Robust filtering using the method of local approximations of power spectral densities

The application of the method of local approximations of power spectral densities to robust filtering problems with bounded variances of signal derivatives is considered. The algorithm design is based on the assumption that the signal is formed as n-times integrated white noise. It is important that the upper bound of the root-mean-square filtering error increases by no more than 1% as compared with the exact solution of the robust filtering problem. An example is given to illustrate the effectiveness of the algorithm proposed. © 2014 Pleiades Publishing, Ltd.