ABSTRACTS OF ARTICLES OF THE JOURNAL "INFORMATION TECHNOLOGIES".
No. 4. Vol. 23. 2017

To the contents

A. F. Valeeva, D. Sc., Professor, aida_val2004@mail.ru, Ju. A. Goncharova, Postgraduate,
Ufa State Aviation Technical University

The Development of the Logistic Transport System

The operational planning problems solving of homogeneous goods delivery to different customers such as inventory management and looking for rational routes of goods delivery are considered in this article. The algorithm for searching the best inventory management strategy and a modified population based ant colony optimization algorithm (P-ACO) for getting the rational routes of homogeneous goods delivery to different customers are presented. Some numerical experiments on random generated data, taking into account such restrictions as vehicle capacity, time windows, split delivery, delivery period, multi depot, pickup and delivery, and tested examples, taken from International OR-library tests, are resulted. Numerical experiments results confirm the effectiveness of the developed algorithms based on the P-ACO algorithm. The problem of rational goods packing in vehicles (the problem of three-dimensional packing) solved in the constructing routes process.
Keywords: inventory management, routing, goods packing in vehicles, population based ant colony optimization algorithm

P. 243–250

N. P. Demenkov, Ph. D., Associate Professor, e-mail: dnp@bmstu.ru, E. A. Mikrin, D. Sc., Professor, e-mail: eugeny.mikrin@bmstu.ru, I. A. Mochalov, D. Sc., Professor, e-mail: intelsyst@mail.ru, Department of Automatic Control Systems Bauman Moscow State Technical University

Fuzzy Transformation of Laplace in Tasks of Fuzzy Mathematical Modelling. Part 1

Article is devoted to a class of models of linear and nonlinear control systems which are represented by traditional mathematical models in their fuzzy realization that provides them adequacy at different impact of indignations.
Linear and nonlinear objects of control, sensors, executive mechanisms and other components of control systems in the conditions of uncertainty lead to various types of fuzzy models: an fuzzy initial task for the linear and nonlinear differential equations with constant and variable coefficients; fuzzy one — point and two — point boundary tasks for the differential equations of the first and second orders; the fuzzy integrated equations of Voltaire like convolution of the 1 and 11 sort; the fuzzy equations in private derivatives of the 2nd order; fuzzy integro-differential equations and others.
The purpose of the real work consists in synthesis of representations of fuzzy transformation of Laplace, its properties and its application for the solution of the listed above various fuzzy linear and nonlinear tasks.
Definitions of fuzzy direct and return transformation of Laplace are given and are specified signs of his difference from traditional transformation. Types and number of the fuzzy differential equations are defined and which are solved by application of fuzzy transformation of Laplace.
The solution of these equations is closely connected with representation of a derivative in fuzzy interpretation.
The generalized fuzzy derivatives of Hukuhara are defined of various types: the fuzzy generalized type gH derivative Hukuhara, the indistinct strong generalized derivatives of gH(i) — gH(ii) of types of Hukuhara. The theorem of differentiation is formulated for fuzzy derivative of the first order.
Application of fuzzy transformation of Laplace is shown for the solution of various fuzzy linear and nonlinear tasks.
Keywords: fuzzy boundary task, fuzzy transformation of Laplace, fuzzy direct and return transformation of Laplace, derivative on Hukuhara

P. 251–258

G. S. Veresnikov, Senior Researcher, e-mail: veresnikov@mail.ru, L. A. Pankova, Senior Researcher, e-mail: ludmila_pankova@bk.ru,
V. A. Pronina, Senior Researcher, e-mail: pron@ipu.ru,
Institute of Control Sciences V. A. Trapeznikov Academy of Sciences, Moscow

Use of Uncertain Programming in Design

The paper is devoted to application of the uncertainty theory offered by B. Liu in design. The generalized model of uncertain multiobjective programming is considered. The use of the model is demonstrated on the two-bar truss design optimization.
Keywords: uncertainty theory, uncertain programming, uncertainty measure, uncertain variable, epistemic uncertainty, expected value of uncertain variable, variance of uncertain variable, critical values of uncertain variable, generalized model of uncertain multiobjective programming, Pareto-front

