Evaluation of a Deep Learning-based Orthogonal Frequency Division Multiplexing (OFDM) Scheme for Undersea RF Communication

• Published in: 2024 IEEE Asia-Pacific Conference on Applied Electromagnetics, APACE 2024
• Main Author: 2-s2.0-85219752966
• Format: Conference paper
• Language: English
• Published: Institute of Electrical and Electronics Engineers Inc. 2024
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85219752966&doi=10.1109%2fAPACE62360.2024.10877407&partnerID=40&md5=1a3a48398794bbc7e785e9f76910d6ca

Undersea RF communication suffers from poor signal-to-noise ratio due to high signal attenuation, and various noises from the propagation channel and devices operating undersea. This paper presents a deep learning (DL)-based orthogonal frequency division multiplexing (OFDM) scheme in an undersea RF communication system to combat these issues. The proposed OFDM transmission scheme utilizes a long short-term memory (LSTM) network at the receiver to replace conventional channel estimation and equalization. The LSTM network is trained to model simplified undersea channels emulating both deep sea and shallow sea conditions in short distance RF communication. It was found that the proposed DL-based method produced improved bit-error-rate (BER) against Eb/No than conventional method in both AWGN and Rician channel, approximately 1 to 2 dB. © 2024 IEEE.