Design of estimators using covariance information with uncertain observations in linear discrete-time distributed parameter systems

This paper presents recursive least-squares (RLS) estimation algorithms using the covariance information in linear discrete-time distributed parameter systems. The signal is estimated with the observations containing some uncertain observations. In the uncertain observations, there are cases where the observed value does not contain the signal and consists of observation noise only. The probability that the signal exists in the observed value is used in the estimation algorithms. The algorithms are derived based on the invariant imbedding method