Software & Systems

Articles of journal № 2 at 2021 year.

Order result by: Public date | Title | Authors

11. The platform for modeling large-scale movement of objects and subjects in an urban environment [№2 за 2021 год]
Authors: D.S. Parygin , S.S. Burov , A.O. Anokhin , A.G. Finogeev , A.V. Golubev
Visitors: 2496
Decision-making on the development of infrastructure and the resource allocation to optimize living conditions in developing cities is based on the study of statistical data, opinion polls, specialized sen-sors-counters, video cameras, and other methods of accounting for the people movement. Models and situation projected growth are based on this data. However, carrying out such studies and their subse-quent analysis are extremely expensive and time-consuming projects, therefore they are not carried out regularly and may inadequately represent the actual situation. As an alternative to the monitoring data using, this study proposes to use a multi-agent model. The paper discusses existing solutions for modeling processes in systems with human interaction. Implementing the platform for modeling the people’s movement in a city is proposed. The key problem of the model is to simulate the behavior of any complexity for various object categories, up to the ob-servance of the overwhelming majority of traffic rules. The paper describes the process of building the architecture of an application that will allow model-ing the movement of individual people in the city map, provides specific project requirements and fea-tures of implementing the multi-agent systems. The exploratory studies in the development of the dis-placement model, the approach to the selection of the grid step (1, 5, 30 pixels) for constructing the route graph, and the rationale for the balance of realism and performance are described separately. An approach to visualizing the work of a model on an online map of a web / mobile client is consid-ered. The characteristics of the applied software solutions are given. Conclusions are made about the prospects for the development of the model of movement in the urban environment and the platform for multi-agent modeling in general.

12. Application of the neural network model to support acceptance decisions of the applicant by choice of specialty [№2 за 2021 год]
Authors: Zubkova, T.M., Tagirova, L.F.
Visitors: 3295
The problem of the admissions office is to advise applicants in existing specialties upon admission. Af-ter a brief conversation and document analysis, the employee can offer specialties for which the appli-cant is best to apply. However, the consultation of an employee of the admissions committee is often subjective. The paper presents the developed automated information system (AIS), which allows, based on the questionnaire of the applicant, to form recommendations when choosing a major. Using the neural net-works is proposed as a method for selection. Proposed is a model of neural network – multilayer per-ceptron. Perceptron is to be trained, i.e. a change in synaptic weights is required. We trained the neural net-work on data vectors of the best graduates or students. Neural network training occurs once before the start of the admissions office, after which the values of the synaptic scales are preserved for subse-quent use. After the applicant submits the education document and the completed questionnaire, in-formation is entered into the database (DB) and the classification algorithm is started. Each output of the neural network corresponds to a separate specialty. The neural network works with analog data. To determine the class number to which the input vec-tor of the applicant data should be assigned, they used the principle of greatest response, that is, the largest value of the outputs corresponds to the recognized class. The recommended specialty is the one that corresponds to the output with the highest response value. Introducing the AIS will allow the applicant to provide intellectual support when deciding on pro-fessional self-determination, as well as facilitate the employee’s labor of the education institution admissions office.

13. Threat projection to the future in complex distributed systems based on the mining of big data and automated monitoring tools [№2 за 2021 год]
Authors: E.V. Palchevsky, O.I. Khristodulo, S.V. Pavlov
Visitors: 3193
In connection with the emergence of new technical possibilities for automatic measurement of the pa-rameters of the state of the external environment (including water levels), a method is proposed for predicting a flood situation in complex distributed systems for which there is a threat of material dam-age, using the recovered data from automatic stations based on stationary hydrological posts for moni-toring water levels. The relevance of the selected research topic is substantiated from the point of view of the applica-tion of the recovered data at the automatic stations for control and monitoring of water levels corre-sponding to the given condition for predicting a flood situation. Based on this, a mathematical formula-tion of the problem was formulated (based on short-term forecasting of water levels), within the framework of which an algorithm for searching for automatic stations and interpolation (restoration) of historical values of water levels was implemented to predict water levels in complex distributed sys-tems. The analysis of the effectiveness of the implemented threat forecasting method in complex distrib-uted systems as one module of an artificial neural network (as an example, the result is shown at the automatic station Bulgakovo located between the stationary hydrological stations Lyakhovo and Okhlebinino), according to the results of which the water level when forecasting for one day at the automatic station varied from 7 to 53 cm. Thus, as part of an artificial neural network, this method allows predicting water levels with ac-ceptable accuracy to predict a flood situation (for example, the 2020 flood in the Republic of Bashkor-tostan), which allows special services to carry out specialized measures to counter this threat.

