Deep Learning-Based Single-Carrier Frequency-Domain Equalization

Single-carrier frequency-domain equalization (SC-FDE) is an effective algorithm against inter-symbol interference, which has been widely used in wireless communication systems. Conventional linear SC-FDE mainly includes three modules: channel estimation, noise power estimation and channel equalization, where each module is optimized individually. In order to jointly optimize the three modules, this paper proposes a SC-FDE method based on deep learning. To reduce the amount of training data for network convergence, one subnetwork is particularly designed for each module. Further, a training mechanism is designed to enhance the generalization capability of the proposed approach by treating each path equally. Numerical results show that the proposed algorithm can converge at a small training data set and has robust channel generalization ability. Moreover, the proposed algorithm achieves better performance in terms of bit error rate than the schemes based on least square channel estimation and minimum mean-square error channel equalization.