Risk-sensitive filtering, prediction and smoothing for discrete-time singular systems (Q1858371): Difference between revisions

The paper studies in a unified form steady-state risk-sensitive filtering, prediction and smoothing for discrete-time singular systems. In particular, it is shown that the risk-sensitive estimation problem is equivalent to a minimax optimization problem for an indefinite quadratic form. This allows a simple derivation of the required estimators. A numerical example illustrates the procedure described.