Cyber Center

ONR Grant Awarded

February 2, 2016

Here is a summary of the ONR grant:

With pervasive sensors continuously collecting massive amounts of information as well as advances in computing, communication, and storage technologies, this is an era of data deluge. Modeling, processing and ultimately making sense of these massive-scale data sets is expected to bring ground-breaking advances in many defense applications, including distributed control, communication and decision support for tactical networks; assisted navigation and tactical surveillance applications; and multimodal information networks, just to name a few. The sheer volume of the data to be processed, together with the growing complexity of the data models (possibly nonconvex and nonlinear) and the increasingly distributed nature of the data sources presents major challenges to the modern big data analytics.

To address such challenges, this project envisions a “fog computing” architecture enabling nonconvex streaming analytics using parallel processors and distributed in-network processing over massively distributed data sets. The crux of the proposed algorithmic framework is a novel convexification-decomposition technique for a general class of nonconvex unstructured (possibly stochastic) problems with nonseparable objective functions and constraints. This new class of algorithms addresses the shortcomings of current (nonparallel and nondistributed) convex approximation techniques by enabling full control of the degree of parallelism and distribution of the computation/signaling among processors/network nodes as well as offering a more flexible selection of the design parameters such as convex approximants, step size schedules, and communication protocols.

This project embarks on an ambitious transformative multidisciplinary research that aims at advancing the state-of-the-art of in-network big data processing. Insights gained from analyzing the proposed algorithmic framework will be beneficial for a gamut of exciting problem domains far beyond big data analytics, and would positively impact nearly all of the ONR deployed settings including but not limited to distributed multi-agent optimization and inference over networks, autonomous navigation and tactical surveillance, as well as distributed control, communication and decision support for tactical networks (e.g., ad-hoc networks, UAVs, underwater networks) in real-world scenarios (e.g., in the presence of random or dynamic topologies, stochasticities, communication failures, and quantized internode transmissions).

Bookmark and Share

Contact Details

More Information