Machine Learning-Based Performance Analysis for IoT Attacks in IDS

Abstract

Internet of Things(IoT) security is one of the main issues when executing and creating IoT platforms. The significant increment of the IoT devices in smart homes and other smart infrastructure make numerous attacks on these devices. With new and interesting attacks equipped for trading off the IoT platforms, intrusion detection and forensic systems really should be created. Security systems, for example, cryptography and validation are difficult to apply obliged IoT devices and organizations. With predominant innovations like the IoT, Cloud Computing, and Social Networking, a lot of organization traffic and information are created. Subsequently, there is a requirement for Intrusion Detection Systems that screen the organization and break down the incoming traffic powerfully. IDS assumes a significant function as a high-security answer for intrusion detection in IoT networks. Building an efficient network-based IDS, feature selection is an essential step as irrelevant and redundant features may adversely affect the classification performance of the system. The proposed system implements using Weka Tool based on BoT-IoT Benchmark dataset. The aims of the system to detect attack utilizing three different machine learning algorithms such as J48, Hoeffding Tree, and Naïve Bayes (NB). The solution for this problem may be provided by calculating precision, recall and, F1-score based on confusion matrix.