Deep Deterministic Policy Gradient-Based Load Balancing Method in SDN Environments

Abstract

Software-Defined Networking (SDN) is a transfor-mative technology that separates the control and data planes, facilitating network control and management through centralized or distributed SDN controllers. Recent advancements in machine learning, specifically Deep Reinforcement Learning (DRL), drive research towards integrating these techniques into SDN for intelligent network management. Effective load balancing in networks is a critical aspect of network management, aiming to enhance Quality of Service (QoS). While previous research has explored load balancing for SDN using DRL, they persist in small-scale topologies, and challenges remain in adapting to dynamic changes in network traffic on links in large-scale topolo-gies. This paper proposes a novel approach to load balancing in large-scale SDN networks utilizing Deep Deterministic Policy Gradient (DDPG). Additionally, the paper proposes a method to extract optimal parameters, enhancing the overall effectiveness of network management.