Identifying False Injection Packet Transmission using SVM and Secure OLSR in Wireless Sensor Networks

Wireless Sensor Networks (WSNs) are systems of wirelessly communicating, geographically scattered sensor nodes that collect, process, and transmit data. Examples of applications for these networks are smart cities, industrial automation, healthcare, military, and environmental monitoring. Several research studies have been conducted on securing WSNs against false injection attacks, anomaly detection, and optimizing routing protocols. If a legitimate packet is accidentally discarded, vital information can be lost, and applications that rely on data accuracy can be compromised. To resolve the issue, this paper proposes an integrated security framework that combines Support Vector Machines (SVMs) for anomaly detection and Optimal LinkState Routing (OLSR) to achieve secure and efficient routing to avoid false injection in the WSN environment. Before detecting the anomaly, a Fuzzy clustering method is used to create the cluster and also select the appropriate cluster. After selecting the suitable cluster, the SVM model is trained to detect spuriously injected packets and anomaly nodes. Simultaneously, OLSR ensures secure and energy-efficient routing through cryptographic authentication and trusted path selection. The proposed model outcomes demonstrate that the suggested method accomplishes a Packet Delivery Rate (PDR) of 92% and an accuracy of 95% in detecting false injection attempts. In addition, energy consumption is reduced by 14% compared to traditional routing methods, significantly extending the network lifetime. The results show that integrating SVM and OLSR can effectively suppress false packet injection and improve WSN performance.