Errors of recovery of signals in multi-channel communication systems with orthogonal modulation

Abstract

The errors of restoration of original continuous signals in digital communication systems with orthogonal modulation of equidistant subcarriers (the so-called OFDM technology) are considered. The theory of radio engineering circuits and signals is applied as a scientific and technical basis of the problem. The spectra of discrete signals were analyzed, the influence of violations of the conditions of Kotelnikov's theorem and the asymptotics of the recovery errors of the original continuous signals were investigated. Methods of analyzing composite signals with arbitrary selection of the sampling frequency of spaced subcarriers have been developed, and quantitative estimates of the quality of signal recovery on the receiving side using Euclidean distances have been obtained. . It is shown that when the amount of frequency shift of the basic series decomposition functions and the width of the spectrum of the main and side lobes in the spectrum of the output signal are matched, it is possible to count on the minimal influence of mutual interference and intersymbol interference of the composite signal. It was established that the correct selection of the sampling frequency of OFDM signals allows increasing the accuracy of signal recovery on the receiving side and the overall efficiency of the wireless communication system. In addition, with a dynamic change in the number of members of the decomposition series in wireless networks, the influence of frequency collisions, Gibbs ripples, and the level of intersystem interference is reduced. Graphical analysis of the results of the series expansion on the quasi-orthogonal system of basis functions shows that the errors of restoration of the original signals can reach unacceptable values. It was also established that when expanding into a non-orthogonal series with a larger number of members, the situation with the restoration of the original signals becomes a little more suitable, but guarantees of improving the situation with recovery errors can be given only when combining the correct selection of the sampling frequency with the number of members of the series expansion. This problem requires additional research of an analytical and computational nature.