SDN-based Solutions for Improving Network Performance

Abstract

The integration of Software-Defined Networking (SDN) into the Internet of Things (IoT) offers a promising solution to address the growing challenges of scalability, complexity, and flexibility in modern networks. owever, this convergence introduces significant management challenges, particularly regarding performance, latency reduction, and resilience. A critical issue is the Controller Placement Problem (CPP), which directly impacts the efficiency of Software-Defined Internet of Things (SD-IoT) networks and becomes more complex in erogeneous environments with dynamic traffic and load variations, especially in mission-critical applications like healthcare and autonomous transport. This thesis proposes a set of SDN-based approaches to improve Software-Defined IoT (SD-IoT) network performance by addressing the Controller Placement Problem (CPP). First, the rogeneous Traffic Flow–based Controller Placement (HTF-CP) method optimizes controller locations by accounting for traffic diversity under flexible deployment assumptions. Second, a population-aware SD-IoT architecture for the Algerian network is introduced, combining K-means clustering with enhanced K-center and K-median algorithms to incorporate population-weighted traffic demands. Third, a Weighted Betweenness Centrality–based approach (WBC-CPP) identifies influential nodes to enable efficient and load-aware controller placement with reduced search space. This approach is further enhanced through the integration of Grey Wolf Optimization (GWO), resulting in the GWO-WBC-CPP method, which considers the dynamic nature of IoT traffic over time. Finally, the Particle Swarm Optimization with Heterogeneous Switch Load (PSO-HSL) method applies a meta-heuristic strategy to jointly optimize latency and reliability while avoiding exhaustive search. The effectiveness of these approaches is validated through extensive simulations on realistic network opologies (Internet Topology Zoo and OS3E), a newly designed population-aware Algerian network, and synthetic topologies for scalability analysis. The results show notable performance gains in terms of latency, load balancing, reliability, and overall network efficiency when compared to conventional CPP and baseline SDN placement methods