DISTRIBUTED DENIAL OF SERVICE ATTACK MITIGATION USING REINFORCEMENT LEARNING

Abstract

Cybersecurity is threatened by Distributed Denial of Service (DDoS) attacks that destabilize network services by flooding systems with wrongful traffic. Unlike more conventional threat countermeasures, they fail to manage dynamic attack trajectories. In contrast, reinforcement learning provides a dynamic approach since systems improve their learning and response to the emerging threats in a real-time exercise. In this paper, reinforcement learning is used to study DDoS attack prevention and the study including the method, data set and measure used is discussed. Primary conclusions confirm strategies and algorithms achieve desirable detection accuracy, low false positive rates while being able to accommodate large networks. The outcomes of the paper demonstrate that reinforcement learning has a high potential for the further development of DDoS mitigation and improvement of security in networks.