Kalman Filtering for LTI Systems with State Dependent Packet Losses

Omanshu Thapliyal, Purdue University

Abstract

Due to recent advances in networked control systems and wireless networked control systems, most systems of interest operate in environments susceptible to sensor malfunctions. The increased dependency on communication channels to transfer measurement data due to the geographical separation of sensors from actuators and plants further contributes to the issue. Most existing methods that deal with the problem of estimation under packet losses assume the packet arrival process to have stationary statistics. In this work we relax this assumption and address the problem of state estimation under state dependent packet losses. This problem presents inherent modeling of practical applications of systems operating in hostile environments, sensor denied environments, or the presence of sensor disruptors/jammers. This estimation problem is formulated as a linear system which a state dependent hybrid measurement model. An optimal estimation algorithm is proposed using a Projection based approach. A special case of this estimation problem is also presented, in the form of a measurement model with Markovian sensor malfunctions (packet losses—full and partial, and multiplicative sensor degradations), which is modeled as a Markov jump linear system. Stability results for the Markovian problem are presented in the form of bounds on the error covariance. Finally, the two estimators demonstrated using illustrative and practical instances and compared against existing estimation algorithms.

Degree

M.S.A.A.

Advisors

Hwang, Purdue University.

Subject Area

Aerospace engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS