记忆非线性功率放大器的神经网络预失真

Neural Network Predistortion Technique for Nonlinear Power Amplifiers with Memory

数字预失真是克服高功率放大器(HPA)非线性失真最有前途的一项技术。早期对预失真技术的研究大多局限于无记忆非线性,但对于宽带应用,放大器的记忆特性明显。该文提出了一种新的有记忆非线性功率放大器的神经网络预失真技术,预失真器利用输入信号的同向和正交分量作为输入,采用带抽头延时的双入双出两层前向神经网络结构,根据非直接学习结构和反向传播算法实现自适应,可同时补偿放大器的记忆失真和非线性失真。仿真结果表明,建议的方案能有效抑制带外谱扩散,降低误码率,实现有记忆非线性HPA的自适应预失真。

Digital predistortion is the most promising technique to overcome the nonlinearity of High Power Amplifier(HPA).The early work on the field of predistortion is mostly limited to memoryless nonlinear.However,memory effects are typically observed in high power amplifiers for wideband applications.In this paper,a two-layer forward neural net-work predistorter with two inputs and two outputs is proposed for HPA with memory.The predistorter is realized using indirect learning architecture associated with the Backpropagation algorithm.This technique allows us to correct for gen-eral nonlinearities and memory effects simultaneously.Simulation results show that the proposed neural network predictor can effectively suppress spectral regrowth and reduce in-band distortion for HPA with memory.