Explainable AI-Driven Intrusion Detection and Blockchain-Based Trust Evolution for Secure Routing in Underwater Sensor Networks

Abstract

Acoustic communication constraints, high latency, and dynamic topology of underwater wireless sensor networks (UWSNs) make them highly susceptible to routing disruptions, Sybil behaviour, selective forwarding, and false data injection, making them highly vulnerable to routing disruption. Currently implemented cryptographic and learning-based defenses lack transparency and cannot justify their decisions due to the lack of centralized trust authorities or black-box intrusion detection systems. Consequently, they are unable to be accepted in the field. A novel protocol, Blockchain-empowered eXplainable AI-based Trust and Intrusion-handling (BEXAIT-Trust), is proposed to perform decentralized trust management, explainable intrusion detection, and attack-aware routing in UWSNs. A BEXAIT-Trust system differs from previous methods that sequentially combined blockchain and Intrusion Detection System (IDS). The system has three main parts: (i) a two-stage explainable artificial intelligence intrusion detection system that uses both supervised attack classification and unsupervised anomaly detection; (ii) a blockchain-integrated trust evolution mechanism that uses lightweight smart contracts to update trust between nodes.; and (iii) a trust-and-risk-adaptive routing framework that informs traffic routing decisions based on articulated risk rather than raw intrusion detection system labels. This methodology does not exist in current UWSN security frameworks, which typically lack either co-optimized trust–IDS coupling or decision explainability. Based on extensive simulations, it has been demonstrated that detection accuracy is 94–99%, false-positive rate is 1–6%, and precision/recall is 92–98%. To accomplish security improvements, mere latency increases of 5–12% are required, energy overhead increases of 6–15%, and throughput increases of 8–20%. With XAI, mission operators can interpret and trace forensic evidence based on blockchain-backed evidence to 90–97% accuracy. This study confirms that BEXAIT-Trust strengthens the security, transparency, and resilience of UWSNs, providing a deployable path toward accountable underwater cyber-physical systems.