In the core of 5G networks, softwarization is a main trend. Software Defined Networking (SDN) is one of the most promising solutions for the Future Internet. SDN is a perfect domain for the application of machine learning tools to solve problems in the NOC/NSC, as evidenced by a wide corpus of scientific publications studying problems such as anomaly detection, traffic classification and performance optimization, to cite a few. This combination of SDN and machine learning was recently captured by the concept Knowledge-Defined Networking (KDN). KDNs are SDNs where management decisions are driven and/or supported by machine learning and data analytics. For the first time in this project, networkmetrics will be applied to SDN.

Big Data processing will be essential in the delivery of the network programmability in SDNs. Data is often referred to as “Big” if it is a collection of large, complex and fast growing data streams. Big Data tools can support SDN in various tasks, such as traffic monitoring, prediction, intrusion detection, quality of service implementation and optimization. One promising approach in this domain is the Multivariate Big Data Analysis (MBDA). MBDA is a general framework for networkmetrics in Big Data. In this project, we will extend for the first time the MBDA methodology with sparse, multi-way and supervised models, which have been succesfully applied to other domains of knowledge.

There are tremendous opportunities for research and innovation in network engineering by looking for inspiration in areas like chemometrics or econometrics, which during decades have devised a plethora of multivariate approaches that can be adopted within networkmetrics. We believe that the adoption of new technology can be promoted by the interaction of the right people in the right context. For this reason, we devise in this project a strategy to push the international adoption of networkmetrics, creating a forum for discussion between network engineers and multivariate analysis specialists, and providing access to tools to facilitate new networkmetric projects.