Bi-LSTM-Based Signal Detection Method for Underwater Acoustic OTFS Communication System

The orthogonal time frequency space (OTFS) modulation is a new multi-carrier modulation technology proposed in recent years. The sparse and invariant signals in the Delay-Doppler domain make OTFS to be a promising technology to solve the Doppler shift in the time-variant underwater acoustic (UWA) channel. In the UWA-OTFS system, the traditional channel estimation and equalization methods need accurate channel information, and the performance can be easily influenced by the channel model. This paper proposes a Bi-LSTM-based UWA-OTFS signal detection method to replace the traditional explicit algorithms. This method can fully use the previous channel information and data to improve the reliability of the UWA-OTFS system, which has been verified on the Watermark underwater acoustic dataset. The result shows that the proposed method performs with a lower symbol error rate (SER) than traditional algorithms.