Title: Hub location under the risk of Interdiction
Speaker: Dr. R.Prasanna, Post-doctoral fellow at the John Molson School of Business, Concordia University.
Venue: IEOR seminar room (2nd floor)
Date and time: 15 March (Friday) at 2:30 pm
Abstract: In this work, we study the problem of designing a hub-and-spoke network to defend it against deliberate attacks (interdictions). The problem takes the form of a three-level, two-player Stackelberg game, in which the network designer (defender) acts first to locate hubs to route a set of flows through the network. The attacker (interdictor) acts next to destroy (interdict) a subset of the hubs in the designer's network, followed again by the defender, who now acts to route the flows through the remaining hubs in the network. We present a trilevel optimization model of the uncapacitated p-hub location problem with multiple allocation, wherein the defender wants to locate a set of p hubs in a hub-and-spoke network by taking into account the risks posed by an attacker. The attacker's response is modeled as a bilevel r-hub median interdiction problem (r-HMIP), wherein the attacker interdicts a subset of r hubs located by the defender so to maximize the latter's post-interdiction minimum routing cost. We present three alternate single-level reformulations of the bilevel r-HMIP, and provide dominance relationships to identify the best among them. Using computational experiments, we demonstrate that the Benders decomposition of our best reformulation can efficiently solve large instances of the bilevel r-HMIP. Further, from the solution to r-HMIP, we generate supervalid inequalities, which are utilized in an exact cutting plane algorithm to efficiently solve the trilevel hub location-interdiction problem. Using our proposed solution method, which is also the first exact method in literature for trilevel hub-location interdiction problems, we are able to solve large problem instances that are otherwise intractable.
Based on our computational results, we present interesting implications of incorporating interdiction risks in the hub location decision.
Bio: Prasanna is currently employed as a Post-doctoral fellow at the John Molson School of Business, Concordia University. He graduated in the year 2018 as a fellow in Production & Quantitative methods from the Indian Institute of Management Ahmedabad. In his doctoral thesis, he studied preventive and reactive strategies to safeguard hub-and-spoke networks against targeted attacks by a rational attacker. Part of his thesis is published in the European Journal of Operational Research, while the remainder is communicated to a well-known journal. His research interest lies in modelling and solving optimization design problems arising in supply chain applications like Material Good logistics, Urban Transportation and Disaster Preparation & Evacuation. Methodologically, he is interested in Discrete Optimization, Optimization under data uncertainty (Stochastic and Robust), and Multi-level/Multi Objective optimization. Prior to joining the fellow program, he worked as an Assistant Manager in Mahindra Auto Sector for a period of four years. He also holds a Bachelors degree in Production Engineering from Madras Institute of Technology campus, Anna University.