Deep-Learning Based Signal Detection for MIMO-OTFS Systems
Yosef K. Enku, Baoming Bai, Shuangyang Li, Mengmeng Liu, Isayiyas Nigatu Tiba
Abstract
In this paper, we propose a DL-based detector model based on two-dimensional convolutional neural network (2D-CNN) which can readily exploit the delay-Doppler channel to learn the multiple-input multiple-output orthogonal time frequency space modulation (MIMO-OTFS) input-output relationship. Thanks to the beneficial properties of OTFS, only a few CNN layers are sufficient to learn the channel features at the cost of a low complexity. Furthermore, we also use a data augmentation technique based on an existing computationally cheaper linear detector to enhance the learning and detection ability of the proposed model. Unlike most DL-based detectors, the proposed model does not need to take the channel as its input to analyze the characteristics of the randomly varying channel during training and online deployment. Simulation results show that 2D-CNN can achieve a better performance compared to existing OTFS detectors.