Eng | Rus

“Systems and Means of Informatics” scientific journal

Volume 32, Issue 2, 2022

Content   Abstract and Keywords   About Authors

ISSUES OF INCREASING THE NETWORK BANDWIDTH AND TERRITORIAL ACCESSIBILITY OF NATIONAL RESEARCH COMPUTER NETWORK OF RUSSIA
  • A. G. Abramov
  • A. V. Evseev
  • A. A. Gonchar
  • B. M. Shabanov

Abstract: The paper is devoted to the discussion of plans for the development of the National Research Computer Network of Russia (NIKS) for the period 20212024 within the framework of the National Project "Science and Universities."
The emphasis in the work is made on the current regulatory status of the NIKS project, significant aspects of the management procedure developed by the network administrator and approved by the line Ministry, including goals, objectives, and expected results in the development process. Special attention is paid to the issues of modernizing the backbone telecommunications infrastructure of NIKS and expanding the territorial accessibility of the network within Russia.

Keywords: National Research Computer Network; NIKS; National Research and Educational Network; regulatory legal status; telecommunications infrastructure; regional expansion

MACHINE LEARNING FOR EXTRAPOLATION PROBLEMS WITH SMALL DATASET
  • A. O. Belozerov
  • A. I. Mazur

Abstract: A new method of extrapolation of variational calculations which is based on training a large number of artificial neural networks with subsequent filtering to select the best networks is proposed. This method is used to evaluate ground state energy for model problem and to calculate the ground state energy of a 4He nucleus based on calculations in no-core shell model with realistic potential Daejeon16. The results convergence with increasing number of input data was studied. It is shown that the method allows one to achieve a sufficiently high prediction accuracy even in the case of small model spaces. The method can be applied both in finding various characteristics of nuclei and in solving other problems not related to nuclear physics.

Keywords: machine learning; extrapolation methods; ground state energy; nuclear shell model

CONSTRUCTING OF THE BRIEF REACHABILITY TREE FOR PROGRAM MODELS IN TERMS OF PETRI NETS
  • D. V. Leontyev
  • D. I. Kharitonov

Abstract: The article deals with the problem of building a state space for analyzing the imperative programs behavior. The state explosion problem of analyzed states number in the automatic program models construction is the main problem for finding errors in the programs source code. This explosion is induced mainly due to the composition of the program variables states. The article proposes an approach to reducing the number of states of the reachability tree of program models by separating the program control flow model from the variable models and then adding only variables that affect the control flow and reducing the states of these variables. The example considered in the article shows how such an approach can be applied in practice.

Keywords: Petri nets; reachability tree; checking programs correctness; modeling program behavior

ON COMBINING THE SOFTWARE-DEFINED NETWORKING WITH THE NETWORK FUNCTIONS VIRTUALIZATION
  • V. B. Egorov

Abstract: Many firms delivering software to the networking market offer products for the software-defined networking (SDN) and network functions virtualization (NFV). In advertising, the frequent neighborhood of both acronyms, up to their merging into SDN/NFV, creates the impression that the technologies behind them are closely related, if not the same at all. The ascertainment of the actual interrelation between SDN and NFV is complicated by difficulties with understanding the objects themselves. In publications on the subject, some common statements have entrenched, many of them turning out to be dubious at a captious exploration. The article impugns the assumed in various publications affinity of both technologies. They originated independently developing in diverse directions and at different rates: NFV is actively demanded in provider networks, while SDN has been implemented mainly by some big internet companies, with the interest in the technology not growing last years because of a number of uncertainties. Actually, both acronyms, SDN and NFV, gravitate towards each other, even up to sticking together into SDN/NFV, basically in advertisements of networking software suppliers for purely marketing reasons.

Keywords: network functions virtualization; NFV; SDN; software-defined networking

ON THE USE OF EXPERT METHODS IN EVALUATING EFFECTIVENESS AND QUALITY OF INFORMATION SYSTEMS
  • A. A. Zatsarinny
  • Yu. S. Ionenkov

Abstract: The article is devoted to the application of expert methods in assessing effectiveness and quality of information systems. The general issues of the application of expert assessments, including the conduct of the expert survey procedure and the establishment of consistency of expert opinions, are considered. Various approaches to determining the number of experts, including formulas for calculations, are presented and comparative calculations based on these formulas are carried out. A formalized approach to assessing the competence of experts is presented, including their scientific qualifications, reasonableness of assessments, and the degree of familiarity with the problem being evaluated.

