Semi-supervised deep-ELM for DDoS attack detection and mitigation using the OptimalLink model in IoT networks

Abstract

Human-machine interaction is becoming smarter because of an emerging technology called the Internet of Things. Internet of Things devices are made by different manufacturers, which may lead to a lack of security standards. The attackers use this lack, such as unpatched vulnerabilities, to form botnets by simply hacking the Internet of Things devices. Of the several security breaches, distributed denial of service attacks are quite tricky, dismember the network, and offer end consumers a variety of services. For instance, sapping bandwidth, depleting server resources, and ruining the end-user experience. As a result, in many Internet of Things use cases, distributed denial of service might create the possibility of catastrophe. This paper explores encountering the distributed denial of service attack mooting by malicious Internet of Things systems. To detect and prevent distributed denial of service attacks, our security strategy modifies the software-defined network paradigm. To identify and counteract distributed denial of service attacks, we have suggested a unique semi-supervised deep-extreme learning machine-learning technique for detection with unique dataset features and a unique optimal link mitigation algorithm. These detection and mitigation methods are incorporated into the software-defined network controller, which is located in the internet of things and application layers. Compared with other solutions results, our detection and mitigation strategy increases the throughput and bandwidth level and decreases the network load. We tested the semi-supervised deep extreme learning machine algorithm using an emulated topology and testbed, and then we compared the outcomes to cutting-edge solutions. We improved the accuracy rate for distributed denial of service attack detection to 99.97 %.