Sequential Decision-Making in Decentralized Systems
Mar 12, 2013
from 10:00 AM to 12:00 PM
|Where||Engr. IV Bldg., Shannon Room 54-134|
|Contact Name||Prof. Mihaela van der Schaar|
|Add event to calendar||
University of California, Berkeley
Decentralized systems are ubiquitous in the modern world. Communication systems, sensor networks, power systems and economic systems like markets and auctions are all examples of decentralized systems. Such systems are characterized by the presence of multiple decision-making agents acting on different information. In this talk, I focus on the problem of finding optimal decision-strategies for co-operative agents in a decentralized system. In particular, I consider a decentralized stochastic decision-making problem with multiple decision-makers that share information with each other with a fixed delay. Such decision problems arise in queuing networks, wireless communication networks, distributed control systems, sensing and surveillance systems etc. In spite of initial conjectures as early as 1971, finding the general structure of agents' optimal decision-strategies with delayed information sharing had remained an open problem for 40 years. My research provides a conceptual framework that not only identifies the structure of optimal decision strategies but also provides a sequential decomposition of the optimization problem. Moreover, the methodology developed here is shown to be applicable to a broader class of decentralized decision making problems arising in diverse application domains.
Ashutosh Nayyar received the B.Tech. degree in Electrical Engineering from the Indian Institute of Technology, Delhi, India in 2006. He received the MS degree in in Electrical Engineering and Computer Science in 2008, the MS degree in Applied Mathematics in 2011 and the PhD degree in Electrical Engineering and Computer Science in 2011, all from the University of Michigan, Ann Arbor. He worked as a post-doctoral researcher at the University of Illinois at Urbana-Champaign from Fall 2011 to Summer 2012. He is currently a post-doctoral researcher at the University of California, Berkeley. His research interests include decentralized stochastic control, game theory, mechanism design and their applications in sensing and communication systems, decentralized control systems and electric power systems.