Software & Systems

Articles of journal № 4 at 2013 year.

Order result by: Public date | Title | Authors |

11. Application of first-order logic in assumption-based truth maintenance systems [№4 за 2013 год]
Authors: Vagin V.N., Зарецкий Д.С.
Visitors: 7234
The paper is devoted to application of the first-order logic in assumption-based truth maintenance systems (ATMS), performing a function of the reasoning support system. The formal ATMS description and the basic ATMS algorithm are stated briefly. Main shortages of the proposition-based basic ATMS algorithm are shown. Three different ways (ground atoms, valid predicates, unions of ground atoms related to the predicate) of expansion for ATMS application areas to the first-order logic are presented. Strong and weak sides of used methods are illustrated by examples. Necessary adaptations of the basic algorithm to use new possibilities are viewed. Possibility of various applications of considered methods are studied. In conclusion, main weak and strong sides of using the predicates or propositions in truth maintenance systems are summarized.

12. Software pipelining of loops for a floating point accelerator in the Komdiv128-RIO processor [№4 за 2013 год]
Authors: Вьюкова Н.И., Galatenko V.A., Самборский С.В.
Visitors: 10716
The paper presents a method of software pipelining for a specialized floating point accelerator in the Komdiv128-RIO processor with MIPS architecture. Programs for the accelerator may be developed in Assember only. Implementation of a high level languages compiler is impossible due to architectural peculiarities of the accelerator. Manual writing of efficient computational loops for the accelerator is complicated. A programmer should keep in mind miltiple factors, such as lack of hardware delays on unready input data and complicated structure of the register file. The purpose of the work is to develop an instrument for automated software pipelining of loops for the specialized accelerator. A programmer may write quite simple internal loops which implement the calculations correctly but probably inefficiently. The loop is then transforming to a pipelined one with maximal calculating efficiency or close to the maximal one. Use of a precise approach based on integer linear programming (ILP) methods allows achieving the optimal efficiency of a code. The article is focused on specific points of the ILP task formulation connected with peculiarities of the accelerator architecture. The authors discuss the issues of exact calculation of the number of required registers as well as a problem of reducing the unroll factor of pipelined loops.

13. The possibility implementing mobile DVB-RCS standard satellite communication network with the spatial frequency resource time division in the Х-band [№4 за 2013 год]
Authors: A.A. Genov, V.V. Osipov, S.B. Savilkin
Visitors: 10560
Nowadays satellite networks for mobile earth stations are implementing in the spacecrafts on geostationary orbit in direct relay mode with frequency-division multiplexing. The DVB-RCS standard with a spatial frequency TDMA involves placing on board a spacecraft multipath receiving-transmitting antenna and multi-board DVB-RCS standard digital platforms, providing multi-channel demodulation of earth stations sygnals. Using the DVB-RCS standard with MSF-TDMA can significantly extend the functional characteristics of mobile networks with satellites in a geostationary orbit. To use higher data transmition rate that as in DVB-RCS standard satellite communication networks, there is a need to increase the energy potential of satellite radio links "Statsionar" spacecraft. It can be achieved only whem establish multibeam receiver-transmit antennas on-board the "Statsionar" spacecraft. The "Statsionar" spacecraft modernisation is possible with the implementation of Multiservice onboard digital platforms and LAM in the spacecraft. It will allow using DVB-RCS (MSF-TDMA) mobile satellite communications standard in the X-band based on the AP-type "Belozer", which can help to organize high-speed direct connection between AP subscribers.

14. Development of agent behavioral adaptive models for multi-agent system of telecommunication enterprise management [№4 за 2013 год]
Authors: Gimarov V.V., Гимаров В.А., Dli M.I., Ivanova I.V.
Visitors: 13214
The article considers a market as an open socio-economic system involving a large number of interacting elements. Market research and analysis requires processing massive amounts of statistical data and information support to forecast marketing situations in order to select strategies and decision-making for their implementation based on formalized knowledge. At the same time, known methods based on using analytical expressions and games theory, describe the state of the elements statistically excluding changes in their behavior and interaction. The dynamics of the system behavior under conditions of uncertainty, risk, illegible or incomplete information is mostly unpredictable. That is why to make effective management decisions it is appropriate to use the agent-oriented approach to the description and simulation of the system elements interaction. These elements alter the properties and behavior depending on the environment. One of the fundamental problems of multi-agent systems construction is the organization of agents’ interaction. If there are restrictions to use procedural approaches, declarative languages are preferred for interaction of agents. The paper proposes an ontological domain model of agent interaction that defines a set of communicative acts. It also shows a model of "request" act in the notation of colored Petri nets. The adaptability of the system of agents' behavior is achieved due to the classification of indicators of external and internal environment using neural networks and selecting the appropriate behavioral strategies.

15. Application of soft computing for to analyse the knowledge base structure of information system [№4 за 2013 год]
Authors: Гордеев Р.Н., Бурилин А.В.
Visitors: 14377
The article investigates the possibility of using genetic algorithms and neural networks for preliminary analysis and optimization of the structure of the information system knowledge base. It describes an example of constructing the structure of a neural network. The structure of the links between analyzed objects and its clustering representation is described after performing of the analysis. As a result, there is a list of classified objects according to given list of features. The study can be used by public and private organizations involved in the management of scientific and technical complex. For example, in the areas of patents, program and project management, storage and processing scientometric information.

