Semantic Communication in Underwater IoT Networks for Meaning-Driven Connectivity

The Internet of Underwater Things (IoUT) is revolutionizing marine sensing and environmental monitoring, as well as subaquatic exploration, which are enabled by interconnected and intelligent subsystems. Nevertheless, underwater communication is constrained by narrow bandwidth, high latency, and strict energy constraints, which are the source of efficiency problems in traditional data-centric networks. To tackle these problematic issues, this work provides a survey of recent advances in Semantic Communication (SC) for IoUT, a novel communication paradigm that seeks to harness not raw symbol information but rather its meaning and/or contextual significance. In this paper, we investigate the emerging advanced AI-powered frameworks, including large language models (LLMs), diffusion-based generative encoders, and federated learning (FL), that bridge semantic compression with context-aware prioritization and robust information reconstruction over noisy underwater channels. Hybrid acoustic-optical-RF architectures and edge-intelligent semantic encoders are also considered enablers of sustainable, adaptive operations. Examples in underwater archaeology, marine ecology, and autonomous underwater vehicles (AUVs) coordination are provided as a relief to illustrate the merits of meaning-driven connectivity. The paper concludes with some recommendations, including semantic representations standardization, cross-domain interpolation, and privacy-support schemes. These issues must be addressed in the future before trustworthy SC-enabled IoUT systems can be developed for underwater communication.