一种新颖的动态选择输入向量的仿射投影算法

Novel Affine Projection Algorithm with Dynamic Selection of Input Vectors

仿射投影算法利用多个输入向量更新滤波器系数,不同输入向量对滤波器性能影响不同。以均方偏差下降最快为目标,推导了动态选择输入向量的理想条件,该条件适用于仿射投影算法类中的多种算法。在此基础上,提出了一种便于实现的动态选择输入向量的仿射投影算法。理论分析与仿真结果表明:与传统的仿射投影算法相比,该算法具有收敛速率快、稳态误差小和计算复杂度小等优点。

Multiple input vectors are used to update the filter coefficients in the affine projection algorithm （APA） family,and different input vector affects the filter differently. The ideal selection condition of the input vectors was derived by the largest decrease of the mean-square deviation. This selection condition fits for different algorithms of the APA family. An APA family with dynamic selection of input vectors was proposed,which could be realized easily. The theoretical analysis and computer simulation results illustrate that,compared with the conventional APA,the proposed algorithm has a fast convergence speed,a small steady-state error and a small computational complexity.