QoSentry: A Reinforcement Learning Framework for QoS-Preserving DDoS Mitigation in Software-Defined Networks

Abstract

In the realm of telecommunications, the advent of 5 G technology brings unparalleled speed and efficiency. Leveraging Software-Defined Networking (SDN), 5 G networks are able to dynamically allocate resources and optimize performance. However, the rise of Internet of Things (IoT) devices has expanded the threat landscape due to the increased number of connected endpoints, providing more targets for potential exploitation. Traditional security measures struggle to defend against these evolving threats, necessitating the adoption of more sophisticated and agile approaches. Artificial Intelligence (AI) and Deep Reinforcement Learning (DRL) techniques offer rapid and adaptive responses as they can learn from and adapt to evolving threats in real-time, surpassing traditional security methods. In this article, we propose a DRL-based approach with the goal of mitigating the malicious impact of DDoS attacks within an SDN framework. Our proposed model leverages the DDQN algorithm to preserve accessibility and performance for legitimate users during different attack scenarios. The experimental setup emulates real-world user behaviors to simulate practical network conditions. Our mitigation strategy employs adaptable countermeasures based on the current network state, thereby ensuring flexible and effective responses to varying threat levels.