Towards WRSN Security: Probabilistic Approach for Malicious Node Identification

The advancement of Wireless Power Transfer (WPT) has led to the widespread adoption of Wireless Rechargeable Sensor Nodes (WRSN s) for distributed reporting, making these networks attractive targets for adversaries seeking to disrupt their functionality. Numerous solutions have been proposed, such as machine learning, cryptography, and verification protocols. However, these methods are typically designed for Wireless Sensor Networks (WSNs) and may not be suitable for resource-constrained rechargeable sensor nodes. In this paper, we focus on preventing malicious nodes from establishing legitimate connections with normal nodes in WRSN, which can severely disrupt the network. We propose a sensor node registration mechanism with multiple Cuckoo Filter-based verification and malicious node isolation. We also employ MinHashing, k-shingling, and Jaccard similarity techniques at the Mobile Charger for effectively detecting Replicated nodes and fake connections. We also perform a comprehensive performance analysis of our approach, demonstrating its efficiency in detecting and mitigating false nodes.