Link-Awareness-Based Routing Protocol for Underwater Wireless Sensor Networks

Abstract For underwater wireless sensor networks (UWSNs), constructing reliable and high-performance multi-hop routing paths among submerged sensor nodes continues to be a fundamental problem that significantly impacts the overall communication capability of the network. Given the constrained energy budgets of sensor nodes and the challenging propagation conditions, poorly selected forwarding routes and unexpected link failures during packet transmission may result in excessive delays and degraded delivery performance. This paper proposes a Link-Awareness-Based Routing Protocol (LARP) for UWSNs. In particular, the protocol leverages a Q-learning framework to enhance link-level transmission quality within UWSNs. The next-hop selection procedure is cast as a reinforcement learning task, where every network node operates as an independent learning agent. A composite reward function accounting for remaining energy, forwarding progress, and link quality is developed, with the best route determined via iterative Q-value updates. Extensive simulation experiments confirm that, relative to conventional routing schemes, the proposed LARP protocol yields notable gains in packet delivery ratio, transmission delay, and overall energy utilization.