Estimation of correlation functions and spectra from randomly sampled data

A theory has been developed for estimating the autocorrelation function of a signal from a set of samples taken at random times. The estimator operates on the random samples to obtain a correlation function and the Blackman-Tukey method is then used to obtain the spectral estimate. The advantage of random sampling is that there is no "aliasing" error as occurs when equispaced sampling is used. Thus good spectral estimates can be obtained when the average sampling rate is a small fraction of the sampled signal\´s bandwidth. The estimator assumes no prior knowledge of the sampling statistics and the sampling sequence can be either a stationary or nonstationary process.