自适应增益APA-AG算法的研究  

A Better Affine Projection Algorithm(APA) Using Adaptive Gain(AG)

作　　者：智永锋[1] 范蟠果[1] 张骏[1] 邓正宏[1] 张怡[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072

出　　处：《西北工业大学学报》2011年第3期449-453,共5页Journal of Northwestern Polytechnical University

基　　金：西北工业大学基础研究基金(JC20100217)资助

摘　　要：为了提高自适应滤波器的收敛性和跟踪性,提出了一种自适应增益APA-AG算法。通过分析估计输出误差信号,获得了参数迭代步长的一个特定值估计,以实现集平均代价函数最小化。仿真结果表明,相比较于传统的APA算法,文中研究的APA-AG算法具有更快的收敛速率和更好的跟踪特性。Aim. To improve the convergence rate and tracking performance of an adaptive filter, esPecially for the highly colored input signals, Refs. 3 and 4 adopt the APA algorithm. However, its gain does not change, and thus its convergence is not as rapid as is possible, in our opinion, with adaptive gain. We explain mathematically in sections 1, 2 and 3 of the full paper our APA-AG algorithm and discuss how to minimize the ensemble-average cost function by analyzing the estimated output-error. The core of section 2, which is entitled the estimated output-error analysis of the APA algorithm, is that we derive eq. （23）. The core of section 3, which is entitled APA-AG algo- rithm, is that, to minimize the ensemble-average cost function, we obtain the estimated iterated step-sizes of our APA-AG algorithm; the needed equations are eqs. （23）, （3c） , （28） , （27） , （26） and （30）. Section 4 presents two numerical examples to simulate our APA-AG algorithm; the simulation results, shown in Figs. 1 and 2, and their analysis indicate preliminarily that our APA-AG algorithm has faster convergence rate and better tracking performance compared with the conventional APA algorithm.

关 键 词：自适应滤波 算法 APA算法 自适应增益 

分 类 号：TN713[电子电信—电路与系统]