14. Software solution for modeling of a phased antenna array radiation pattern and coding of phase shifter parameters [№2 за 2021 год]
Authors: T.V. Kalinin, D.V. Bogomaz , D.V. Khrestinin, A.V. Vikhrev
Visitors: 3333
For detection and tracking of space objects, it’s necessary to control the position of transmitting and receiving radiation patterns (RP) within a section of the radar coverage area. In modern radar stations (RS) with phased antenna array (PAA), positioning of RP is performed by phase method of determining azimuth and elevation angles. For the phase method implementation, the phase of every PAA emitter must be determined. The paper presents a mathematical model and software implementation of phases and digital control code calculation for every phase shifter. The software solution allows for researching PAA RP accord-ing to the beam direction, the number of emitters, and phase shifter discreteness. The software solution is an application with a graphical interface, written in the C++ programming language in Qt Creator programming environment. Analysis of existing RP PAA simulation software systems (Sensor Array Analyzer package from Matlab 2013 and above, Mathcad, CST MICROWAVE STUDIO, ANSYS HFSS) showed that they do not have the possibility of changing the discreteness of phase shifters, calculating phase codes of phase shifters and visual representation of errors in setting the direction. Thus, the novelty of the software model lies in the visual representation of the theoretical radiation pattern and the radiation pattern, considering the errors that arise because of the discreteness of the phase shifters, which allows it to be used for training specialists in the phased array antenna.

15. Developing rules for detecting hidden data in PDF documents [№2 за 2021 год]
Authors: A.V. Sergeev , Khorev P.B.
Visitors: 2519
The paper discusses the definition and implementation of rules for analyzing and detecting hidden in-formation inside PDF documents. Currently, there is a lot of work devoted to detecting hidden elements in images, sound, and video, but little attention is paid to hiding data in electronic documents of vari-ous formats. A widespread format in modern electronic document flow is PDF (Portable Document Format), which is also of interest for studying the possibility of its use in steganography (the science of methods of hiding confidential messages inside container objects). Documents in PDF format have not been sufficiently researched (especially in Russian literature) from the point of view of the possibility for introducing hidden data into them. Systems for the prevention of data leaks (DLP, Data Leak Protection) are designed to prevent leak-age of confidential information from the information system to the external circuit through various in-formation transmission channels. DLP systems are based on the analysis of data streams crossing the perimeter of the protected information system. When confidential information is detected in this stream, an active component of the system is triggered and the transmission of a message (packet, stream, session) is blocked. Modern DLP systems are unable to detect data hidden in steganographic containers and do not al-low adding the necessary functions to their composition. For corporate networks with high network traffic and high value of transmitted and processed data, means of detecting steganographic attach-ments in the information flow are required. The paper discusses the PDF document format, as well as one of the popular open-source intrusion detection and prevention systems – Suricata, which has pow-erful support for scripting in the Lua language for detecting information security threats. A script in Lua has been developed to detect PDF documents with unwanted attachments as scripts, various ob-jects, and the use of inter-container space with the possibility of obfuscating (obfuscating) the name.

16. The access control system development based on face recognition [№2 за 2021 год]
Author: S.A. Antipova
Visitors: 3441
The basis of this work is the developed access control system based on face recognition. The program comprises four independent components: face detection in a video stream, face recognition based on a convolutional neural network, door/turnstile opening control, and a client web service. Each module works in its own process, so during the development of the project, you can transfer each module to a separate server. The agile development methodology (agile approach) ensures high speed and system quality. The key feature of the work is the substantiated application of deep learning technology on the ex-ample of the created model of the software tool using the technology of one-shot learning or siamese networks, implemented using the PyTorch framework. MobileNetV3 was used as a pre-trained neural network. Siamese networks are convenient for use from the point of view of the absence of the need to form a huge dataset with data, which is especially important for face recognition. The architecture of such networks comprises two identical neural networks with the same weight and structure, and the working results are transferred to one activation function - thus, the similarity of the input data (simi-larity assessment) is determined based on the comparison of the values of two vectors. The enterprise’s point of entry tested the system, considering the biometric data of employees, on which the neural network was trained. The system showed high accuracy in identifying individuals. The proposed service-oriented architecture allows scaling the authentication and verification sys-tem horizontally and vertically. If necessary, system components can be physically placed on different servers, increasing the throughput of the system as a whole.

17. The core design of an integrated information system [№2 за 2021 год]
Authors: Chernysh, B.A., A.S. Kartamyshev
Visitors: 3394
The paper deals with a common problem in the design of corporate information systems, when it is necessary to solve simultaneously two conflicting objectives: the effectiveness of application in one specific subject area and flexible adaptation to the requirements of another area. To solve these prob-lems, there is an approach that consists in forming a software framework or platform that implements generalized functionality with the possibility of flexible configuration under evolutionary require-ments. The platform must comply with the requirements of modern realities, such as flexible service-oriented architecture, versioning, auditing and maintaining the history of changes, the ability to store binary data, flexible management of the life-cycle and business processes. The paper outlines the design concepts of such systems, underlying approaches, and technologies, interfaces of data exchange protocols. The author touches upon the application of modern domain-specific languages on top of these protocols. There are examples of existing domestic and foreign sys-tems, their application areas, advantages, and disadvantages. The paper describes the implementation of a comprehensive information platform based on a sys-tem developed by the authors in the production of technically complex product platforms. There are some applicable solutions and techniques, data structures, and algorithms. A comparative measurement of performing the developed system and one of the commercial platforms confirms the effectiveness of the proposed methods. The resulting assessment is the basis for conclusions about the prospects of the development, its applicability in various subject areas. The author has developed a roadmap for further development and optimization of the platform in the most significant areas, considering the current requirements of the industry.

◄ ← Preview | 1 | 2