Min- max linear estimation of band-limited sequences in the presence of random noise (Corresp.)

A problem in band-limited signal estimation is presented. The signal is assumed to be a band-limited sequence subject to an energy constraint. The observed signal components are assumed to be corrupted by the components of a discrete-time stochastic noise process with zero-mean and known antocorrelations. The error metric is the energy in a fixed linear combination of the error components. Linear estimators are considered, and an explicit expression for the optimal (in a min-max sense) estimation scheme is derived. An interesting feature of the optimal estimation scheme derived in this correspondence is that it is identical to one based on stochastic models for both the signal and the noise.