Electrical Engineering-Systems Dept.

סמינר מחלקתי

You are invited to attend a lecture by

Prof. Hanoch Lev-Ari

(Center for Communications and Digital Signal Processing

Northeastern University, Boston)

on the subject:

Networked Multi-Sensor State Estimation with

Delayed and Irregularly-Spaced Time-Stamped Observations

Progress in communication technology has opened up the possibility of large scale control systems in which the control task is distributed among several processors, sensors, estimators and controllers interconnected via communication channels. Such control systems may be distributed over long distances and may use a large number of actuators and sensors that communicate with one (or several) estimation/control hubs. Communication delays, loss of observations and asynchronous sensor sampling are some of the unavoidable effects that degrade the performance of such networked estimation systems.

The length of transmission delays can be determined from the time-stamp attached to each sensor observation, so that the minimum MSE state estimate based on delayed asynchronous multi-sensor observations with irregular sampling patterns can be determined by a standard continuous-discrete Kalman filter augmented with additional storage capacity. The performance of such a Kalman filter is analyzed in terms of the associated error-covariance matrix. We show that the ensemble-averaged error covariance depends only on system parameters and on the characteristic functions of (a) the delay period and (b) the irregular multi-sensor inter-observation interval. We provide explicit expressions that quantify the effect of these channel imperfections on the state error covariance matrix, as well as a necessary condition for its stability, expressed in terms of the region of convergence of the associated characteristic functions. Our results indicate that the average error-covariance depends primarily on the average observation rate F_av , which can therefore serve as the primary control parameter in networked estimation systems with delayed and irregularly-spaced observations: the effects of these (and other) imperfections can be countered by a suitable adjustment of F_av .

Bio: Hanoch Lev-Ari received the B.S., Summa Cum Laude, in 1971, and the M.S. in 1978, both in electrical engineering from the Technion, Israel Institute of Technology, Haifa, Israel; and the Ph.D. in electrical engineering from Stanford University, Stanford, CA, in 1984. During 1985 he held a joint appointment as an Adjunct Research Professor of Electrical Engineering with the Naval Postgraduate School, Monterey, CA and as a Research Associate with the Information Systems Laboratory at Stanford; he stayed at Stanford as a Senior Research Associate until 1990.  He is currently a Professor with the Department of Electrical and Computer Engineering at Northeastern University.  During 1994-1996 he was also the Director of the Communications and Digital Signal Processing (CDSP) Center at Northeastern University.

His present areas of interest include adaptive filtering under the non-stationary regime, model-based spectrum analysis and estimation for non-stationary signals, multi-rate networked estimation and control, and dynamic time-frequency analysis, with applications to over-the horizon (OTH) radar, time-variant system identification, dynamic phasors, dynamic power decomposition, and adaptive power flow control in electric energy systems.  Dr. Lev-Ari served as an Associate Editor of the IEEE Transactions on Circuits and Systems I, and of Circuits, Systems and Signal Processing. He is a Fellow of the IEEE and a member of SIAM.