Keywords: expert methods; efficiency; quality; number of experts; qualification of experts; information system

ANALYTICAL MODELING AND ESTIMATION OF NONSTATIONARY NORMAL PROCESSORS WITH UNSOLVED DERIVATIVES
  • I. N. Sinitsyn

Abstract: Methodological and algorithmic support for analytical modeling, estimation, identification, and calibration for essentially nonstationary (e.g., shock) stochastic systems with unsolved derivatives is proposed. Basic theorems are given. Special attention is paid to shock disturbances. An example is provided. Some possible generalizations are presented.

Keywords: calibration; estimation (filtering and extrapolation); identification; Kalman and Bucy filter and estimator; shock system with unsolved derivatives; system with stochastically unsolved derivatives

SOME APPROACHES TO NETWORK DLP ANALYSIS
  • A. A. Grusho
  • N. A. Grusho
  • M. I. Zabezhailo
  • D. V. Smirnov
  • E. E. Tmonina
  • S. Ya. Shorgin

Abstract: The work deals mainly with network DLP (Data Lost/Leakage Prevention) systems. One of the main tasks of DLP systems is to recognize the appearance of valuable information that can appear on the network at the output of the computer system. Network-based DLP systems are becoming more important due to the development of remote work of employees of various organizations. The proper functioning of the DLP is determined by recognizing short messages as random sequences and leaking valuable information such as personal data as a short sequence satisfying structural constraints. The main problem of these methods is false alarms. In the work, in the simplest models, the volume of data is estimated in which it is possible to search for valuable information at small probabilities of false alarms. The examples of simple but fast methods for detecting leaks of valuable information are built. The methods are divided by complexity and their applications. Such methods can be used not only to analyze DLP systems, but also to search for service inserts and signals during the transfer of information when searching and analyzing poorly protected or unprotected personal data where depersonalization is required according to the rules. The estimates of the probabilities of false alarms are built.

Keywords: information security; DLP systems; estimates of probabilities of false alarms

ARCHITECTURES OF INTELLIGENT AGENTS OF COHESIVE HYBRID INTELLIGENT MULTIAGENT SYSTEMS
  • S. V. Listopad
  • I. A. Kirikov

Abstract: The paper develops research in the field of modeling the cohesion of a team of specialists who solve problems "at a round table" with hybrid intelligent multiagent systems that combine the dynamic synthesis of heterogeneous models of intelligent agents and the methods of group dynamics. The formation of the cohesion of agents of these systems is provided by such mechanisms as exchange of information about interests, knowledge, experience with other agents, development and coordination of goals, values, norms, means, and methods of activity common for the system which is relevant to the processes taking place in long-term teams of specialists who jointly solve practical problems.
The implementation of such mechanisms will significantly simplify the process of integrating heterogeneous intelligent agents into a single system which is especially important when they are created by independent development teams.
The paper proposes the architectures of agents necessary for the implementation of the listed mechanisms.

Keywords: cohesion; hybrid intelligent multiagent system; team of specialists

STUDY OF METHODS FOR DETECTING KEY POINTS WHEN CREATING PANORAMIC IMAGES
  • P. O. Arkhipov
  • A. K. Trofimenkov
  • M. V. Tsukanov
  • N. Yu. Nosova

Abstract: The article is devoted to the problem of creating panoramas in conditions of differences in the color palette and illumination level of stitched images caused both by the features of the optical system and by the weather and other factors. A procedure for correcting brightness and color is proposed to minimize differences between stitched images and to reduce negative impacts.
The features of detection of key points by various methods during the stitching of panoramic images under conditions of shooting at different heights are presented.
The effect of an additional procedure for correcting brightness and color on the number of found key points is described. The quality of the final panoramic images and the time of their stitching were evaluated. In the course of the work, statistical data were obtained and their comparative analysis was carried out.

Keywords: brightness correction; color correction; detection; key points; panoramic images; image processing

DETERMINATION OF THE REQUIRED NUMBER OF OPTICAL DISKS ORGANIZED INTO HYBRID STRUCTURES FOR THE PURPOSE OF DEVELOPMENT AND MAINTENANCE OF LONG-TERM ELECTRONIC ARCHIVES
  • A. V. Chernyshov

Abstract: Hybrid structures of optical disks are a method of archival storage of information with increased reliability, similar to RAID arrays of disks. Unlike classic RAID arrays in hybrid structures, one of the copies of information is stored on single disks that provides quick and easy access to information using just one standard drive. During long-term storage, optical disks with recording can degrade. Knowing the number of optical disks that will need to be purchased yearly to create an archive and then to maintain its integrity is necessary for planning the life cycle of the archive and the costs of its maintenance at the design stage. The paper presents analytical models that allow one to determine the required number of optical disks yearly for four possible options (strategies) for filling the archive as well as time points that characterize the process of filling the archive and to assess the principal achievability of the planned archive capacity. An example of a real calculation is given.

