Deep Learning Based Target Parameters Estimation for OTFS-ISAC System

Orthogonal time frequency space (OTFS) modulation is a promising solution to support integrated sensing and communication (ISAC). In this paper, a two-step parameter estimation method based on deep learning is proposed for target sensing in OTFS-ISAC. A pulse compression operation is performed to cancel the influence of multi-path effects and overlapping transmission symbols. Subsequently, a deep residual is designed to discover targets and estimating integer Doppler and delay. Finally, a deep neural network is employed to estimate fractional Doppler and delay. Simulations results show that proposed method is capable of effective target detection and accurate delay and Doppler estimation.