Events Calendar

Stochastic Network Optimization and the Theory of Network Throughput, Energy, and Delay

Speaker: Michael J. Neely,
USCAbstract: This talk has two parts:
(i) We first summarize the landmark results in the theory of stochastic network optimization over the past two decades. This theory treats routing, scheduling, resource allocation, and flow control in general networks, including ad-hoc mobile networks with time-varying topologies and unreliable channels. We consider the history of network capacity and stability theory, as well as our more recent contributions on joint stability and performance optimization. This allows for optimization of time averages of network utilities and costs (such as throughput, fairness, energy, reliability, etc.) subject to general time average constraints. Simple techniques of backpressure, max-weight decision making, and virtual queues can be used to optimize these performance metrics to any degree of accuracy, with an explicit tradeoff in end-to-end average network delay. (ii) We then focus on the delay metric itself by treating a particular network: A multi-user wireless downlink (or uplink). We generalize the Berry-Gallager bound to this multi-user case, establishing a fundamental tradeoff between average power expenditure and average delay. Scheduling to achieve the optimal tradeoff is a problem that is notoriously complex, and the complexity quickly explodes as the number of users is increased beyond 1. Nevertheless, we overcome this complexity explosion through a novel dynamic control policy that aggressively steers drift in desired directions. The policy works with low complexity, is real-time implementable, does not require knowledge of the channel or traffic statistics, and has quick convergence for any number of users.Details of these results can be found in the following references: 1. L. Georgiadis, M. J. Neely, L. Tassiulas, "Resource Allocation and Cross-Layer Control in Wireless Networks," Foundations and Trends in Networking, Vol. 1, no. 1, pp. 1-144, 2006. http://www-rcf.usc.edu/~mjneely/pdf_papers/NOW_stochastic_nets.pdf2. M. J. Neely, "Optimal Energy and Delay Tradeoffs for Multi-User Wireless Downlinks," IEEE Transactions on Information Theory, vol. 53, no. 9, pp. 3095-3113, Sept. 2007. http://www-rcf.usc.edu/~mjneely/pdf_papers/energy-delay-it.pdfBiography: Michael J. Neely received B.S. degrees in both Electrical Engineering and Mathematics from the University of Maryland, College Park, in 1997. He then received a 3 year Department of Defense NDSEG Fellowship for graduate study at the Massachusetts Institute of Technology, where he received an M.S. degree in EECS in 1999 and a Ph.D. in 2003. During the Summer of 2002, he worked as an intern in the Distributed Sensor Networks group at Draper Labs in Cambridge. He is currently an Assistant Professor in the Communication Sciences Institute (CSI), within the Electrical Engineering Department at the University of Southern California. His research interests are in the areas of stochastic network optimization and queueing theory, with applications to wireless, satellite, mobile ad-hoc networks, and switching systems. Michael received the NSF Career award in 2008. He is a member of Tau Beta Pi and Phi Beta Kappa.

Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248

Audiences: Everyone Is Invited

Contact: Gerrielyn Ramos