Effectiveness of time and digital versions of radiometry when detecting sources of ionizing radiation

In radiometric practice, experiments to detect sources of ionizing radiation with random time characteristics are usually carried out in two ways, in which sliding filtering of the data at m fixed intervals is used: (1) the time interval t0 is fixed, and the filter is organized for the number of readouts k in the time t0 (henceforth called the open-quotes digitalclose quotes version of radiometry), and (2) the range of numbers k0 is fixed, and the filter is organized for time intervals t, during which k0 successive readouts occur (henceforth called the open-quotes timeclose quotes version). It is usually stated that both versions are equivalent, but the conclusion is drawn that the second version is preferable, since, in the author's opinion it provides better time resolution for the same detection efficiency, i.e., for an equal level of significance and power of the criterion for background-signal classification. Nevertheless, there are indications in the literature that there are differences in the conclusions when processing data using one version or another, for example, in the standard problem of determining the intensity of the readouts in the detector. These differences are presented in this paper. Analysis of the set of detection characteristics concludes that the time and digital versions of radiometry and, obviously, the statistical models of the Poisson flow are not equivalent. Filtering of the data with a fixed time exposure provides a better signal/noise ratio when using the radiometer both to estimate the radiation parameters and for detection. The use of the simplest adaptive processing employing modified maximum-likelihood estimates of the signal level increases the efficiency with which nonstationary signals can be detected in both cases