“Optimal remote estimation over the collision channel”
Wednesday, September 07, 2016 EEB 248 2:00 pm
Abstract: Cyber-physical systems often consist of multiple non-collocated components that sense, exchange information and act as a team through a network. In this talk, we will discuss a one-shot decentralized remote estimation problem where a team of sensors make independent measurements and decide whether to transmit them or not to a fusion center over a collision channel. The communication constraint imposed by the collision channel is such that only one transmission may reach the fusion center and, if more than one sensor transmits, a collision is declared. We will show that the search for solutions that minimize a mean squared error criterion can be restricted to a class of policies with a deterministic threshold structure. Then, we will derive an algorithm based on quantization theory to compute person-by-person optimal policies and show its global convergence to a locally optimal solution. Finally, we will show how to extend these results to a model with dependent observations where sensors make private and common measurements.
Bio: Marcos Vasconcelos received the Ph.D. degree in Electrical and Computer Engineering from the University of Maryland, College Park, in 2016. He previously obtained his B.Sc. and M.Sc. degrees in Electrical Engineering from the Federal University of Pernambuco (UFPE), Recife, Brazil, in 2004 and 2006, respectively. He is a postdoc in the department of Electrical Engineering at the University of Southern California hosted by Prof. Urbashi Mitra. His current research interests include networked control and estimation, optimization and information theory.