Abstract: A minimum separation between successive samples is a practical constraint that often comes in the way of sampling of a continuous time stationary stochastic process for the purpose of spectrum estimation. It is known from a recent study that additive random sampling subject to the said constraint can be alias-free for bandlimited spectra with any specified support, but known estimation approaches do not work. In this paper, we propose a new spectrum estimator for this purpose and show that it can accurately and precisely estimate any power spectral density limited to an arbitrarily large but known support.
Key words and phrases: additive random sampling, bandwidth, smoothing, spectral density.