Optimization on OTFS Modulation Channel Estimation Path Employing CNN-based Self-Adjustment Model

Channel estimation performance is an important evaluation parameter in Orthogonal Time Frequency Space (OTFS) modulated systems, in which the pilots in the delay-Doppler (DD)-domain are estimated at the receiver point. The proposed adjustment model will collect results of the history transmission progress and utilize these records to adjust the channel estimation path for the next transmission progress. A BP-based learning algorithm is introduced into the adjustment model, which completes the adjustment progress by analyzing pilot final channel selection and other not selected channels in history transmission. Feedback from the adjustment model is used in the next OTFS transmission for reducing channel estimation paths when data/signals with the same DD-domain index are transmitted in another transmission set. The simulation results indicate that the improvement of our proposal under classical Bit Error Rate(BER)-based scheme is competitive in various learning-based OTFS optimization solutions.