On Optimal Linear Filtering in the Presence of Correlated Measurement Noise

摘要

The problem of optimal linear filtering for discrete linear processes in which the measurement noise is correlated is reformulated as a linear filtering problem that involves uncorrelated noise, the latter problem is of higher dimension than the former, but it can be solved by standard tecnhniques, the resulting optimal filter and estimation-error covariance equations are partitioned and placed in a form that is especially suited for digital computation, the result is that the effects of increased dimensionality on computer program complexity are minimized. (Author)