参差周期序列的格形滤波器

Lattice Filters Used for Stagger-Period Sequences

本文提出参差周期序列的格形滤波器结构。借助于提出的参差周期Levinson递推关系式(见附录),证明了该格形结构的最佳性。给出了高效的加权调和平均算法的参差格形反射系数估计;讨论了减少反射系数估值方差的递推平均和块平均的方法。通过对它们在平稳和非平稳序列中的学习特性的研究,说明了该类滤波器在自适应处理中的良好性能。

This paper presents a structure of lattice filters with stagger-period sequences. By means of the presented stagger-period Levinson recursion relations (proved in the appendix), the opti-mality of this lattice structure is shown. Based on an efficient weighting-harmonic-mean algorithm, the estimation of staggered lattice reflection coefficients is provided; the methods of recursion-averaging and block-averaging for reducing reflection coefficient estimate variances are discussed. Through researching their learning characteristics in stationary and nonstationary sequences, the good performances of this class staggered lattice filters in adaptive processing are demonstrated.