MHSA-CNN-Bi-LSTM Based Interference Cancellation Algorithm for MIMO-FTN-OWC System

In MIMO-FTN-OWC systems, inter-symbol interference (ISI) caused by the fast-than-Nyquist (FTN) technique is a significant issue. To address this problem, we propose an ISI cancellation algorithm a hybrid network consisting of MHSA-CNN-Bi-LSTM. This algorithm effectively eliminates ISI by leveraging the feature extraction capability of CNN, the bi-directional sequence information capture capability of Bi-LSTM, and the multi-head self-attention (MHSA) mechanism. Simulation results demonstrate that our proposal achieves improved bit error rate (BER) performance compared to traditional algorithms without deep learning-based ISI cancellation, with a maximum improvement of 7 dB under various conditions. Furthermore, when operating within the Mazo limit for acceleration factor, our algorithm successfully eliminates the FTN caused ISI and achieves BER performance comparable to that of orthogonal systems.