF的非线性Volterra滤波

结合Volterra滤波的多人单出(MISO)模型,本文将多级维纳滤波(MWF)应用于非线性Volterra滤波,提出三种实现结构及其降秩滤波算法.MWF将期望信号连续投影到正交的低维子空间,利用嵌套的一组标量维纳滤波器实现滤波,避免求解观测矢量的自相关矩阵及其逆;而基于MWF的降秩滤波算法不需要进行计算复杂的特征值分解.结合中继卫星信道的非线性均衡对算法进行仿真,仿真结果表明三种结构均可以很好地收敛;降秩算法在减小运算量的同时,性能接近全秩算法.

Multistage Wiener Filter Based Nonlinear Volterra Filter

Associating with the Multiple Input Single Output (MISO) model of Volterra filter,this paper applies the Multistage Wiener Filter (MWF) to nonlinear filter,namely MWF based Nonlinear Volterra Filter (MWF_NVF).Three implementation structures are proposed,and the corresponding reduced-rank algorithms are described.MWF successively projects the desired signal onto orthogo- nal,lower dimensional subspaces,and implements the Wiener filter by a nested chain of scalar wiener filters,without the need to calcu- late the observod-data covariance matrix and its inverse.MWF based reduced-rank algorithms needn't complicated eigenvalue decompo- sition.Simulations are carried out in the case of nonlinear satellite channel equalization.The results show that all the three structures converge very well;the reduced-rank algorithms are close to full-rank algorithms in performance,and the computation complexity reduces greatly.