K-Means++ Clustering-Based Approach for SDN Controller Placement in LEO Satellite Networks

Abstract

Software-Defined Networking (SDN) offers a solution for the efficient management of Low Earth Orbit (LEO) satellite networks through centralized control and programmability. However, the dynamic nature of LEO satellites has a unique requirement for balanced controller placement, considering low-latency communication and balanced load distribution. Therefore, a strategic, low-latency controller placement is essential to ensure optimal performance and scalability in LEO satellite networks. The proposed study presents a clustering-based technique, K-Mean++, for placing SDN controllers in the satellite network, aiming to minimize communication delay between the controller and the Open Virtual Switch (OVS) satellites. The satellites are divided into clusters, and a candidate satellite (cluster-head) is selected for controller placement. The proposed technique works in collaboration with the K-Means clustering algorithm. The performance is evaluated on switch-to-controller delay, controller-to-controller delay, controller load coefficient of variance, and traffic distribution efficiency. The results show that the proposed technique outperforms its counterparts with a 59% reduction in worst-case switch-to-controller delay, a 5% reduction in control domain delay, a 26% improvement in controller load distribution, and a 5% enhancement in control domain traffic distribution, making it an optimal solution for SDN controller placement in LEO satellite networks.