The weighted wavelet packet based LMS algorithm and its application to channel estimation

A weighted wavelet packet based least mean square algorithm (WWP-LMS) applied to estimate slowly time varying channels is proposed. In this algorithm, the training and desired signals are both projected onto wavelet packet subspaces. The mean square errors (MSEs) are weighted in different subspaces and combined to act as the system error criterion; while the LMS algorithm is used to approximate the channel response. Theoretical analysis and simulation results show that the proposed algorithm is superior to the traditional LMS in estimation error performance and convergence rate, especially as unexpected narrowband interference is present.