Design and Experimental Validation of Semantic-Aware OFDM for Sonar Image Transmission over Underwater Acoustic Channels

Underwater sonar image transmission is vital for marine exploration, environmental monitoring, and underwater robotics. However, it faces severe challenges due to the limited bandwidth and multipath fading of underwater acoustic (UWA) channels [1]. Traditional image compression combined with UWA communication often results in artifacts and loss of detail, primarily due to the complexity of UWA channels [2]. Recently, with the rapid development of artificial intelligence (AI), semantic communication has emerged as a promising alternative to extract essential semantic features and offer enhanced robustness in complex UWA environments [3, 4]. In this work, we design a semantic-aware orthogonal frequency-division multiplexing (OFDM) system that incorporates joint source channel coding (JSCC) for sonar image transmission. Laboratory experiments over UWA channels are conducted to evaluate its performance. Experiments show that the proposed semantic-aware OFDM system outperforms conventional approaches, ensuring reliable sonar image transmission in UWA channels even under low-SNR conditions.