SPEAKER:  Ketan Savla, Faculty Candidate

ABSTRACT:

Queueing systems, along with flow network approximations, provide a fruitful framework
for several applications such as transportation, production and data networks. In this talk,
we present a novel generalization of this framework that explicitly incorporates dynamical
aspects inspired by well-known empirical findings. In particular, two scenarios will be
discussed. First, we present a novel dynamical queue model in which the service times
depend on the utilization history of the server. For such a queue, we show that a simple
threshold policy, that releases a task to the server only if its state is below a certain fixed
value, is throughput-optimal. Second, we consider a dynamical flow network where the flow
dynamics is driven by the difference between the inflow and outflow on the links. For such a
flow network, we show that the node-wise routing policies that respond cooperatively to
variations in flow densities on local links in fact provide maximum global robustness
guarantees under local information constraint. These results rely on technical tools at the
intersection of dynamical systems, queues and network flows, and provide key insights into
the fundamental performance limits in presence of dynamical effects.

(joint work with E. Frazzoli, G. Como, D. Acemoglu and M. A. Dahleh)

Bio
----
Ketan Savla is a research scientist at the Laboratory for Information and Decision Systems at MIT. He obtained his Ph.D. in Electrical Engineering and M.A. in Applied Mathematics, both in 2007, from UCSB, as well as M.S. in Mechanical Engineering from UIUC in 2004. His current research interest is in control and optimization techniques with applications in mobile robotic networks, humans-in-loop systems, intelligent transportation systems and computational neuroscience. His awards include CDC-ECC'05 best student paper finalist and best CCDC thesis award from UCSB.