An Evolutionary-Based Approach for Low-Complexity Intrusion Detection in Wireless Sensor Networks

Abstract

As security threats are increasingly diversified, a critical problem in Wireless Sensor Network environments (WSNs) is detecting anomalies. WSNs are affected by several limitations, such as limited energy, insufficient memory, weak computation power, and short communication range. Hence, it is necessary to improve the detection accuracy and convergence speed of intrusion detection in such environments. In this article, we propose an intrusion detection model based on Time-Varying Parameter Improved Particle Swarm Optimization (TVP-IPSO) with Principal Component Analysis (PCA) and Support Vector Machine (SVM). The PCA is applied aimed at the data dimension reduction by compressing the data to reduce energy consumption, and an intrusion detection algorithm based on SVM is considered to ensure high detection accuracy. To optimize the SVM algorithm and identify its optimal parameters, the TVP-IPSO is used to improve the intrusion detection algorithm's detection precision and convergence speed. Experimental results show that the detection accuracy of TVP-IPSO-SVM is higher than GA-SVM and IPSO-SVM, demonstrating that the proposed research has better adaptability, higher detection accuracy, and faster convergence speed when compared to other works presented.