Distributed Processing in Sensor Networks

Abstract

Sensor networks present the opportunity to instrument large scale critical physical infrastructures. Within CenSCIR, at Carnegie Mellon, we launched Sensor Andrew, a campus wide sensor network with the goal of supporting universal sensing, and sustaining heterogeneous applications and technologies. We describe briefly Sensor Andrew and then focus on the challenges of distributed processing. Sensor networks operate under resource constraints and broad imperfect conditions–noisy links that may fail at random times, quantized transmissions, non Gauss noise. We present a unified treatment, in the framework of controlled Markov processes and stochastic approximation, to prove almost sure convergence of distributed processing algorithms in sensor networks under these inter-sensor communication distortions. For clarity, we focus on distributed average consensus. We consider the design of the topology of the sensor network that maximizes the rate of convergence of distributed processing algorithms when communication among sensors has costs and there is an overall cost constraint. We show that, under simplified assumptions, the optimal topology is given by a class of expander graphs – the Ramanujan graphs, and that, under more general conditions, it is obtained by solving a semidefinite programming convex optimization problem.

Work with Soummya Kar.

Biography: José M. F. Moura is a Professor at CMU where he founded CenSCIR, the Center for Sensed Critical Infrastructure Research, and ICTI, the Information and Communication Technologies Institute that manages the CMU-Portugal Program. He holds a D.Sc. in EECS from MIT and an EE degree from IST (Lisbon, Portugal). His interests are in algebraic and statistical signal/ image processing. Current projects are on sensor networks and critical physical infrastructures, time reversal imaging, bioimaging, and intelligent compilers for signal processing applications.

He is President of the IEEE Signal Processing Society. He was the Editor in Chief for the IEEE Transactions on Signal Processing and was on the Boards of the IEEE Proceedings and the ACM Sensors Journal. He is a Fellow of the IEEE and AAAS, and a corresponding member of the Academia das Ciências of Portugal. He received the IEEE 3rd Millennium Medal, the IEEE SPS Meritorious Service Award, an IBM Faculty Award, and the 2007 CIT (College of Engineering at CMU) Outstanding Research Award.