Skip to main content

Talk by Umang Bhaskar

Title: Using Tolls and Signals to Obtain Good Equilibria in Routing Games

Speaker: Umang Bhaskar, TIFR, Mumbai

Venue: IEOR Teaching lab (ground floor)

Date and Time: 26th February 2018, 10.30 - 11.30am

Abstract: Network routing games are game-theoretic models of traffic in networks. In a network routing game, each agent picks a route from its source to its destination in the network to minimize its latency, where the latency on each edge depends on the population of agents that use the edge. Most analysis of routing games approach however assumes precise, detailed information about the edge latency functions, which may be unavailable and difficult to obtain. We study two problems related to incomplete information on the latency functions.

Firstly, we study whether one can induce a given target flow as the equilibrium in a network routing game without knowing the underlying latency functions. We are allowed to place tolls on the edges of the network, and observe the behaviour of the players in response. Secondly, we study the Bayesian setting, where the edge cost functions depend on the state of nature, which may be hidden. Instead, the system designer is allowed to send the players a signal regarding the state of nature, which they use to form beliefs on the state of nature and choose their strategies. Our goal is to find the optimal signaling scheme, which minimizes the social cost.

Bio: Umang Bhaskar is a faculty member at the School of Technology and Computer Science, Tata Institute of Fundamental Research, Mumbai. Before this, he was a postdoc at Caltech and at University of Waterloo. He obtained his PhD from Dartmouth College. His research is on algorithmic game theory, especially on the properties of equilibria in network congestion games.
 

 

 

 

 

News Category