Andres Ferragut

AI4OPT Seminar Series

Date: Thursday, February 9, 2023

Time: Noon – 1:00 pm

Location: Instructional Center 115 (Scale Up Room) - (759 Ferst Dr, Atlanta, GA 30318)

Join Virtually: https://gatech.zoom.us/j/99381428980

Speaker: Andres Ferragut


Load Balancing for Spatially Distributed Demands With Applications to Electrical Vehicle Charging

Abstract: We will discuss the problem of managing a stochastic and spatially distributed demand of service, for instance, electrical vehicles needing charge. A spatially distributed set of stations are available to serve this demand, and users may need to weigh the transportation cost (time to reach the station) against the congestion cost (e.g. delay until service).

Based on Lagrangian duality, we design a control law for load balancing based on station congestion signals which combine naturally these costs. A fluid, differential equation model is proposed for the resulting closed loop dynamics. Two alternatives arise, depending on the model for client departures from the system: given sojourn times, or given service times. In both cases, the dynamics are shown to converge to an equilibrium, which may be interpreted as a solution to a suitable global optimization problem: the allocation a central planner would obtain for the steady-state distribution of load.

Time permitting, we will also discuss recent work on how to estimate this demand based on Machine Learning, and recent advances on the scheduling problems that may arise within a station when service and power considerations come into play.

Bio: Andres Ferragut is a professor of networks at Universidad ORT Uruguay. He works in the Mathematics Applied to Telecommunications and Energy research group. Ferragut obtained his PhD in Electrical Engineering from Universidad de la República (Uruguay) in 2011. He served as editor of IEEE/ACM Transactions on Networking (2016-2020) as well as TPC Chair of Valuetools (2020). His research work navigates the interface between Queueing theory, Optimization and Control Theory, with a focus on obtaining stability and performance guarantees for complex networked systems.

To meet with Andres, sign up here: https://docs.google.com/document/d/1c9VTd6pd9EQRwCCzjmzjtvslHg0IIwnzkHCTKz6uPJo/edit

Lunch will be served at the seminar. So, please stop by 15 minutes before the seminar to pick up lunch.

To receive AI4OPT seminar announcements, please sign up to our mailing list. at https://lists.isye.gatech.edu/mailman/listinfo/ai4opt-seminars.

Past seminars can be found at https://www.ai4opt.org/seminars/past-seminars.