一种新型自适应最小均方算法

针对最小均方(LMS)算法应用于功率放大器时存在收敛速度与收敛精度相矛盾的问题,提出了基于步长比较的最小均方算法。在带有P因子的变步长最小均方算法的基础上,采用简化的Sigmoid函数对步长进行改进,通过对前后两次步长的比较来确定是否更新权系数,以误差的自相关时间均值及均方误差的时间均值来调节算法步长,可以加快算法的收敛速度,降低算法的计算量。仿真结果表明,与最小均方算法相比,经过自适应预失真处理后,该算法的误差向量幅度(EVM)值提高了2.653 2%,系统邻信道功率比(ACPR)减少了4 dB。

A Novel Adaptive Least Mean Square Algorithm

Aiming at the problem that the convergence rate and the convergence precision were contradictory when the least mean square(LMS) algorithm was used in the power amplifier, a kind of least mean square algorithm based on the comparison step was proposed. Based on the least mean square algorithm of variable step size with P factor, the step size was improved by the simplified Sigmoid function. Whether to update weight coefficient was determined by comparing the two steps of the before and after step. The convergence rate of the algorithm could be accelerated, and its calculation amount could be reduced by adjusting the algorithm steps with autocorrelation time mean values of error and time domain average values of mean square error. The simulation results showed that after adaptive pre-distortion processing, the error vector magnitude(EVM) of the proposed algorithm was improved by 2.653 2 % compared with that of the least mean square algorithm, and the adjacent channel power ratio(ACPR) of the system was reduced by 4 dB.