Análise de técnicas de otimização multiobjetivo para o posicionamento de controladores em redes SDN

Ana Carolina de Oliveira Christófaro, Marcelo Menezes de Carvalho, Daniel Guerreiro e Silva

DOI: 10.14209/sbrt.2017.78

Evento: XXXV Simpósio Brasileiro de Telecomunicações e Processamento de Sinais (SBrT2017)

Keywords: SDN NSGA-II PSA

Abstract

Software Defined Network is a traffic engineering solution with an increasing interest by industry, which raises issues such as the use of multiple controllers. Considering that the positioning of controllers within a given topogoly constitutes a multiobjective optimization problem in order to satisfy certain performance metrics, this work presents a comparative analysis between PSA and NSGA-II, based on the initial study developed by Lange et al. [1]. The results yield good results for both techniques, in terms of accuracy and processing time, but with an advantage to NSGA-II algorithm, in terms of convergence and exploration degree of the search space. Such properties are interesting when one considers real time traffic optimization and a resilient network operation.

Download