Back to Search Start Over

Multi-criteria decision-making for controller placement in software-defined wide-area networks.

Authors :
Abdi Seyedkolaei, Ali
Hosseini Seno, Seyed Amin
Source :
Journal of Supercomputing. Nov2021, Vol. 77 Issue 11, p13447-13473. 27p.
Publication Year :
2021

Abstract

Software-defined networks have many benefits such as more control over the control plane and reduced operating costs through separating the control plane from the data plane in network equipment. One of the most critical problems in software-defined networks is a controller placement problem, which significantly influences its overall performance. The purpose of this problem is to determine the number of controllers required and how to assign switches. This paper attempts to solve this problem, aiming to reduce the network's operational cost and to improve their survivability and load balancing. Hence, we have tried to divide the network into several subdomains using segmentation. Then, we used multi-criteria decision-making methods to solve the controller placement in each subdomain. For this purpose, we considered criteria such as reliability rate, cost, delay, and processing capacity. To assign switches to controllers, we used the proposed mathematical model to minimize the objective function while observing the defined constraints. Furthermore, according to the proposed architecture, we used distance to the central controller, adjacent neighborhood, and controller memory to select the cluster head controller for each subdomain to communicate with the central controller. Finally, we performed experiments on topologies Uran, SwitchL3, and Sinet of the Internet Topology Zoo to evaluate the proposed method. In these experiments, we compared the results of the proposed method with our work-related methods, including Controller Placement Analytic Hierarchy Process (CPAHP) and Controller Placement Modified Density Peak (CPMDP), in terms of criteria such as cost, survivability, number of controllers, connection failure probability, average delay, and controller load-balancing rate. The results show that the proposed method outperforms CPMDP and CPAHP. Thus, the proposed method has a 24.97% and 19.76% improvement in reducing network implementation costs than CPMDP and CPAHP, respectively. [ABSTRACT FROM AUTHOR]

Details

Language :
English
ISSN :
09208542
Volume :
77
Issue :
11
Database :
Academic Search Index
Journal :
Journal of Supercomputing
Publication Type :
Academic Journal
Accession number :
153206747
Full Text :
https://doi.org/10.1007/s11227-021-03815-3