Operational forecasting of optimal operating frequencies for long­range radio communications under the influence of non­fluctuating interference

Abstract

The analysis of the methods of moving average and moving median, which are widely used to build empirical models for forecasting optimal operating frequencies, is carried out. The reasons for the resulting distortions are shown. The methodological bases for processing the results of discrete observations developed for the operational forecasting of the values of optimal operating frequencies for a dedicated radio communication channel under the influence of non-fluctuating interference are described. The spectral analysis used to calculate the mathematical expectation shows that, due to averaging, such models do not make it possible to detect the influence of external sources of perturbation on the ionosphere. That is, in the case of such averaging, together with the random component, which is zero, the effect of ionospheric non-fluctuating disturbances on radio communications is excluded from consideration. To account for the influence of ionospheric disturbances in a standard ionosphere model, these non-fluctuational disturbances are added separately. In addition, in geophysics, much attention is paid to the calculation of various averages for days, months, years, etc. It is important to establish to what extent the results obtained can be free from insurmountable distortions due to discretization. A condition for the extrapolability of the process is its spectrum limitation, so to obtain a limited spectrum, the observation data must be subjected to digital low-pass filtering (instead of the operations of equal-weighted averaging or moving median). For the initial data processing, it is suggested to connect two filters in series: a continuous weighted averaging and an ideal low-pass filter. The Chebyshev filter is the best fit for the characteristics of the corrected continuous averaging filter. Using the method of digital extrapolation, a method for forecasting observation data is presented. The conditions of extrapolability of the real process are given. A set of programs for realizing the forecast of optimal operating frequencies in real time has been developed.