Skip to Main Content
Algorithms are developed for the error analysis of the optimal filtering solution for colored noise. The types of error sources considered are those both in the imprecise specification of the model and in the improper choice of the noise statistics. The matrix differential equations have been derived permitting the calculation of the actual error covariance and the additional error due to the preceding error sources. For a case where only the error source based on the incorrect choice of the noise statistics is present, a conservative design criterion has also been established, which makes it possible to prescribe the performance of the filter for colored noise. A simple example demonstrates the utility of these results.