Filtering, predictive, and smoothing Cramér-Rao bounds for discrete-time nonlinear dynamic systems (Q5950245)

The authors consider the discrete-time nonlinear system \[ \begin{aligned} x_{k+1} & =\varphi(x_k,w_k)\\ z_k & =\gamma_k(x_k,v_k),\;k=0,1,2, \dots, \end{aligned} \] where \(x_k\) is an \(n\)-dimensional state vector, \(z_k\) is an \(r\)-dimensional measurement vector, \(\{w_k\}\) and \(\{v_k\}\) are mutually independent white sequences. The aim is to find a recursive algorithm for the Cramér-Rao lower bounds (the Cramér-Rao bound is defined as the inverse of the Fisher information matrix) of the mean-square error matrix \(E\{(x_k-\widehat x_k)(x_k- \widehat x_k)^T\}\), where \(\widehat x_k\) is an estimate of the state \(x_k\). The authors derive recursive relations for predictive and smoothing the Cramér-Rao bounds as well. A discussion of the Cramér-Rao bounds can be found in [\textit{B. Z. Bobrovsky}, \textit{E. Mayer-Wolf} and \textit{M. Zakai}, Ann. Stat. 15, 1421-1438 (1987; Zbl 0642.62018)].