Robust Tuning of Kalman Filters

The usual method of tuning a Kalman-Bucy filter is to utilize both fictitious process noise (or plant disturbance vector), of intensity matrix Q, and independent fictitious measurement noise, of intensity matrix R, and to vary Q in relationship to R. The recent analysis by Bass [1] of a famous counter-example of Doyle [2] shows that in terms of Bass´s 1956 definition of robustness margin ¿ (cf. [1]) there may be no way to improve ¿ as long as the process and measurement noises are assumed to be uncorrelated. However the following scheme increases ¿ toward its upper bound as the design parameter ¿¿ + ¿.