Exploring the Performance of Receiver Algorithm in OTFS Based on CNN

Abstract

Communication systems based on the Orthogo-nal Time Frequency Space (OTFS) technology can provide reliable information transmissions in the high-speed mobile environment. This paper proposes and analyzes a Convolutional Neural Network (CNN) structure for data-driven OTFS receivers. After using the received signals in the delay-Doppler domain for training, the presented OTFS receiver based on CNN can attain excellent Bit Error Rate (BER) performance. In addition, in real-world communication systems, channel parameters change with the transmission environment and operating frequency variation. The proposed OTFS receiver is an effective scheme to deal with channel distortion, which can achieve good robustness for the change of channel parameters. Through experiments, simulation results demonstrate that the OTFS receiver with the proposed CNN structure better outperforms the comparison algorithms.