Research Article
Centrality Based Algorithms for Controller Placements in Software Defined Wide Area Networks
@INPROCEEDINGS{10.1007/978-3-031-34896-9_1, author={Isaiah O. Adebayo and Matthew O. Adigun and Pragasen Mudali}, title={Centrality Based Algorithms for Controller Placements in Software Defined Wide Area Networks}, proceedings={Towards new e-Infrastructure and e-Services for Developing Countries. 14th EAI International Conference, AFRICOMM 2022, Zanzibar, Tanzania, December 5-7, 2022, Proceedings}, proceedings_a={AFRICOMM}, year={2023}, month={6}, keywords={Centrality Controller placements D-S theory of evidence Latency Propagating capability}, doi={10.1007/978-3-031-34896-9_1} }- Isaiah O. Adebayo
Matthew O. Adigun
Pragasen Mudali
Year: 2023
Centrality Based Algorithms for Controller Placements in Software Defined Wide Area Networks
AFRICOMM
Springer
DOI: 10.1007/978-3-031-34896-9_1
Abstract
Most controller placement algorithms require as input the estimated number of controllers to be placed in software defined wide area networks (SD-WANs). However, determining the correct number of controllers is NP-hard, as it requires selecting the set of nodes whose propagation capabilities exceed a given threshold. To this end, we propose in this study, a number of centrality based algorithms for estimating the capability of nodes to propagate end-to-end traffic flow. Specifically, we explore the Dempster-Shafer (D-S) Theory of evidence as a framework for estimating the capability of nodes by combining the properties of multiple centralities together to derive new joint properties. Nodes whose estimated capabilities exceed a given probability threshold are then selected as controllers. Based on the set of selected controller locations, we evaluate the performance of each joint evidential centrality (JEC) algorithm in terms of latency-related metrics. Experimental results show the superior performance of the combination of degree, node-strength and betweenness centralities in estimating the number of controllers required when worst and average case latencies are to be minimized.
