Adaptive Equalization for QAM Signals Using Gated Recycle Unit Neural Network
Hongtai Shi, Tianfeng Yan · 2021
In this paper, an adaptive equalization algorithm for 4-Quadrature Amplitude Modulation (QAM) signals based on the Gated Recycle Unit (GRU) neural network is proposed. The GRU can avoid the problem of gradient disappearance in the Recurrent Neural Network (RNN), and the amount of parameters of GRU is small which reduces the risk of overfitting. Experimental results show that the proposed GRU based adaptive equalization algorithm can effectively reduce the distortion caused by the non-ideal characteristics of the channel to the transmitted 4-QAM signals. Compared with other neural network-based adaptive equalization algorithms, it has lower Mean Square Error (MSE) and faster convergence speed.