A Novel Spectral Estimation Method by Using Periodic Nonuniform Sampling

In this paper we present a method of estimating power spectrum density of random ergodic signals. The method allows use of arbitrarily low sampling rates to achieve the goal. We compare our method with similar schemes reported in research literature and argue superiority of our approach in terms of its suitability for practical implementations. The most visible difference between our approach and the previously reported ones consists in replacing Poisson additive random sampling with deterministic sampling. Comparing with the approaches based on the Poisson additive random sampling, where theoretically infinitely large resources are needed to implement them accurately, our approach clearly relies on limited and well defined resources.