16. Locally defined planning disciplines [№4 за 2013 год]
Author: Грюнталь А.И.
Visitors: 7587
The paper deals with systems, considering as a number of tasks, which are to be executed in a predefined period of time. Planning function points to a particular task of a system which is to be executing in a current moment. Planning discipline is a rule prescribing a specific planning function to a system. Planning function, although it is a global object, could be considered locally, i.e. it could be built based on local conditions but not the whole scope of data, describing all tasks in the past and in the future (according to a current moment). Inherited and locally defined disciplines are under consideration. The paper clarifies corresponding notions in a mathematical manor. Inherited disciplines are those which depend on system state in critical points, i.e. moments when new tasks appear. A theorem establishing coincidence of algorithmic and axiomatic definitions of inherited disciplines is formulated. Locally defined discipline prescribes to a system its planning function in a way that depends on local state of the system in an arbitrary moment of time. A theorem is formulated which determines a structure of an arbitrary locally defined discipline. The theorem states that a locally defined discipline is inherited in relation to a locally defined discipline applied to systems with a single critical point.

17. Preprocessing the video sequence to publicate in digital libraries [№4 за 2013 год]
Authors: Damov M.V., V.V. Buryachenko, Pakhirka A.I., Zotin A.G.
Visitors: 11590
This paper presents methods for increasing the quality of embedded video sequences when preparing digital books and other digital library multimedia resources. They are presented as processing stages: non-linear correction of image brightness, scene replacement detection with preliminary motion estimation in the scene and preliminary quality estimation of the frame, image stabilization with frame borders restoration methods. Non-linear correction of images brightness performs using modified MSR-algorithm which allows increasing the contrast of image details in of high brightness areas. The algorithm of video sequence stabilization is based on block matching method and allows removing frame-to-frame jitter effect. An interpolation of neighboring frames is used when stabilizing. It provides less image content loss on frame borders. It is offered to evaluate the quality of video sequence frame that allows detecting cases of scene changes and high level noise. A description of testing and results of the increase of video sequences quality it presented.

18. Intelligent expert method for determine the optimal route of product transportation [№4 за 2013 год]
Authors: Dli M.I., Gimarov V.V., Глушко С.И., Ivanova I.V.
Visitors: 11498
To reduce transport costs a modern enterprise needs to select the optimal combination of different technical and economic indicators of the product delivery route. Searching rational track for goods movement is the foundation for building an efficient logistic system, which determines the timeliness of production output, production process duration, production costs. The optimal route is often determined by cost and transit time criteria. The combined use of these criteria causes the need for a two-criterion optimization algorithm. An NP class task affiliation is the reason for using heuristic methods (simulated annealing, genetic and ant algorithms, tabu search method, and others). Ant algorithms are based on the modeling of the behavior of individual ants acting according to simple rules, while the total system (colony) has an intellect. Using an ant colony to optimize the route of transporting products with two-criteria requires using two different colonies for each criterion, and the third generalizing colony is for route selection based on two criteria. To consider uncertainty when solving these problems it is appropriate to use fuzzy sets for unspecified parameters of mathematical models and fuzzy inference methods with expert judgment. The ant algorithms peculiarity is providing transport costs information on the site, as well as the weighting factors that determine the degree of algorithm greed in the form of fuzzy numbers and then applying the operations of fuzzy-logic convolution. The developed algorithms can improve the efficiency and reliability of enterprise logistics system.

19. Optimal routing using the network congestion criterion [№4 за 2013 год]
Authors: Dmitriev G.A., Margolis B.I., Muzanna M.M.
Visitors: 11814
The article considers the network information flow model with reroute messages for any number of switching nodes and links. There are criteria for optimizing with time-delay messages and the maximum network congestion, restrictions for the capacity of the channels and the conservation of flows in the network. An example of a network consisting of five nodes and seven channels of communication is described. It is shown that the problem of optimal routing refers to non-linear optimization of several variables function with inequality constraints and equations. Inequality constraints are formulated for channel capacity and the type of equations for the conservation of flows in the network nodes according to their division into input node, output node and intermediate nodes. Based on the standard fmincon function, a program as developed in MatLab. It allows finding the optimal distribution of flows in the network consisting of an arbitrary number of nodes and links according the topological structure of the network, the input flow and capacity of channels. The article shows the results of the optimum flow separation calculation using criteria of congestion and delays for the considered example. The analysis of changes in the criteria of maximum load at different intensities of the input stream is made. It is shown that the determination of the optimum distribution of channels load while separating each channel to any number of lines can be done using analytical optimization techniques. There are results of the calculation of the optimum flow separation on two links by load for the considered example.

20. Data protection in self-organizing information system without using special facilities [№4 за 2013 год]
Authors: Drozhdin V.V., Зинченко Р.Е.
Visitors: 9531
The safe access in self-organizing information system is described. The current state of information security and data protection is shown. To increase a level of safety of self-organizing information system three components are offered: interaction of information system with system of hierarchically connected users, interaction of users with the system via visibility areas, regular users' authentication based on their behavior. Interaction of information system with the system of hierarchically connected users united in a network structure, on the one hand, simplifies system creation and provides control of subordinated users’ activity by higher users. On the other hand, it gives an opportunity to the system to gain the help of higher users in case of a difficult or uncertain situation with the specific user. Using the areas visibility mechanism for all users allows to form the user representation of data domain easily and to delegate powers of higher users to the subordinated users, and the self-organizing information system will provide the correct functioning and the effective data processing according to conceptual data domain model. Regular users' authentication in self-organizing information system based on their behavior will allow to identify each user more precisely and to confer confidential powers more correctly. Thus, the offered approach combines naturally both program (automatic) and organizational means of information system security.

◄ ← Preview | 1 | 2 | 3 | 4 | 5 | 6 | Next → ►