Traffic engineering is critical for optimizing network performance and resource utilization. Software-Defined Networking (SDN) offers a flexible architecture that allows dynamic traffic management. However, traditional traffic engineering methods can struggle as SDN networks grow in size and complexity, making congestion management a key challenge. This paper presents a novel traffic engineering solution for SDN that utilizes Particle Swarm Optimization (PSO) within a scalable hierarchical framework. The approach addresses the difficulties of dynamic traffic patterns and resource allocation in large-scale SDN deployments by breaking down traffic management tasks into manageable domains. By leveraging PSO’s ability to explore large solution spaces, the algorithm optimizes routing paths and bandwidth allocation based on a fitness function designed to minimize congestion, latency, and other key performance metrics. The proposed method, called Particle Swarm Optimization for Congestion Management (PSOCB), focuses on optimizing cluster switches using meta-heuristic techniques. PSOCB achieves a 13% increase in throughput and an 8% improvement in link performance, while reducing delays compared to traditional methods. Although PSOCB occasionally showed higher congestion than baseline approaches, it quickly restored normal network conditions through its congestion removal strategies. Simulations on realistic network topologies demonstrate substantial performance improvements over conventional static routing, highlighting the effectiveness and scalability of our PSO-based hierarchical traffic engineering solution.
