Adaptive filter with variable parameters and zero attractors

Abstract

The invention discloses an adaptive filter with variable parameters and zero attractors, and belongs to the field of digital filter design. The filter mainly uses time-varying step parameters and regularization parameters that control the intensity of the zero attractors to speed up the convergence speed of the adaptive filter and reduce its steady-state offset. The adaptive filter with variable parameters and zero attractors disclosed by the invention has a great improvement in performance compared with the conventional filter when the input is white signal input. The adaptive filter with variable parameters and zero attractors disclosed by the invention can be applied to applications such as echo cancellation and active noise control.

Description

technical field [0001] The invention discloses a variable parameter zero attractor adaptive filter, which belongs to the field of digital filter design. Background technique [0002] Traditional least mean square filter (LMS) and normalized least mean square filter (NLMS) adaptive filters have a wide range of applications in adaptive filters, such as adaptive echo cancellation, active noise control and adaptive noise cancellation Wait. In some applications, the system estimated by the adaptive filter may be sparse. If the LMS or NLMS filter is used, when the step size parameter is large, the convergence speed of the adaptive filter is slow, and when the step size parameter is large When the time is small, the steady-state misadjustment of the adaptive filter is large. [0003] In order to solve the above problems, the literature with l 1 Norm-constrained LMS filter (denoted as l 1 -LMS) [YChen, Y Gu, A O Hero. Sparse LMS for system identification, Proceedings of the IEEE...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the contradiction between the selection of the step size parameter and the convergence speed and steady-state imbalance, the literature proposes a variable parameter l 0 Norm-constrained LMS filter (denoted as VP-RZA-LMS) [Danqi Jin, Jie Chen, Cedric Richard, Jingdong Chen. Model-driven online parameter adjustment for zero-attracting LMS, Signal Processing, 152:373-383], This method has better convergence performance. Compared with the traditional LMS filter, the convergence speed and steady-state misalignment have been greatly improved, but the shape of its zero attractor cannot be adjusted according to the sparsity of the unknown system.

Images