Abstract
The effects of distributed sampling periods on hybrid data processing are studied theoretically. It is assumed that the difference Δτ of each sampling period deviates from its mean according to a probability density function, p(Δτ). Firstly, it is shown that the characteristic function P(ω) of p(Δτ) behaves as a frequency window if the exact sampling times are used. Then the effects of neglecting the randomness are examined and statistics of errors of resultant spectra are discussed. It is clarified that the effects are essentially equivalent to those which appear when a signal is sampled with an exactly equal period after pre-filtered by a filter, whose transfer function is P(ω). Some applications of a sampler with distributed sampling periods and results of computer simulation are also presented.
