Polynomial interpolation and prediction of continuous-time processes from random samples

We consider the interpolation and prediction of continuous-time second-order random processes from a finite number of randomly sampled observations using Lagrange polynomial estimators. The sampling process (t/sub 1/) is a general stationary point process on the real line. We establish upper bounds on the mean-square interpolation and prediction errors and determine their dependence on the mean sampling rate /spl beta/ and on the number of samples used. Comparisons with the Wiener-Hopf estimator are given.