Performance comparison of H-infinity and Kalman filters and equalisers
GA Einicke, L.R. White · Information Sciences, Signal Processing and their Applications · 1996
This paper compares the performance of Kalman and H infinity solutions for filtering and equalisation problems. It is argued that the cost of H infinity designs is an increase in mean square error. In the case of known time-invariant models, it is shown that the H infinity solutions are closer to the high SNR asymptotes. In the case of unknown models, it is shown how uncertainty may be accommodated in the filter and equaliser designs.