A Framework of Novel Feature Set Extraction based Intrusion Detection System for Internet of Things using Hybrid Machine Learning Algorithms

Smart objects associated with the Internet, constitutes the concept of Internet of Things (IoT), are reforming people's collaboration with the world. As innovation comes to all around, anybody can abuse it, and it is constantly fundamental to secure it. Various intrusion detection approaches have been established with their individual merits and demerits. The greater part of study in the zone of IoT intrusion detection is quiet built on either benchmarked KDD cup dataset or simulated datasets due to non-accessibility of real datasets. For this reason, there is a necessity to create test data set, which depends on real network data. The proposed work builds a novel data set from the wireless network packet traffic flow captured through Wireshark that holds different attack profiles. The captured traffic is filtered and preprocessed to generate a featured dataset. The proposed Intrusion Detection System (IDS) uses hybrid machine learning technique by combining K-means clustering and Support Vector Machine (SVM) classification to achieve low false positive rate with improved detection rate.