A confidence ellipsoid approach for measurement cost minimization under Gaussian noise

The well-known problem of estimating an unknown deterministic parameter vector over a linear system subject to additive Gaussian noise is studied from the perspective of minimizing total sensor measurement cost under a constraint on the log volume of the estimation error confidence ellipsoid. A convex optimization problem is formulated for the general case, and a closed form solution is provided when the system matrix is invertible. Furthermore, effects of system matrix uncertainty are discussed by employing a specific but nevertheless practical uncertainty model. Numerical examples are presented to discuss the theoretical results in detail.