Bounding estimators for systems with colored noise

Attention has been directed to the problem of estimating the state of a linear dynamic system which has correlated noise. The cross correlations between the random vectors affecting the system are assumed unknown. A linear estimator is developed which bounds the corresponding error covariance matrix for any allowed cross correlation between the random disturbances in the plant and/or in the measuring device. The computational requirements are of the order of the corresponding Kalman filter assuming white and uncorrelated plant and measurement noise processes.