Spectral analysis of randomly sampled signals using a correlation-based slotting technique

Abstract

A detailed investigation of correlation-based methods for generating spectral estimates from randomly sampled signals is described. Particular attention is paid to the effects of quantising the time instants, or lag time, associated with the data samples. Explicit formulas are given for the bias errors associated with various quantisation schemes, and a simple derivation of an asymptotic expression for the variability of the spectral estimates is given. Comparisions with the results of processing simulated data validate the theoretical conclusions and demonstrate the practical value of the proposed estimation algorithms.