An Efficient Feature Selection-Based Approach for Intrusion Detection in IoT Environment

With the current scenario Internet of Things (IoT) is not only a luxury but seems essential to improving human lives. Security and privacy have always been a critical concern for any kind of computer communications and networks. IoT-based systems can be exploited by its inherent security vulnerabilities. One vital security mechanism is Intrusion Detection System (IDS). In this study, we explore the possibilities of developing a solution for the implementation of attack detection mechanism using machine learning algorithms with minimal feature selection techniques. In the proposed system, we incorporate three popular feature selection methods, namely, ANOVA F-test, mutual information, and Spearman's correlation coefficient to achieve better performance with reduced features. The proposed system obtains comparatively better results than existing systems (unnecessary features affect the performance of the system adversely).