判决反馈递归神经网络自适应均衡器研究

Study of the decision feedback recurrent neural network equalization with the self adjusting structure

针对存在严重符号间串扰及轻度非线性畸变的数字信道，提出了两种判决反馈递归神经网络均衡器结构，将传统的线性信道判决反馈结构巧妙地融入递归神经网络中，在自适应训练时用期待信号代替判决反馈信号，并对学习步长自适应调节进行了研究．实验结果表明：在自适应参数数目相同的情况下，该均衡器较传统的递归神经网络均衡器具有更好的特性．

Two decision feedback equalizer structures using recurrent neural networks for non linear channels with severe Inter Symbol Interference (ISI) and mild non linear distortion are proposed, which put the traditional decision feedback structure for linear channels equalization skillfully into the recurrent neural networks, substitute the training signal for the decision feedback signal in the learning process and adaptively adjust the learning step. Simulation results of the first type of two new equalizer structures have shown that it has better equalization performance than the traditional recurrent neural network equalizer with the same number of parameters.