Graphical Reasoning for Decentralized Systems

Decentralized systems are ubiquitous in many applications Including communication networks, sensor networks, networked control systems, etc. Optimal design of decentralized systems poses new challenges, both conceptual and computational. In this talk I present how to reason about optimal design of decentralized systems using graphical models. This method allows us to easily extend the intuition of centralized systems to decentralized systems. Moreover, this method can be automated, allowing an automated derivation of structural results of optimal decision makers. I will present examples from real-time communication and decentralized control.

(Joint work with Sekhar Tatikonda at Yale University)

Aditya Mahajan received B.Tech. in Electrical Engineering from Indian Institute of Technology, Kanpur in 2003, and M.S. and Ph.D. in Electrical Engineering and Computer Science from University of Michigan, Ann Arbor in 2006 and 2008, respectively. He is currently a post-doc at Yale University. His research interests include decentralized systems, communication with delay constraints, feedback communication, discrete event systems, and resource allocation and scheduling.