A Naive Bayes-Driven Mechanism for Mitigating Packet-Dropping Attacks in Autonomous Wireless Networks

Autonomous wireless networks, characterized by peer-to-peer connectivity and dynamic topology, enable efficient internet access, irrespective of geographical constraints. Their applications span across disaster relief, military operations, road safety, and healthcare, areas where secure communication is indispensable. This paper addresses the crucial challenge of packet-dropping attacks in these networks, a security issue that has not yet been thoroughly explored in current literature. Conventional mechanisms for preventing packet-dropping often fail to differentiate between malicious attacks and system faults, underscoring the need for an effective classification system. Such a system should discern whether packet-dropping incidents are a result of malevolent attacks or system faults, and appropriately penalize only the malicious nodes. In this light, we introduce an Intrusion Detection System (IDS) based on the Naive Bayes algorithm to mitigate packet-dropping attacks in autonomous wireless networks. This algorithm has demonstrated efficacy in distinguishing between different categories based on statistical probabilities. By successfully mitigating both malicious attacks and system faults, our proposed IDS significantly enhances network performance. Simulation results confirm the effectiveness of the proposed IDS, showing notable improvements in packet delivery, delay, and energy efficiency. This IDS, therefore, not only detects and eliminates packet-dropping nodes that disrupt network operations but also extends the overall network performance.