Estimation with Intermittent Observation Using Both a Priori and a Posteriori Information

For the systems in which missing observation occurs, we drive a new estimator which uses a priori information in state update and a posteriori information in covariance update. The resulting filter is a minimum variance filter which minimizes the expectation of error covariance with respect to noises and the random variable representing missing observation. The merit of presented filter is that the covariance update equation does not include random variable representing missing observation and hence the covariance can be calculated off-line in time invariant system.