Data Mining Techniques used for Evaluation an Efficient DDoS Attack Detection System: A Deep-Learning Model

Abstract

Developed recently to solve the shortcomings of traditional networks, software-defined networking is a new paradigm (SDN). Decoupling the control plane from the data plane, which is the SDN's core property, makes network management simpler and promotes effective programmability. The new design, on the other hand, is vulnerable to a number of attacks that could exhaust the system's resources and prohibit the SDN controller from offering services to authorized users. One of these threats, the Distributed Denial of Service (DDoS) assault is one that is gaining popularity. A DDoS assault has a severe negative effect on emphasize servers have no ability to access their network facilities as a result. accommodate legitimate users. We introduce DDoS Net, a DDoS attacks system for intrusion detection for SDN systems. Our approach is based on data mining software (Rapdiminer) and neural networks working along with auto encoders. As a result, we have a lot of faith in protecting these networks thanks to our methods. We assess the CICIDS 2019 dataset, which includes network behavior associated with DDoS threat and benign network activity, is publicly available, and satisfies certain requirements. It evaluates a variety of data mining algorithms' efficacy as well as internet usage characteristics to determine the best attributes for spotting the most typical assault kinds.