Software & Systems

Articles of journal № 3 at 2012 year.

Order result by: Public date | Title | Authors

31. Parallel algorithms designed for the strength analysis of hydrogen-charged structures [№3 за 2012 год]
Authors: Федотов В.П., Spevak L.F., Nefedova O.A.
Visitors: 9700
The work provides a model for estimation of hydrogen brittleness for flat elements used in metal structures with defects. It is known that defective metal parts the hydrogen concentration is distributed fairly uneven – defect area is surrounded with molecular hydrogen of high concentration under high pressure. This model can be used for analysis of destruction process of a metal structure that is exposed to intensive external action. In this case, internal pressure in defects grows quickly, and this leads to destruction of such elements. The model uses algorithm of coupled diffusion-deformation problem. Coupled problem includes equations that describe hydrogen diffusion process in high field, and mechanic equations for stress and deformation tensors that consider diffusion process. Solution of deformation problem in linear elasticity theory is made with modified boundary element method. Collection of the hydrogen around defect area is described by convective diffusion in the field of high mechanical stress. Diffusion problem can be solved with modified boundary element method. Solution of the coupled problem can be obtained step by step in time. In final stage of calculation, each step produces value of molecular hydrogen pressure in every defect and then destruction condition of boundaries is checked. For reduction of computing time, parallelizing procedure was performed. The program was designed with this algorithm. This program can be implemented on multiprocessor computing system. This algorithm was illustrated by solution of two-dimensional problem of hydrogen diffusion around the defect – pores for the metal sample exposed to external tension stress. Calculation outputs are shown in charts.

32. Tablet computer for outfit of the industrial enterprise workplaces [№3 за 2012 год]
Author: Матвеев А.И.
Visitors: 8987
There are two variants of the development of a portable computer engine operated on the basis of the motherboard as a desktop computer or as a portable computer (laptop). In consequence of the devices’ characteristics analysis, which is made on the data obtained during assembly, the choice is made in favor of the computer engine, working on the basis of the motherboard of a desktop computer. Created Tablet Computer is a mobile and is powered by a rechargeable battery. It causes to changes in all motherboard power system and eliminates to use the standard power unit. Power supply of the created Tablet Computer is based on the PWM generator. This method is easy to implement because of the large variety of MAXIM's chips and ISL (possibility of the programmatic control of the voltage level), the long hardware component base as of PWM-generating chips as well as several field transistors in a single package. The matrix (display) of the portable computer (laptop) is also used; the image is transmitted to the matrix through the DVI-output. The solution of a number of technical problems ensures compliance with the required functional characteristics and cost performance, which allows you to equip the required number of concerned industrial enterprise workplaces.

33. Increase of storage coefficient of computing complex efficiency when using virtualization [№3 за 2012 год]
Author: Тетюшев А.В.
Visitors: 10872
The article gives a mathematical description of the effect of increased storage coefficient of computing system efficiency using virtualization tools if certain conditions are performed. These conditions are determined by the efficiency win of the virtual system compared to the physical system at the stage of data processing. It is the efficiency win of the virtual system (service) for certain computational tasks that can be used as a weighting factor that determines the reliability of the whole computing complex. To prove theoretical research has been made an experiment, which included the local area network to the physical server and several workstations. On the physical server a virtual server was raised. During the experiment, the servers reset so that the virtual system exceeded the physical one in the data processing phase. Measuring the efficiency of servers was performed using software agents Ziff-Davis «NetBench». The result of the experiment demonstrated the feasibility of the conditions under which the efficiency of the virtual system can exceed the physical one in the data processing phase. In this case the total reliability of the system is increased without additional funds. The value of this research is that it is possible to develop highly reliable computing systems based on the virtual servers (services), comparable in reliability to existing, but with a much lower cost.

