Variable Step Size LMS Adaptive Algorithm Based on Exponential Function
Donghong He,Mingjiang Wang,Yufei Han,Shifei Hui
DOI: https://doi.org/10.1109/icicsp48821.2019.8958492
2019-01-01
Abstract:The value of the step factor affects the performance of all aspects of the LMS algorithm, including convergence speed, steady-state error, and anti-interference ability. The traditional fixed-step LMS algorithm performs a weight update step factor with a fixed value. If the step factor has a large value, the convergence speed is fast but the steady-state error is also large. If the step factor value is small, the steady state error is small but the convergence speed is reduced. It is difficult for the fixed step to satisfy the requirements of both the convergence speed and the steady-state error of the algorithm. Therefore, this paper proposes a variable step size LMS adaptive algorithm based on exponential function. The algorithm leads into controllable parameters to adjust the step size factor, and analyzes the influence of the values of each parameter in the algorithm on the step factor. The final model of the algorithm is obtained through multiple experiments. The simulation results show that compared with the existing algorithms, the proposed algorithm has faster convergence speed and smaller steady-state error.