稀疏自适应Volterra滤波的QRD-RLS算法

QRD-RLS Algorithm for Sparse Adaptive Volterra filtering

求解非线性问题时,非线性Voherra滤波的性能明显优于线性滤波;稀疏结构Volterra滤波器有效降低了Volterra滤波的工程应用复杂度;本文针对稀疏Volterra滤波,改进了基于QR分解的RLS算法,在更换滤波器抽头的时候保留了先前的数据信息,从而加速当前抽头的收敛,加快了滤波器核矢量的更新和替换。仿真结果表明改进的QRD-RLS算法具有快速自适应能力;一同验证了稀疏Volterra滤波的有效性。

Nonlinear Volterra filtering outperforms linear filtering in dealing with nonlinear problems. Volterra filter with sparse taps reduces the hardware complexity of applications effectively. Combining with the sparse structure of Volterra filter, we improved the QR decomposition based RLS algorithm by reusing the foregoing decomposed results when replacing the tap contributing least to the output. The proposed algorithm accelerates the convergence of current taps coefficient, quickens the update and replace of the sparse Volterra kernels. Simulation results show that sparse Volterra filtering is efficient and the improved QRD-RLS algorithm converges faster.