A Runtime DDoS Attack Detection Technique Based on Stochastic Mathematical Model

Distributed Denial of Service (DDoS) attacks are increasingly prevalent, targeting various entities. Detecting DDoS attacks is still an evolving and open challenge, despite considerable efforts. Existing solutions, including those employing artificial intelligence techniques, require significant computational resources and present limitations in handling realtime data. Hence, this paper presents a novel technique founded on a stochastic model to detect DDoS attacks during runtime. For evaluation, the technique focuses on SYN flood DDoS attack, and it has been implemented in a software-defined network given its programmability feature. Results have compared the proposed technique to representative ones from the literature, as Fuzzy Logic, MLP Neural Network, and Shannon Entropy. The new technique outperforms the other methods, opening up possibilities for application in different scenarios.