Optimal observations for minimum variance filtering
Stanley R. Pliska · IEEE Transactions on Automatic Control · 1974
Consider optimal filtering for a linear, discrete-time, dynamical system with scaler state xkand observation variance nk1at time tk. Then the error varianceP(k|j)corresponding to the minimum-variance estimate of xkgiven the observations through tj,k \geq j, is a convex function of (n1. . .,nj) on the nonnegative orthant of Ri.