A DEEP Q-NETWORK (DQN) APPROACH FOR DYNAMIC FLOW ROUTING AND CONGESTION CONTROL IN SOFTWARE-DEFINED WIRELESS NETWORKS
DOI: https://doi.org/10.65725/JCISE/1/2/005
JOURNAL OF COMPUTATIONAL INTELLIGENCE SCIENCE AND ENGINEERING (JCISE)
ISSN: 3107-8168
Volume 1 Issue 2, October 2025

Abstract
The convergence of Software-Defined Networking (SDN) and wireless technologies has given rise to Software-Defined Wireless Networks (SDWN), offering unprecedented programmability and centralized control. However, the dynamic and unpredictable nature of wireless environments, characterized by fluctuating link quality and bursty traffic, poses significant challenges for efficient traffic management. Traditional routing protocols like OSPF, which rely on static metrics, are inadequate for these conditions. This paper proposes a novel Deep Reinforcement Learning (DRL) framework integrated within the SDWN controller to achieve adaptive and intelligent traffic management. We formulate the joint problem of flow routing and congestion control as a Markov Decision Process (MDP). A Deep Q-Network (DQN) agent is trained to learn an optimal policy that dynamically assigns paths to data flows based on real-time network state, including link utilization, delay, and packet loss. Simulation results using the Mininet-WiFi platform and a custom traffic generator demonstrate that our DQN based agent significantly outperforms traditional shortest path and load-balancing algorithms. The proposed model reduces average end-to-end delay by 32% and packet loss by 45% under high load conditions, while improving network throughput by 28%, proving its efficacy in enhancing Quality of Service (QoS) in dynamic wireless environments. 

Authors: A.Afrin Safna, A.Anbarasi , and A.Gobinath

Keywords: Software-Defined Wireless Networks; Deep Reinforcement Learning; Deep QNetwork; Traffic Engineering; Congestion Control; Quality of Service; Markov Decision Process