Compromise Node Detection Using Linear Regression Model in Wireless Sensor Networks

This paper introduces a method for detecting compromised nodes based on the characteristics of sensor nodes using a machine learning model in Wireless Sensor Networks (WSN). WSNs are emerging as a highly adaptable solution; however, numerous applications utilizing these networks involve sensitive information. Consequently, ensuring security is paramount in many of these scenarios. If a sensor node is breached, the overall security of the network can deteriorate rapidly unless appropriate countermeasures are implemented. Various strategies have been explored to address this challenge. This paper examines an anomaly-based intrusion detection system aimed at identifying compromised nodes within wireless sensor networks. We have used a Linear Regression Model (LRM) technique specifically for detecting these compromised sensor nodes, and simulations have been performed to validate the proposed design. Performance evaluation metrics, including accuracy, recall, and root mean square, are used to assess the detection system, demonstrating the model's effectiveness in identifying compromised nodes.