Keywords: long-term electronic archives of information; write once optical disks (WORM); hybrid structures of optical disks; replacement of failed optical disks; yearly required number of optical disks

A MODEL OF DISCOVERING NOVEL TERMS AND SENTIMENTS IN TEXTS
  • I. M. Zatsman

Abstract: The paper considers a model focused on the design of information technology and systems that ensure discovering novel terms in the process of computer-human analysis of texts (hereinafter referred to as the information technology-oriented model, or the ITO model). A modification of the ITO model is proposed to describe discovering novel terms in scientific documents and sentiments in the messages of social network users. Consideration of these two models is presented in the paper as a step-by-step transformation of the spiral model of new knowledge generation. The main purpose of the paper is to describe the step-by-step transformation which is considered by the example of discovering novel terms in the texts of scientific documents and sentiments in the messages of social network users.

Keywords: spiral model; novel terms; ITO model; sentiments; ITO model modification; spiral model transformation

AN APPROACH TO THE DEVELOPMENT OF A CONCEPTUAL SCHEME OF A GEODATA DATABASE WITH SUPPORT FOR A MULTIVARIATE REPRESENTATION OF GEOOBJECTS
  • D. A. Nikishin

Abstract: The structure of the conceptual scheme and the main design solutions for constructing a variant of the geodata database that provides support for a consistent multiscale and multivariate representation of geo objects are presented.
The essence of the variation of geomodels is considered and the typology of relationships between variant representations of geodata is highlighted. Combining the options for representing geodata in the context of a common geodatabase will allow for direct access to them and, on this basis, implement new methods for solving problems of data analysis and processing as well as ensure synchronization and coordination of changes in geodata.

Keywords: geodata; geodata databases; conceptual scheme of the geodata database; modeling of geo objects; variation of geomodels

APPROACH TO RECORD LINKING IN TECHNOLOGY OF CONCRETE HISTORICAL INVESTIGATION SUPPORT BASED ON FUZZY SETS
  • I. M. Adamovich
  • O. I. Volkov

Abstract: The article is devoted to the further development of the distributed technology of concrete historical investigation support based on the principles of crowdsourcing and focused on a wide range of users which are nonprofessional historians and biographers. The development is carried out by modifying the algorithm for automatic record linking when processing nominative documents of family structure based on the Family reconstitution method, successfully used in historical demography, and the structure of family relationship graphs. The modification consists in relying on the theory of fuzzy sets and using the similarity measures to make a decision about records linking in the case when the algorithms of the family reconstitution method do not allow an unambiguous conclusion about the linkage. The article describes the methodology for constructing membership functions and substantiates the use of a similarity measure based on the Lukasiewicz and Zade similarity measures.

Keywords: concrete historical investigation; distributed technology; record linking; sources of family structure; fuzzy sets

S-MODELING AND DIGITAL TECHNOLOGIES OF ECONOMIC ACTIVITY
  • V. D. Ilyin

Abstract: The application of the theory of symbolic modeling of arbitrary objects (s-modeling) to identify economic processes in solving problems of managing them in a digital environment is considered. S-models of financial, trade, and other processes of economic activity are represented by a set of self-learning digital twins. To construct a digital twin, the values of parameters characterizing the behavior of the simulated object are used. Since its creation, digital twin has been accumulating knowledge about the simulated object, exchanging messages with it and updating. Digital technology of economic activity of a certain type is considered as a set of methods, means, and rules for the implementation of the processes of this activity through online services operating in the environment of digital twins. The examples of updated technologies of the normalized economic mechanism are given.

Keywords: symbolic modeling; s-modeling; digital technology; economic process; normalized economic mechanism

COMPARATIVE ANALYSIS OF MULTIPATH TRANSPORT PROTOCOLS
  • E. P. Stepanov
  • R. L. Smeliansky

Abstract: The paper deals with the problem of ensuring the network service quality by using multipath routing protocols by splitting the application data flow to several subflows, each of which is transmitted through its own route in the network. There are two classes of multipath protocols: the static one with fixed number of subflows during the connection lifetime and the dynamic one which varies the number of subflows depending on the current network state and application requirements. The paper presents the comparative analysis of the effectiveness of static and dynamic multipath transport protocols. The comparison methodology is invariant to the network topologies, load, and transport agent settings. The areas of static and dynamic multipath protocol efficiency are derived based on study results. The experimental study has shown the advantage of a dynamic approach.

Keywords: MPTCP; FDMP; QOS; ISP









 

RUS