Figure 2.

The function f is a highly stylized example of a signal with distinguishable temporal epochs. The first, second, and final third have different means, variances and characteristic frequencies. The stochastic process F₁, however, whose realizations are obtained through uniform random circular shifts of f, is statistically stationary in that that the statistical summaries assessed at distinct timepoints over large numbers of realizations are the same. All realizations have spectral and epochally clear variations, which would be reflective of FRBD, but as a stochastic process the collection of phase-shifted versions of f are statistically stationary.