P. 258–265

V. I. Gorbachenko, D. Sc., Professor, Head of Chair, e-mail: gorvi@mail.ru, S. N. Katkov, Senior Lecturer, e-mail: skat.pnz@yandex.ru, Penza State University

Memristor Cellular Network for Solving Boundary Value Problems

Before the advent of high-performance digital computers primary mean of solving boundary value problems described by differential equations in partial derivatives (PDEs), was some analog simulators especially the network model. The advantages of network models are high concurrency, low sensitivity to failures of individual elements of the simulation environment and the possibility of solving ill-conditioned problems. The main drawback of the network model is the lack of solution accuracy.
Currently in the field of solving boundary value problems in a certain degree of network models "successors" are linear analog cellular network — a type of cellular neural networks.
The purpose of this paper is to analyze the possibilities of solving boundary value problems in the analog cellular networks based on new elements — memristor. The ways are offered to overcome the main shortcoming of analog models — the lack of solution accuracy.
There's given the analysis of linear analog cellular network with a discrete time representation. It's the seen connection between the mathematical description of the cellular network and the system of difference equations approximating the implicit scheme for the boundary value problem. There are presented schemes of analog cellular network cells.
It's shown the possibility of realization the cellular communication network weights using memristor. Application memristor will reduce the time to solve the problem by reducing conductivities setup time, and due to the short duration of the transition process in nanoelectronic circuits. Computing system architecture comprising memristor cellular network must be a two-layer: the first layer is a memristor network, the second layer is the system of parallel processors and data preparation processing results.
To compensate the analog cellular network error it's proposed the additive correction algorithm. In this algorithm on the net it's just formed an amendment to the decision and the decision itself is calculated in the digital part of the system that provides the accuracy of the solution, in general only limited with opportunities of the digital system. We prove the convergence and stability of the algorithm.
Keywords: boundary value problem, partial differential equation, analog cellular network, memristor, additive correction algorithm, convergence, stability

P. 266–273

A. N. Rodionov, D. Sc., Computer Centre of Far-Eastern Branch of RAS, Khabarovsk, ran@newmail.ru

Some Syntactic and Semantic Limits on Scenary Business Process Models

The main goal of this paper is to obtain the simple and descriptive method for base workflows verification on soundness. This method can be used during immediate development or reengineering of workflow models. Formally, any workflow can be described as graph that contains vertices of two types: actions and logical connectors. By means of the latter is given alternativeness of instantiation at the business process. We expand the standard conditions of soundness such as absence of dangling refrences, deadlocks and livelocks and consider that workflow model is sound if all initial facts for sake of model is a progress will be derived after process termination.
In order to eliminate unreal graph links we impose some functional restrictions on the potential maps between elements of the finite sets such as the set of actions, the set of join connectors and the set of split connectors. The syntactic control comes down to the checking existence of only given restrictions. The semantic control assumes that a workflow net doesn't contain livelocks. We propose the method using Boolean variants and functions for rapid unveiling total and contingent livelocks at workflow models. Join connectors are the points of expectative livelocks. Under the total livelocks any paths that pass through the join nodes are blocked and the business process is terminated. In case of contingent locking only some of the paths aren't interrupted.
As a result, syntactic and semantic controls are correspondently necessary and sufficient conditions of net correctness.
Keywords: workflow net, soundness, verification, lovelocks, syntactic and semantic correctness, syntactic and semantic control

P. 273–281

K. A. Shcheglov, Graduate Student, A. Yu. Shcheglov, Professor, e-mail: info@npp-itb.spb.ru,
University ITMO, St. Petersburg, Russia

Integrated Information-Operational Security Parameters and Characteristics Modeling Approach

We did research modern secure fault-tolerant information systems mathematical modeling problems in general. We formulated modeling problem, defined mathematical mechanism used for modeling, suggested basic integrated information-operational security parameters and characteristics modeling approaches — characteristics and parameters informational systems reliability and security. Such parameters and characteristics are determined on single complex model which allows to calculate their interference. Suggested modeling approaches can be applied while building failover secure informational systems.
Keywords: informational system security, information security, attack threat, vulnerability threat, informational system element failure, reliability, redundancy, mathematical modeling, comprehensive assessment, Markov model, integrated security

