Thursday, April 21, 11:00-noon, EE conference room (13th floor of Mudd)
Abstract: Power efficiency has emerged as an important concern in the design and operation of networks. One important concept is rate adaptivity, or "power follows load", i.e. using low power in times of light traffic and higher power in times of heavier traffic.
Power-follows-load can be accomplished globally by rerouting traffic streams to minimize network-wide power consumption. The resulting problem is a multi-commodity flow problem where the cost function displays diseconomies of scale. For this problem we present a polylogarithmic approximation.
Another approach takes place at a local level by adjusting the clock speed of each individual network element, and hence its power consumption as a function of traffic. In this context, we study how to devise rate-setting algorithms in combination with existing scheduling policies so as to tradeoff power minimization against competing goals such as network stability, delay and queue minimization.
This work is joint with M. Andrews and S. Antonakopoulos.
Speaker Biography: Lisa Zhang is a Member of Technical Staff in the Algorithms Research group in Bell Labs, Alcatel-Lucent. Her research broadly concerns algorithmic and complexity issues of networking, for example network design, routing and scheduling protocols. Her current focus is on energy-efficient computing and communications. She has twice won Bell Labs President's Gold Award for innovation and technical excellence. Lisa received her BA in mathematics from Wellesley College in 1993 and Ph.D in theory of computation from MIT in 1997.