Lower order optimal linear filtering of nonstationary random sequences

The following deals with the discrete-time linear minimum-variance filtering of nonstationary random processes. The dynamics of the signal and colored noise processes are represented by a combined random process model.[1] Some of the measurement elements contain additional white noise, others do not. Similar to the continuous-time case of Bryson and Johansen,[3] the white-noise-free measurements will be used to reduce the order of the Kalman filter,[1],[2].