摘要
在卫星通信中,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。提出一种基于小波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,该算法不仅能明显改善信号星座图,并与基于RBF神经网络的自适应预失真算法相比能提高收敛速度,同时HPA线性化性能也有所提高。
The more bandwidth efficient modulation schemes are used for high speed data transmission systems in the satel lite 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 predis- tortion based on wavelet neural network for HPA onboard the satellites is proposed to linearization HPA. And the adaptive a rithmetie of stochastic gradient method is derived. The simulation results show that the scheme can not only correct the con stellation distortion obviously, but also converge faster than the adaptive predistortion based on REF neural network, and improve the linearization performance greater.
出处
《现代电子技术》
2007年第24期23-26,共4页
Modern Electronics Technique
关键词
小波神经网络
高功率放大器
非线性
预失真
wavelet neural network
high power amplifier
nonlinearity
predistortion
