This project aims are to study open issues in SDN related to scalability and complexity of large deployments and to promote knowledge creation increasing the critical mass in this subject. The key challenges are:
Scalability: Modern networks are subject to a high amount of data to transport. Having programmable devices such as SDN controllers encourage the gathering and the handling of large amounts of control information. The situation can become unfeasible if every decision is taken centrally.
Network data measurement and treatment: SDNs can provide network data that needs to be treated. Just a few attempts have been made to define ways to classify information, infer data correlations and predict future events.
To address the scalability issue the approach is to study solutions where some level of local control and decision exists. Realtime decisions can be made locally based on available information without any interaction with a centralised entity while more strategic actuations are done at a more central level of control that deals with orchestration and strategic tasks for the particular problem at stake benefiting from a more global view of the network.
In the network and data measurement case we will study the use of big data tools like classification, clustering and association rules algorithms to infer correlations between measured network data.
The project methodology will be to apply this vision to three uses cases that cover some of the most important problems of the current networks: managing policy based multipaths between nodes in a network, resource optimisation of the network and Dynamic access control.
|Start Date: 01-04-2016|
|End Date: 01-03-2018|
|Team: Pedro Miguel Figueiredo Amaral, Paulo da Costa Luis da Fonseca Pinto, Luís Filipe Lourenço Bernardo, Susana Isabel Barreto Miranda Sargento, Carlos Miguel Ferreira|
|Groups: Radio Systems – Lx, Network Architectures and Protocols – Av|
|Local Coordinator: Pedro Miguel Figueiredo Amaral|