Modulation and coding scheme mode selection based on machine learning in underwater acoustic channels

This paper proposes a machine learning-based optimal modulation and coding scheme (MCS) mode selection in underwater acoustic communication (UAC) channels. MCS is an efficient method for improving the system efficiency by changing transmission parameters according to channel conditions in UAC channels. As a UAC channel has time-varying characteristics such as intersymbol interference and Doppler frequency shift, it is impossible for a UAC system to overcome with a large variety of communication impairments well by only using fixed MCS. Thus, this paper proposes a machine learning-based approach for optimal MCS mode selection by utilizing various channel state information (CSI) such as received signal-to-noise ratio, Doppler effects, and multipath propagation. CSI datasets of various UAC channels are generated through simulation and are subsequently used to train machine learning models to predict the optimal MCS mode for varying channel conditions. In time-varying underwater channels, we proved the proposed machine learning-based MCS selection method is better than conventional fixed table-based methods in aspect to transmission efficiency and reliability. [This research is supported by a KRIT grant funded by the Korea government (DAPA) (KRIT-CT-23-035-01, Multi AUV operation Technology for Mine Detection ('23–'28))]