This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).
Network matrices, such as traffic matrices, delay matrices, adjacency matrices, and social proximity matrices, are essential inputs to a wide range of network applications. Network matrices and the underlying network often exhibit a multi-faceted scaling behavior. To capture such multi-scale behavior, a particularly promising approach is Multi-Resolution Analysis (MRA), which creates multiple approximate representations of network matrix at different resolutions. This project will (i) develop a novel framework to enable network-centric MRA of network matrices, and (ii) use the framework to develop novel solutions to several network management tasks: missing value inference, design of experiments, traffic synthesis, and anomaly detection.
Intellectual Merit: The project is multi-disciplinary by nature and will foster effective synergy between networking, statistics, data mining, and scientific computing. The MRA framework and its applications will deepen the understanding of the spatial and temporal characteristics of network matrices at different scales, and advance the state of art in several significant network management tasks.
Broader Impact: The MRA framework is valuable to multiple scientific fields. The project is expected to produce publications in leading conferences and journals, and software that will be publicly available online. Through technology transfer, the network management solutions can potentially improve the operations of real ISP networks. The project will provide several graduate students' thesis research and honors undergraduate research projects. The research results will also be integrated into undergraduate and graduate curricula as well as outreach activities.
