T-SAPR: An Efficient Q-Learning Trust-based Secure Routing Protocol for Underwater Acoustic Sensor Networks

Underwater acoustic sensor networks (UASNs) play pivotal roles in diverse civilian and military contexts. Yet, due to their broadcasting nature and the challenging environments they operate in, they are susceptible to a multitude of security vulnerabilities. Additionally, few existing routing protocols in UASNs account for both security and underwater transmission challenges. In response, we introduce T-SAPR, a secure Q-Learning-based routing protocol bolstered by trust management and AUV-driven path restoration. T-SAPR utilizes an attention-based Long Short-Term Memory (LSTM) to construct a multifaceted trust model, encompassing node trust, communication trust, and environmental trust. Further, energy, packet delivery ratio, and latency collectively inform optimal routing strategies via Q-Learning's reward function. Additionally, an AUV-assisted repair mechanism to enhance UASNs' reliability is proposed, particularly in scenarios involving multiple sensor node failures or the detection of malicious nodes. Evaluation results demonstrate T-SAPR's prowess in identifying malicious nodes.