MURI: Information Dynamics as the Foundation for Network Management

MURI: Information Dynamics as the Foundation for Network Management MURI Project: Information Dynamics as the Foundation for Network Management MURI project: Information Dynamics as Foundation for Network Management. The Arizona State University work program will focus on stochastic network optimization and control of wireless ad-hoc/sensor networks and study high-dimensional mathematics for information dynamics. Under the theme of exploiting the underlying geometry to address challenges in Air Force Networking, we will develop robust fastconverging algorithms to guarantee stable control and resilience over multi-scale network dynamics: sessions come and go; packets arrives in bursts; channels vary over time; and network topology changes due to mobility. Given the stochastic nature of wireless networks, there will be a particular emphasis on the impact of stochastic noisy feedback and lossy channel on distributed network algorithms that employ dual decomposition and alternatives. Characterization of delay performance will be carried out using heavy traffic diffusion approximation. We will study the design of software-defined backbone platform responsible for heterogeneous network management, so as to support a variety of services with different quality of service requirements, for both data and real-time traffic. We will also investigate real-time scheduling with hard deadlines for inelastic traffic. Post Doc Sub-Contract with Princeton Dr. Robert Bonneau of the AFOSR has expressed an interest in increasing the presence of smart grid research in the Princeton AFOSR MURI. This work will address that area, as follows: The high variability of wind power has posed significant challenges to power system reliability. It is thus imperative to develop accurate wind generation forecasting approaches. The available big data for wind generation, i.e., more detailed data for turbines output, more diverse metrological data (wind speed/direction, pressure, temperature, etc.) and more comprehensive geographic information system (GIS) data, provide unprecedented opportunity for this task. We will take a novel big data analytics approach to obtain rigorous characterizations of wind generation dynamics, and develop accurate forecast models for single wind farm generation and multiple wind farm generation. Then, we will explore stochastic optimization for joint economic dispatch and interruptible load management, towards seamless integration of wind generation into the power system operations. We expect this framework can significantly reduce the regulating reserves needed to compensate for the uncertainty of wind generation