Mitigating DDoS Attacks via AI Detection and SDN Response

Abstract

Distributed Denial of Service (DDoS) attacks pose a serious ongoing threat to the stability and reliability of computer networks. These attacks have advanced significantly in scale and complexity over the past decades. Recently, massive DDoS assaults regularly reaching terabit levels have targeted prominent blockchain networks and internet services. However, real-time detection and mitigation of such sophisticated distributed threats remains a challenge. This research proposes an integrated artificial intelligence and software-defined networking (AI-SDN) framework to address this problem. The framework utilizes hybrid machine learning models for adaptive detection of attack behaviors and patterns. It leverages the programmability of SDN to dynamically route legitimate traffic away from attacks while rate-limiting suspect botnet sources. A feedback control loop enables swift coordination between detection and tailored mitigation responses. Evaluation through network simulations and emulations demonstrates the framework's effectiveness in achieving multi-layer visibility, early attack recognition, and containment of large-scale DDoS assaults before major disruptions occur.