Nonlinear system identification using DWT/LMS orthogonalizing adaptive algorithms

Considering the deficiency of signal process using traditional least mean square (LMS) algorithm, a discrete wavelet transform algorithm in transform-domain LMS algorithm is presented. The nonlinear filter structure was given based on orthogonal DWT/LMS algorithm. The method is easily implemented and decreases the spectrum dynamic range of auto-correction matrix of input vector, and improves the adaptiveness of the model identification. Compared with traditional LMS algorithm, DWT/LMS algorithm enhanced the convergence speed and stability. The simulation results show that the proposed method is efficient and feasible.