2021 IEEE 5th Advanced Information Technology, Electronic and Automation Control Conference (IAEAC) | 2021
Digital predistortion technique based on improved LMS algorithm
Abstract
Digital predistortion is an effective method to improve the nonlinear distortion of power amplifier. As the traditional LMS algorithm is slow in convergence and easy to be disturbed by noise and there is convergence instability in the optimal solution, it cannot converge quickly and accurately to identify the inverse model of power amplifier behavior, which affects its predistortion effect. In this paper, an improved NVSLMS algorithm is proposed to improve the anti-noise capability of the predistortion system by changing the iterative function to improve the convergence speed and introducing the autocorrelation matrix of the adjacent moment error. The simulation results show that the predistortion system based on the improved New variable step LMS (NVSLMS) algorithm is significantly superior to the traditional adaptive algorithm in anti-noise performance, convergence speed and out-of-band suppression. EVM index is optimized by 4.3%, and ACPR index is optimized by -8.02dB.