Abstract

OVIEDO BAYAS, Bryon Wladimir  and  ZAMBRANO VEGA, Cristian. Bayesian networks applied to error prediction in software defined networks. Universidad y Sociedad [online]. 2021, vol.13, n.2, pp. 419-429.  Epub Apr 02, 2021. ISSN 2218-3620.

This article is focused on the analysis of failures detected in software-defined networks (SDNs), which are a highly efficient, scalable, programmable set of networks with high speed and the ability to manage network resources. The research began with the identification and selection of drivers and emulation tools currently used in SDN. Then it is proposed to design and implement a test scenario based on evaluation metrics. The proposed system is based on learning a Bayesian network from the data extracted and processed from an SDN, looking for causal relationships between data values ​​and the state of the network. To emulate the operation of a real SDN, a simulation has been designed with several linked scale-free networks, in which various types of traffic have been injected. The Bayesian network will be used later to diagnose new failures introduced into the network, reasoning with the data extracted from it. Finally, the results obtained with the OpenDayLight controller, the OpenFlow protocol and the Mininet emulator are analyzed.

Keywords : Software Defined Networks (SDN); Bayesian Network; Troubleshooting; Mininet.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )