NOMADS: Un nuevo paradigma de monitorización y compartición de datos de red

Description
Network monitoring is of paramount importance to both network operators and researchers, to the point that it has recently become a major research area. However, network monitoring still suffers from fundamental problems that leave its rigorousness behind more traditional research fields, such as other experimental disciplines within physics or biology. Most scientific works in the field of network monitoring are evaluated using private, undisclosed data sets. Nevertheless, disclosure of experimental data is a basic principle of the scientific method that enables experiment reproducibility, independent validation and cross-comparison of research results. For this reason, we think that the credibility of research works in this area overly relies on the scientific community bona fide. Our initial hypothesis is that two fundamental barriers have to be overcome in order to increase the rigorousness of this scientific area. First, the mere acquisition of reference data sets is extremely challenging from a technological viewpoint due to the ever-increasing network speeds. Second, sociological reasons discourage their publication, primarily due to privacy concerns. Both the technological and sociological barriers are currently perceived to be insurmountable, which prevents the use of common data sets in scientific works and impedes experiment reproducibility. The main objective of this project is to investigate the fundamental research challenges associated to these two barriers. On the one hand, this project will explore novel network monitoring and traffic measurement techniques in order to address the technological difficulties. Main topics of research in this direction will include complex resource management techniques, such as load shedding and distribution, traffic sampling and specialized streaming algorithms for traffic processing and analysis. On the other hand, this project will propose a novel data sharing paradigm that overcomes the sociological barriers. The research carried out in this project will result in a completely novel evaluation framework for network monitoring research based on what we call the “code-to-the-data” model. This model will enable reproducibility, validation and comparison of scientific works without requiring full disclosure of traffic data sets, thus avoiding most privacy concerns involved in dataset publication. Ultimately, the new monitoring and data sharing paradigm resulting from this project will provide a realistic solution to set the use of common data sets as a standard requirement for scientific publication, as it is in other research areas.

Duration
36 months

Participants
UPC, EPO: Tecsidel, CESCA

Contract number
TEC2011-27474