St. Petersburg State University of Telecommunications

About: St. Petersburg State University of Telecommunications is a education organization based out in Saint Petersburg, Russia. It is known for research contribution in the topics: Computer science & The Internet. The organization has 408 authors who have published 465 publications receiving 2160 citations.

Estimation of Minimum Required Number of Coefficients of Digital Filter for Chromatic Dispersion Compensation in Optical Fiber

Abstract: The 5G technology requires the application of optical fiber in the access network. The chromatic dispersion in fiber causes distortion of transmitted signal. The dispersion can be compensated by digital optical filters. The physical size of the filters depends on the number of coefficients. It is proved that the minimum number of coefficients is limited by the physical characteristics of optical fiber and has a quadratic dependence on frequency band. An analytical equation for estimating the number of coefficients is proposed. And the method for decreasing the total number of coefficients is considered.

Modeling multi-beam radio channel

Abstract: In this work, a simulation model of a radio channel with fading is implemented for use in research of the multipath channel, as well as for assessing the noise immunity of transmission, recep-tion and processing systems in modern and future mobile communication and radio access networks. Formalization of mathematical models of a radio channel with fading, including the Rayleigh amplitude distribution, uniform phase distribution and a given Doppler spectrum, made it possible to visualize the time-frequency and probabilistic characteristics of a radio channel with fading. The implementation of the procedures for delaying and attenuating copies of signals in a multipath radio channel made it pos-sible to reproduce scenarios of flat and frequency selective fading, which are widely used in practice in assessing the noise immunity of signals with given frequency and time characteristics.

The use of a matrix decompositions for dimension reduction of training sample

Abstract: Proposed two-step procedure of dimension reduction training sample for pattern recognition neural networks deep learning. The use of the procedure improves the efficiency of object recognition in control systems due to the substantial reduction of computation.