ذكي الوزن الخفيف تطفل منع نظام ل إنترنت من الأشياء

Abstract

As the number of Internet of Things (IoT) users, services, and applications rises, so does the demand for more security attention. IoT networks have limited power efficiency, necessitating the need of a solution with a limited amount of computing operations. Building a smart lightweight Intrusion Prevention system (IPS) with low False Positive Rate (FPR), accurate detection and without human involvement is necessary for critical and real-time applications -as standard IPSs cannot provide these criteria-. Further more, IoT environments are made up of a diverse set of IoT components from various suppliers and based on various IoT platforms. As a result, interoperability issues prevent the wide use of IoT technology. In this thesis, we present an overview that is focused on IoT security, Machine Learning algorithms mostly used to develop Intrusion detection systems, and software defined networks (SDN) technology. We deeply reviewed some of the most recently created IDS models for IoT, which employed Machine Learning (ML) or Deep Learning (DL) approaches. We gave a comprehensive review of IoT manage ment, including an overview of conventional management platforms and protocols. We used the Extreme Leaning Machines algorithm to develop an autonomous and efficient intrusion detection architecture. We trained and validate our model using SDN and IoTID20 Datasets after an extensive data preprocessing. The results shows that our model can perform well in terms of time and accuracy. We evaluated the performance of three online Ensemble-based Learning models. As the SDN technology is an excellent solution for managing heterogeneity as well as providing monitoring and real-time intru sion prevention, it has been used in this work. We simulated our model using Mininet and Ryu. The experimental findings indicated that the suggested architecture for an intrusion detection system based on Extreme learning machine algorithm can identify DDOS Attacks and improve the security of the IoT environment.