ENERGY-EFFICIENT ROUTING AND ANOMALY DETECTION IN UNDERWATER WIRELESS SENSOR NETWORKS USING RANDOM FOREST-BASED OPTIMIZATION

The open deployment of underwater sensor networks (UWSNs), the restricted capabilities of its nodes, and the hostile environment of the underwater environment make them very susceptible to network assaults. Even though intrusion detection systems may assist in mitigating these risks, the approach that is now being deployed is ineffective, does not detect certain kinds of assaults, and is generally underutilized. Hence, this paper proposed an Adaptive Random forest-based dynamic trust evaluation model (ARF-DTEM) for underwater sensor networks for multiple types of attacks. With our method, cluster head nodes can reduce the amount of processing by using rough set theory to extract vital information from neighboring components. The data that has been processed is used by nodes at lower layers to improve computation. By creating a more representative sample using the Synthetic Minority Oversampling Method (SMOTE) to improve the proposed ability to identify attacks on minority groups. The proposed method makes predictions about the trust states of the sensor nodes and uses deep reinforcement learning to improve the trust updating process. Both of these techniques lead to an increase in the accuracy of the detection process. By reducing the number of false positives and achieving a high degree of trust assessment accuracy, the simulation's findings demonstrate that ARF-DTEM can efficiently identify malicious nodes.