A Lightweight DoS and DDoS Attack Detection Mechanism-Based on Deep Learning

Abstract

Denial of Service (DOS) attacks are one of the major attacks on any network and a potential threat to internet resources and services. This threat amplifies with Distributed Denial of Service (DDoS) attacks as these attacks do not give any alert or time for the victim to act. With the increase in devices connected to the internet, the intensity and frequency of attacks are also growing daily. Existing solutions like intrusion detection systems(IDS) cannot provide better results with the complexity of DDoS attacks because their filtering criteria have been static to distinguish attack traffic from regular traffic. So, with the robustness of attacks, the solutions need to be robust enough. The proposed method here is to use the state-of-art CNN model, i.e. EfficientNet and ResNet. EfficientNet model being a lightweight model, can be integrated with any device to avoid DDoS attacks. Since the prediction time is minimal, the proposed method can pinpoint the attack to act immediately.