A Neural-Based OTFS Channel Estimator

orthogonal time frequency space (OTFS) systems are considered as the reliable solution for addressing the challenges of high mobility in sixth generation (6G) scenarios. By modulating data across both the delay and Doppler dimensions, OTFS efficiently handles the double spread problems caused by high mobility. Accurate and efficient channel estimation is essential for ensuring reliable data reception in OTFS systems. In this paper, we propose a convolutional neural network (CNN) based channel estimator within the standard OTFS transceiver framework to estimate the transmitted channel from the received OTFS signal. Simulation results demonstrate that our neural channel estimator achieves superior channel reconstruction and outperforms existing neural-based methods. Moreover, we analyze its performance with higher-order modulation and show that it effectively supports such modulation under good channel conditions, improving spectral efficiency and throughput.