Reinforcement Learning-based Adaptive Routing in Software Defined Networks under Security Threats

Software Defined Networking (SDN) introduces a flexible and centralized approach to network management by decoupling the control plane from the data plane, thereby enabling enhanced programmability and efficient traffic handling. However, this architectural centralization also introduces critical security vulnerabilities, especially within the control plane, rendering SDN environments highly susceptible to attacks such as Denial-of-Service (DoS) and blackhole routing. Traditional static routing methods prove insufficient in mitigating such dynamically evolving threats. To address these limitations, this paper proposes a Reinforcement Learning (RL)-based adaptive routing framework designed to enable the SDN controller to autonomously learn optimal and secure routing policies through continuous interaction with the network environment. The proposed framework is implemented using the Mininet emulator and the Ryu SDN controller. It dynamically adjusts routing paths in realtime to avoid compromised network nodes and strengthen the network's security posture. Experimental evaluations conducted under various simulated attack scenarios demonstrate that the RL-based approach significantly outperforms conventional static routing techniques in terms of adaptability, network performance, and resilience. The results validate the effectiveness of our proposed framework in establishing a robust foundation for intelligent, adaptive, and security-aware SDN infrastructures.