基于RBF神经网络的HPA自适应预失真算法  

Adaptive Predistortion Based on RBF Neural Network for HPA

作　　者：李爱红[1] 肖山竹[1] 张尔扬[1] 

机构地区：[1]国防科技大学电子科学与工程学院,湖南长沙410073

出　　处：《国防科技大学学报》2008年第3期105-108,共4页Journal of National University of Defense Technology

基　　金：国家部委预研基金资助项目(113030401)

摘　　要：卫星通信中,高速数据传输系统要求使用频谱利用率高的高阶调制技术,但高阶调制对高功率放大器(HPA)的非线性非常敏感,会造成码间干扰和邻信道间干扰。提出一种基于RBF神经网络的自适应预失真算法,以实现HPA的线性化,同时推导了自适应算法的迭代公式。仿真结果表明,该算法能明显改善信号星座图,并能大大提高系统的误比特率性能。The more bandwidth efficient modulation schemes are used for high speed data transmission systems in the satellite communication. But these schemes are sensitive to the nonlinearity of high power amplifiers （HPA）. The nonlinearity of HPA causes inter symbol interference （ISI） and adjacent channel interference （ACI）. A scheme for adaptive digital predistortion based on RBF neural network for HPA onboard the satellites was proposed to linearization HPA . And the adaptive arithmetic of stochastic gradient method was derived. Simulation results show that the scheme can not only correct the constellation distortion obviously, but also improve the system BER performance gready.

关 键 词：RBF神经网络 高功率放大器 预失真 

分 类 号：TN911[电子电信—通信与信息系统]