I'd describe the confluence of some global trends that will drive an architectural transformation of future power networks. In particular, sustainability demands increasing use of renewable energy such as wind and solar generation. One of the key challenges with renewable energy integration is its intermittency and uncertainty. I will present some smart grid research at Caltech to mitigate this uncertainty, in particular, a proposal for real-time dynamic load management, inverter control for voltage support in distribution networks with a high photovoltaic penetration, and optimal power flow problems.
(Joint work with S. Bose, M. Chandy, L. Chen, M. Farivar, D. Gayme, L. Jiang, J. Lavaei, N. Li)
Mark Squillante (IBM Research, United States), Performance Analysis in Information Technology Services
Performance measurement, modeling and optimization are fundamental aspects of computer system/network design and management. Many of these same methods, however, are playing an equally important role in a wide variety of emerging applications. One such area is information technology services, in which service providers typically offer a broad range of service products, each requiring resources with certain capabilities, in markets characterized by highly volatile and uncertain customer demands. We investigate important problems associated with resource management in information technology services and present an integrated suite of performance modeling and analysis methods to address these problems. This includes various forms of resource capacity planning, one capturing the risks of having insufficient capacity and the risks of having too much capacity, and another capturing the evolutionary dynamics of resources. Experience with these methods in practice and real-world numerical results demonstrate the benefits of our performance analysis approach.
This presentation is based in part on joint works with: H. Cao, J. Hu, C. Jiang, T. Kumar, T.-H. Li, Y. Liu, Y. Lu, S. Mahatma, A. Mojsilovic, M. Sharma, Y. Yu; K. Jung, D. Shah; J. Anselmi.
We consider large scale distributed server systems, featuring both memory and bandwidth resource constraints. Our aim is to identify memory management strategies which maximize overall system capacity.
We first propose the so-called proportional content placement strategy, and characterize its efficiency in a large system asymptotic regime. This result features a non-classical loss network in a random environment.
An alternative model is considered next, which measures performance via the size of matchings in suitable graphs. In this context, we first show the impact of the matching strategy by comparing greedy and optimal matchings. In particular, we show that greedy matching undergoes a phase transition, with a severe performance degradation, at critical loading. We also investigate improvements on the previous "proportional" placement strategy. Underlying these results is a characterization of density of maximum matchings in large bipartite random graphs.
This presentation is based on joint work with Marce Lelarge (INRIA/ENS), Mathieu Leconte (Technicolor/INRIA) and Bo Tan (UIUC).