34. Application of parallel programming package Graphplus templet in modeling [№3 за 2012 год]
Authors: Востокин С.В., Литвинов В.Г., Хайрутдинов А.Р.
Visitors: 9673
Software implementation of parallel programming tools Graphplus templet is represented, which automates the development of parallel programs for multiprocessor work stations and supercomputers. This development has been carried out at the department of information systems and technologies of Samara State Aerospace University within the research project «Graph Plus» (graphplus.ssau.ru). The article develops the approach, which makes it possible to expand and simplify the use of high performance computing hardware in numerical simulations. The key concepts of this approach are the following: application of generic solutions (patterns) of parallel programming; automatic parallelizing and deployment of code on different software and harware architectures; allocation of roles of system programmer – application programmer; usage of integrated development environments and code visualization techniques. The approach is based on the authorial methodology of parallel processes design – TEMPLET-methodology. The structure of the new version of software complex of parallel programming automation Graphplus templet is described. The detailed review of functioning principle of a program graphic model translator, input and output data of the translator, its integration with MS Visual Studio environment is given. The usage of software complex for solution of problems of non linear dynamics and computer optics is shown. Efficiency factors for the programs generated in it in the course of executing on multiprocessor work stations run under MS Windows and on the supercomputer «Sergei Korolev» of Samara State Aerospace University operating under Linux are provided.

35. Software reliability analysis based on the model of inhomogeneous poisson process and bootstrap methods [№3 за 2012 год]
Authors: Guda A.N., Chubeyko S.V.
Visitors: 7178
A new mathematical model of software reliability is described, based on mathematical model of inhomogeneous Poisson process. The basic idea of the proposed in the article forecasting method is a method of reproduction of data samplings, containing two original sets: the cumulative program execution time and number of errors committed during this time. As a method of a randomized sampling reproduction a bootstrap technology is taking which uses random quantities, having a Poisson distribution. Algorithms of parameter estimate and forecasting indicators of software reliability are suggested. The first algorithm is used to assess the intensity of errors expected in subsequent versions of the software. The algorithm uses a random number sensor on which basis randomized sampling and random arrays are arranged under Poisson law. The second algorithm makes it possible to evaluate the intensity of error detection. He uses data samples from the first algorithm and operates according to the method of maximum likelihood. The article describes the general procedure for forecasting the expected number of errors that can occur during a subsequent program run at a certain time interval, following a cumulative period of observation. The proposed forecasting method was realized in the form of a program written in Pascal programming language in the free programming environment PascalABC.NET. It also describes examples of using forecasting software with some test data.

36. Program system for visual languages analysis [№3 за 2012 год]
Authors: Шаров О.Г., Афанасьев А.Н., Гайнулин Р.Ф.
Visitors: 6639
The article describes client-server architecture for analysis of consistency of graphic works in the design process of complex automated systems. The system is represented by multilayer structure. Presentation layer is a tool fo creation of diagrams in a visual language (UML, IDEF, eEPC etc) included in design technique software package (RUP, ARIS etc), or any other editor, which can be expanded with plug-ins. Logic layer is designed for data exchange with plug-in. Description of diagram is transferred during sending procedure in XML format. When error list is received in XML format. The server is accessed through public API-interface. Data processing layer contains multipurpose visual language analyzer of the project works, this analyzer operates on the basis of graphic finite state, which discover and eliminate errors within one pass of diagram analysis. Syntax and semantic errors are controlled in analyzed graphic specifications of the work flow. Analyzer has advantage in the form of linear analysis time, full control and visual error presentation. The article shows grammar example of UML-diagram analysis. The system is implemented in Python environment at Tornado server. The work was made in Computing Technics Department, Ulyanovsk State Technical University.

37. Program based traffic generator of the virtual lab resource users [№3 за 2012 год]
Authors: Жевнерчук Д.В., Николаев А.В.
Visitors: 11938
In design process of cloud calculations it becomes necessary to model query stream. Query models provide estimation of server hardware resources for the maintenance of client systems under operation or peak load. The work describes software and hardware for the study of processes happening in cloud environment of simulation modeling. The system is implemented on .NET Framework 2.0 platform. It was designed with MS Visual Studio 2008, GPSS World, MySQL 5. The work provides functional model of this system, ER model. The model implements preparation of the model pattern and generation of the final code. Series experiments was made in order to set up traffic generation subsystem; this was done in university laboratory of simulation modeling using GPSS World, the access to this system was provided with Cloud Computing technology. Time metering was performed with GPSS shell, time meter 1.0, which maintains user’s log. Special attention was paid to interaction of a user with the server. Using obtained consistent patterns they built empiric GPSS functions, which describe user’s interaction with the system. The work provides taxonomy of interaction modes with simulation models; based on experiments, it defines basic types of client queries; the work shows users’ interaction models with simulation model environment; the work offers an architecture of model environment for network traffic. Obtained models can be used for estimation of the server hardware resource; this helps to take into account peak load and idle time. Traffic generators that use empiric models can be used for the study of load algorithms of distributed environment for simulation modeling.

