Signal space geometry (Q1090658)

In the paper a new approach to the problem of estimating a sum of imaginary exponentials (multiple frequencies) in the presence of noise, assumed to be zero mean Gaussian noise, is presented. The process is available by a set of measurements at the times \(t_ 1,...,t_ n\) and this is performed N times, all the measurements are assumed to be independent. It is shown that the geometric least square criterion can be viewed as that of calculating the shortest distance from a point to a set in a Grassmannian manifold with a coordinate free metric. We think that this approach can reveal new applications of differential geometry in estimation theory.