Analysis of discrete-time Kalman filtering under incorrect noise covariances

This paper analyzes the behavior of the discrete-time Kalman filter under incorrect noise covariances. The quality used to measure the filter performance is the actual one step predictor error covariance. Convergence and divergence properties of the actual one-step predictor error covariance are analyzed. Through this analysis, useful insights into the behavior of the Kalman filter designed with inexact values of noise covariances are gained.