摘要： 针对功率放大器的记忆非线性特性给通信系统带来的失真问题,提出一种改进的滤波器查找表(FLUT)预失真方法。在FLUT方法的基础上,改进FLUT预失真结构及简化自适应更新部分。采用传输窄带序列训练神经网络模型弥补功率放大器的非线性特性,利用二维滤波器码表补偿功放的记忆效应。通过将功率放大器的非线性特性和记忆效应分开处理,降低神经网络模型的计算复杂度。仿真结果表明,改进方法能有效降低系统误码率,抑制带外频谱扩展,减少带内失真,与原FLUT方法相比,对记忆功率放大器有更好的线性化效果。

关键词: 预失真, 滤波器查找表法, 神经网络, 滤波器矩阵, 非线性特性, 记忆效应

Abstract: To solve the problem of distortion caused by the memory and nonlinearity of power amplifier in communication system,an improved Filter Look-up Table(FLUT) predistortion method is proposed.The new method improves the FLUT predistortion structure,and simplifies the adaptive update part.It transmits the narrowband sequence to train the neural network to compensate the nonlinear characteristics of Power Amplifier(PA),and uses 2D filter table to compensate the memory effect.It deals with the nonlinearity and memory effects of power amplifier separately,and reduces the complexity of model coefficient identification.Simulation results show that the improved method can restrain the expansion of spectrum effectively,reduce the bit error rate and inner distortion,thus achieve a better linearization for the PA with effects than FLUT method.

Key words: predistortion, Filter Look-up Table(FLUT) method, neural network, filter matrix, nonlinear characteristic, memory effect

中图分类号: 

• TN91