一种改进的集员滤波仿射投影算法

摘要

Affine projection adaptive filtering algorithm based on the set-membership concept allows the reduction of the frequency of updates of the filter coefficients. However, the computational complexity of adaptive algorithm is proportional to the number of filter coefficients. To deal with this obstacle that the large number of filter coefficients diminishes the usefulness of the adaptive filtering algorithm owing to increased complexity, a modified affine projection algorithm is proposed by employing a data selection strategy based on a partial update technique and the setmembership concept. The new algorithm focuses on updating a selected subset of the filter coefficients at each iteration, when the output estimation error is upper bounded by a pre-determined threshold. Echo cancellation simulation results not only verify that the new adaptation algorithm can further reduce the computational complexity when compared with the affine projection algorithm based on the set-membership concept, but also can obtain a faster convergence rates than their conventional counterparts.%在优化算法的研究中,集员滤波的仿射投影算法具有比传统仿射投影算法迭代次数少、计算量小等优点,但实际应用中,算法的计算复杂度与滤波器长度成正比,当滤波器长度较大时限制了实时实现.为减少误差,实时运行,提出了一种改进算法.算法中当输出误差小于给定误差门限时,滤波器系数不必进行调节;否则将滤波器系数分成多个系数子集,通过仅更新权系数某一子集的方法,减少了原算法每次迭代中需更新的抽头权系数个数.采用一个声回波消除方法进行仿真.仿真验证了改进算法的性能,表明新算法与基于集员滤波的仿射投影算法相比,具有更快的收敛速度,并降低了计算复杂度.