Authors:
Sangho Ko
;
Seungeun Kang
and
Jihyoung Cha
Affiliation:
Korea Aerospace University, Korea, Republic of
Keyword(s):
Covariance Assignment, Estimation, Networked Control System (NCS), Packet Loss, Riccati Difference Equation (RDE), Algebraic Riccati Equation (ARE).
Related
Ontology
Subjects/Areas/Topics:
Sensor Networks
;
Signal Processing
;
Statistical and Adaptive Signal Processing
Abstract:
This paper introduces an improved linear state estimator which directly assigns the error covariance in an environment where the measured data are intermittently missing. Since this new estimator uses an additional information indicating whether each observation is successfully measured, represented as a bernoulli random variable in the measurement equation, it naturally outperforms the previous type of covariance-assignment estimators which do not rely upon such information. This fact is proved by comparing the magnitude of the state error covariances via the monotonicity of the Riccati difference equation, and demonstrated using a numerical example.