一种改进的HRLS算法

An Improved HRLS Algorithm

HRLS(Hierarchical Lest-Squares Algorithm)算法是一种改进的RLS算法,他减少了运算量,但对FIR信道估计时,他只适用于输入信号是白噪声而且信道的非零系数少的情况。提出一种改进的HRLS算法,该算法在HRLS算法每级后面(最后一级除外)加一个RLS滤波器,其输入分别为同一级其他滤波器输入的均值和前一级新加滤波器的输出。此滤波器回收了被HRLS分组分级处理所丢掉的组间信号的联系,把握住了输入信号的整体变化规律。仿真结果和分析表明该算法与HRLS相比具有较好的收敛性,而且当把FIR信道划分后所得的每个子信道的自身系数间的差异不大时,其收敛性能与RLS算法近似,同时与RLS相比,其复杂度降低,且具有更强的抗噪能力。

HRLS algorithm makes improvement to RLS algorithm for its complexity reduction, however,it is only adaptive to the FIR channel when it is used for channel estimation. In this paper,an improved HRLS algorithm is proposed. This algorithm adds a RLS unit at the end of each level of HRLS scheme except the last level. The mean of inputs in each elder corresponding unit at the same level and the output of the new added unit of the upper level consist the inputs of the new added RLS unit. Therefore,the new added unit in each level collects some information lost by the hierarchical approach of HRLS, and commands the total changing behavior of the input signals. From the simulations and the analysis we can confirm that in the area of convergence performance,the proposed algorithm is better than HRLS algorithm in whatever circumstance,and close to RLS algorithm when the coefficients of every subchannel obtained after the division of FIR channel have a little difference from each other. Furthermore,the proposed algorithm has much stronger noise resistance than RLS algorithm.