Channel State Information Prediction for Underwater Acoustic Multiple-Input Multiple-Output Communications

Multiple Input Multiple Output (MIMO) transmitters (Tx) leveraging Channel State Information (CSI) can substantially enhance communication quality, such as increasing channel capacity and Signal-to-Noise Ratio (SNR). However, in Underwater Acoustic (UWA) communications, the extremely low underwater sound speed and the continuously changing sea waves render the conventional CSI acquisition methods. Therefore, the UWA MIMO system only utilize CSI at Receiver (Rx). By analyzing the affect factors of UWA CSI, a model-data dual driven channel prediction framework is proposed. First, the problem and key technologies (historical CSI acquisition scheme, spatial and temporal alignments) of the proposed framework are formulated. Second, a realistic historical CSI acquisition scheme, model-driven spatial and data-driven temporal alignments are proposed. Finally, numerical simulations show that the proposed CSI prediction framework can effectively meet the CSI quality requirements of Tx when the ratio of CSI error variance and CSI variance is lower than 0.4.