Vol 18, No 3 (2020)

The frst part of the article provided an overview of the basic principles of the promising direction of modeling and analysing infocommunication networks and systems - Network Calculus theory. The second part of the article considers examples of practical application of Network Calculus theory to the analysis of infocommunication networks and systems. A qualitative comparison was given of trafc delays estimates in the queuing system obtained using the methods of queuing theory, network calculus theory and simulation, with delays in a real system. The results of using network calculus to assess the boundary characteristics of the service trafc quality in the most promising infocommunication networks are presented. The features of the use of deterministic network calculus for estimating end-to-end delays in sensor networks and the upper queues boundaries in the bufers of the controller and switches of software-defined networking are described. Application of the theory of stochastic network calculus to obtain boundary estimates of end-to-end delays in LTE mobile networks for trafc fows of real and unreal time is shown.

The task of research of non-Poisson trafc obtained during the procedure of statistical characterization at the specifed rates of packets arrival is considered. In order to investigate the statistical characteristics of multimedia stream, multicast data transmission is considered, which implies that the server generates one data stream and sends it over the network to connected clients. The load rate of the total packet fow at each point in time depends on the software products serving the outgoing requests and the requests number ratio by taking into account these software products. Based on histograms of measurements, approximate functions of time interval distributions between packets and duration of packets, their approximating expressions are obtained in the form of sum of damped exponents satisfying the properties of distribution density function of random quantity. Values for the average packet delay time in the network and queue length are obtained using the spectral method of solving the Lindley integral equation for G/G/1 queuing system. The solution accuracy is determined by the approximation accuracy of the «heavy» tail distributions used.

Implementing the almost limitless possibilities of a software-controlled network requires additional study of its infrastructure level and assessment of the telecommunications aspect. Therefore the aim of this study is to develop an analytical model for analyzing the main quality indicators of network switches and communication channels. To form a mathematical model of the switch for software-controlled network operating in a stationary mode and serving the requests of variable volume, it is proposed to use the queuing system theory, as well as the apparatus of semi-Markov processes. The switch model is formally presented as an open queuing network with Poisson fow of incoming packets, refusals and blockings. The frst stage of servicing requests includes the processes of writing network packets to the switch multiline bufer memory. In the next center the correspondence of incoming packet address to the records in the fow tables is being checked. The third single-line center implementing the transmission process of the packets leaving the switch can be described by a system with independent single-line queuing systems with n-place bufers. Dependencies of queuing time and service time of incoming network packets on load are obtained, as well as equations for fnding the volume of a switch bufer memory with acceptable probability for message loss, the quality indicators of its transmitting part.

The object of the research is the cluster systems of data centers containing a certain set of application servers, a servers file, data storage systems, an input-output system connected by a switching system and communication channels. The goal of the work is to increase the efficiency of using virtualized cluster systems by developing and applying a rational method for distributing virtual machines among the physical elements of the data centers. The proposed method is based on iterative greedy algorithm and a limited search procedure allowing to increase the performance of virtualized system due to rational data distribution. A simulation functioning model of the virtualized system was developed, its experimental study was carried out. The presented approaches provide determination of the required number of cluster nodes of the virtualized system with unpredictable load intensity and unpredictable level of resources requested by virtual machines.

Algorithms for controlling stream video traffic are considered. The features of video codec traffic, which has a pronounced packet character, are analyzed. The characteristics of such flows and their influence on the size of queues and delays in telecommunication network nodes are investigated. A generalized Khinchin-Pollachek formula is given and linear dependence of the formula numerator on the load factor is shown. The main causes of packet delays in the queues of a telecommunication network are considered and the influence of these delays on the processes of controlling stream video traffic is shown. A video stream control algorithm is proposed based on size measurements of the buffer memory in the client’s equipment. It is shown that such an algorithm eliminates the influence of delays from the feedback loop of the control system. The results of simulation modeling of the considered processes are presented.

The features of the existing classification of queuing systems are considered, its disadvantages are shown. Only the laws of time intervals distribution between neighboring requests and time intervals for processing requests are used as the main flows characteristics. The main disadvantage is the lack of information about the correlation and cross-correlation flows properties. It is shown that flows that have the same distribution laws but different correlation properties can form queues where values differ by several orders of magnitude. Flows controlled by the Markov chain and, in particular, examples of hyperexponential flows classification are considered. It is shown that this kind of flows can have the same probability distribution laws for time intervals between neighboring requests, but significantly different queue sizes caused by the difference in the flows correlation properties. Another disadvantage of the Kendall’s classification is inability to classify systems by the nature of changes in the input load.

An experimental investigation of smooth matching transitions based on inhomogeneous sections of a microstrip transmission line with different wave impedances profiles was carried out in the present work. Comparative tests of smooth matching transitions prototypes are presented in order to confirm the theoretical data obtained by the optimal parametric synthesis method. The possibility of using synthesized piecewise linear transition to ensure good coordination in a fairly wide frequency range has been proven. It has been shown that fabricated inhomogeneous microstrip waveguide transitions provide matching in the frequency range from 1.2 to 2.7 GHz. It has been proven that the use of two-stage transitions allows a 1.5-times increase in the frequency matching band. Locally-optimized matching transitions near the lower boundary of the operating frequency range can give a significant gain in comparison with matching transitions with an exponential wave impedance profile for the same electrical length. The developed matching transitions can be used in the design of filters, matching transformers and connecting elements of the microwave frequencies transmission line.

It is known that the stroke color of a ballpoint pen changes over time (fades), which means that by the color change of the stroke, you can determine the age of the stroke, and therefore the authenticity of the document. It can be noted that the use of objective colorimetry method is widespread during such examinations. In practice, the spectral method is usually used for determining color and chromaticity coordinates. This method has a high measurement accuracy, but unfortunately, it can not be used directly. This fact is related to the aperture of the studied radiation and the entrance slit of the spectral device. Therefore, stroke dye solutions in colorless solvents are used in practice, which causes other difficulties. This paper describes a slightly different method for determining color coordinates, based on the method of computer colorimetry. This method allows to solve the problem.

The article analyzes the process of interaction between the state and society in the context of the development of digital technologies against the background of an emergency situation related to the spread of coronavirus. The article reveals the causes of the crisis of trust in the interaction of social and political institutions that is characteristic of modern society. The most important role of information and communication technologies in the conditions associated with the need for operational management of information flows to prevent political destabilization is shown. The author substantiates the idea that in the face of the threat of a pandemic, society is ready to review some of the guidelines related to the protection of civil liberties and accept new restrictions in exchange for a more effective state policy to ensure the safety of life and health.