An Efficient Controller Placement Algorithm in Software Defined Networks using Spectral Clustering
Mayuri Kalmat, Prerana Shetti, Fiza Khan, Sumedha Narayan Shinde, D. G. Narayan · 2023
Software Defined Networks (SDN) represents the next-generation network technology that enables efficient network management. The core concept of SDN revolves around the separation of the data plane and control plane. Unlike traditional TCP/IP network architectures, SDN employs a centralized controller that oversees multiple forwarding devices. This centralized management distinguishes SDN from conventional approaches. SDN employs a decoupled architecture, ensuring the separation of the control and data planes. In SDNs, the placement of controllers plays a crucial role in network performance. Therefore, it becomes essential to define clusters based on optimal algorithms that minimize network delay and latency. To achieve this, the network is divided into clusters using spectral clustering. The Elbow Method is utilized to determine the optimal number of clusters. Subsequently, the centroid-based controller placement algorithm is employed to position the controllers within these clusters. The Mininet emulator with the POX controller is used for implementation. Finally, the evaluation of various Quality of Service (QoS) parameters, including latency, throughput, and jitter, is conducted using the Internet Topology Zoo (ITZ) dataset.