A ResNet-DNN based Channel Estimation and Equalization Scheme in FBMC/OQAM Systems

Abstract

Due to the intrinsic imaginary interference among subcarriers, the channel estimation problem has become one of the main difficulties of the filter bank multicarrier (FBMC) systems. In this paper, we propose a novel channel estimation scheme based on residual networks (ResNet)-deep neural networks (DNN), called as Res-DNN scheme, for the FBMC systems. In the Res-DNN scheme, the conventional channel estimation and equalization module and the demapping module are replaced by a Res-DNN model of deep learning. Simulation results show that the channel estimation performance of the Res-DNN scheme is greatly superior to other schemes in terms of bit error rate (BER).