Internet of Things-Based Middleware Against Cyber-Attacks on Smart Homes using Software-Defined Networking and Deep Learning
Mohammed Tawfik, Nasser M. Al-Zidi, Belal Alsellami, Aymen M. Al-Hejri, Sunil Nimbhore
Abstract
Internet of Things (IoT) devices are expected to number about 3.5 billion by 2023.; a tremendous amount of Internet of Things (IoT) data that is generating (IoT) devices is estimated to exceed 79.4 zettabytes by 2025. Security challenges will become an increasingly significant issue, especially in smart homes that depend entirely on IoT devices. Due to the weak infrastructure of the IoT, it is vulnerable to various types of cyber-attacks. The most common IoT attacks are distributed denial of service (DDoS). Most traditional security solutions, like intrusion detection systems (IDS), cannot detect most attacks. Complexity is being hidden by a new paradigm that recently arose, called the software-defined system that is brings a significant change to the networking industry, a great solution for mitigation of attacks that can adopt deep learning technique to encounter cyber-attacks based on the attack behavior and by filtering normal and attack traffic by using well-defined rules. This paper proposed a system by suggesting middleware that can help mitigate or prevent various attacks on IoT on smart home environment. Machine learning has included in the middleware to provide automatic protection against cyber-attacks on IoT networks. A promising approach to protecting real-time, highly accurate attacks on SDN-managed IoT networks has been proposed. This middleware allows IoT devices to efficiently handle evolving security threats dynamically and adaptively without impacting the IoT devices.