38. Parallel design algorithm for description of the mass transfer in a pore [№3 за 2012 год]
Authors: Поветкин А.Д., Чан Хыу Куе, Koltsova E.M.
Visitors: 9239
Here is described a model using molecular dynamics approach that describes gas transfer in small pores. Two types of the mass transfer that happen in a pore were taken into account: Knudsen diffusion and molecular diffusion. Two cases of interaction are mentioned that describe molecular movement and interaction. First case describes the variant, where molecules move according to classic mechanics law and where they interact with each other and a wall, on the principle of perfectly elastic collision. It is assumed that the particles can collide with the pore’s wall in two ways: mirror and diffusion, and for each specific collision this way is defined at random fashion, and the ratio of collisions according to both ways presents one of the features of the model. The second case describes molecular interaction using intermolecular potential of Lennard-Jones interaction. Calculation of molecular movement was made with parallel algorithms. Parallel computing is made with the use of CUDA technology. Information about particle features is stored in 3-D array. Number of array cells corresponds to number of particles in the system. Each array cell contains information about particle: coordinates, velocity vector, speed up vector and its type. One dimension array of particle type exists separately. Number of array elements equals to number of types of existing particles. Cells contain information about each substance that is present in the system: particle’s mass and radius, coefficient of molecular interaction potential. The array was arranged with bubble sort algorithm adapted for parallel operation and extended for 3-D case. The article contains basic computation flow charts.

39. Analysis tools of Linux operating system drivers development [№3 за 2012 год]
Authors: Korablin Yu.P., Павлов Е.Г.
Visitors: 8768
The article describes the tool for search of synchronization problems in Linux operating system drivers that uses the detection of error situations by using semantic models. This method is based on theory, described in the work [2], it enables to detect such errors of synchronization as loops, deadlocks, double locking and others. Tool for search of synchronization errors consists of four main parts: the preprocessor, the compiler of driver code to program in the language of asynchronous functional schemes (AFS), the presentation of the AFS program as a system of re-cursive equations, analyzer of recursive equations system. The preprocessor is implemented as a script in the bash language, its purpose is to prepare the driver code to translate into the AFS language. After all the preparatory procedures the preprocessor passes control to the translator into the AFS language. The translator is written using the GNU Flex lexical analyzer generators and GNU Bison syntax analyzers. The driver is translated in the AFS language, based on the specified in this article operational semantics for synchronization objects. The next step is to convert the AFS language programs into the recursive equations system. On the basis of the theory of [3] the program in the AFS language is presented as a system of recursive equations, which sets a lot of computing sequences associated with the driver structure. In order to search synchronization errors there is implemented a tool for analyze of recursive equations system received. The tool has a modular architecture, so it can be adapted for different types of software. In addition, the possibility of applying the method of [2] for races state detection is being currently studied.

40. Task allocation on the basis of auction model in the electronic document flow systems [№3 за 2012 год]
Author: Леднев А.М.
Visitors: 9044
The article describes some approaches to building of problem and business process management system on the auction principle between the employees of the enterprise in P2P network with regard to their specific peculiarities and preferences. It studies the matrix structure of company management according to which solutions are made collaboratively under the results of agreed interactions of employees. To take into consideration the experience, skills, competence and preferences of the company employees in the document flow systems it is proposed to use employee competence knowledge bases, which can be formalized by means of ontologies. In the problems of research and building of enterprise personnel interaction management systems it is proposed to use the model of interaction in multiactor integrated information environment. To study this approach a special virtual environment is proposed which contains P2P network of active programs simulating behavior of specialists and managers. The auction model, where each problem acts as an auction lot and specialists – as participants of the auction, is considered as one of the models of the described problem distribution between specialists. Application of this model can be useful to solve the problem of introduction of electronic document flow systems for the purpose of provision of problem allocation and decision making support in real time mode.

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