P. 282–290 

S. M. Avdoshin, Ph. D., Professor, Head of Software Engineering School, Faculty of Computer Science, e-mail: savdoshin@hse.ru,
A. V. Lazarenko, Undergraduate Student, Software Engineering School,
Faculty of Computer Science, e-mail: avlazarenko@edu.hse.ru, National Research University Higher School of Economics (HSE)

Peer-to-Peer Botnets

Huge computer networks composed of infected machines so called botnets are currently the one of the most dangerous threats for global society. Botnets are a brilliant tool for organizing distributed denial of service attacks, sending spam and phishing email, stealing sensitive information and money from various accounts. Hackers create botnets from any kind of electronic devices connected to the Internet. The most widely used devices are:personal computers with windows operating system, smartphones with Android OS, IoT devices with public IP addresses and open vulnerabilities (IPTV cameras, for example). There are a lot of options for botnet hiring on the darknet. It is possible to hire a botnet, buy a builder kit or any other service in hacker-to-hacker manner.
The easiest way to build a botnet is a creation of central server for spreading commands and creation of clients (bots) that will connect to this server. Such botnets are so called traditional botnets. Traditional botnets with centralized command and control
mechanism are vulnerable to the wide range of attacks. For example, if one particular bot is captured by the security analyst the IP address of command and control server will be compromised with very high probability. Without central command and control
server botnet is useless because the malefactor is unable to spread commands over infected network. In order to make botnets more
robust and steady the peer-to-peer botnet architecture was introduced. P2P architectures could be divided on 3 main categories:
pure, moderated and hybrid. The aim of the current papers is covering peer to peer botnets and its architecture.
Keywords: P2P, botnet, zombie network, peer-to-peer network, malware

P. 290–299

Yu. S. Efimov, Student, e-mail: yuri.efimov@phystech.edu, Moscow Institute of Physics and Technology, Dolgoprudny, Moscow Region, I. A. Matveev, Sci. D., Head of sector, e-mail: matveev@ccas.edu, Federal Research Centre ''Computing Centre'' of Russian Academy of Sciences, Moscow, Russia

Iris Image Segmentation by Paired Gradient Method with Pupil Boundary Refinement

Locating the iris area in an eye image is an important part of any iris recognition system. In many studies iris boundaries are modelled as circles and Hough transform is used to obtain their parameters precisely. The use of mentioned object detection algorithm is limited in case of noisy images or the occlusion of the iris by eyelashes, glasses, hair etc. Original Hough transform for circles also has high computational complexity. Many modifications were proposed to improve the robustness and speed of the method. But the circular iris boundary representation is insufficient in terms of segmentation accuracy, mainly because of human pupil shape variations. Therefore, more precise representation is required.
The proposed method consists of two main steps. The first step is rough iris center and boundary radii search using the modification of Hough methodology, named Gradient Pair method. Image is processed with Canny filter and pairs of pixels are selected from the resulting boundaries, which most likely belong to one circle-like iris area border. The second step is the pupil border refinement using the circular shortest path method. The retrieved circular path is also used to improve the quality of eye center search.
Computational experiment is performed on data from the public iris image databases to obtain the optimal parameters for Gradient Pair method and check the overall efficiency of the algorithm.
The proposed method provides the high quality of eye center and pupil boundary search. The iris segmentation accuracy is comparable to that of modern state-of-art methods and outperforms many of them.
Keywords: iris image segmentation, circle detection, Hough transform, circular shortest path

P. 300–309

A. V. Golov^{1, 2}, Graduate Student, e-mail: golov.andrey@hotmail.com, E. A. Timme^{2, 3}, Researcher, e-mail: alpdem@yandex.ru, A. V. Kozlov^{2, 4}, Chief specialist, Graduate student, e-mail: 89165363985a@mail.ru
¹Moscow Institute of Physics and Technology (State University), Moscow, Russia
²Moscow Center of Advanced Sports Technologies, Moscow, Russia
³State Research Center of Russian Federation — Institute of Biomedical Problems of the Russian Academy of Sciences, Moscow, Russia
⁴Russian State University of Physical Education, Sport, Youth and Tourism, Moscow, Russia

The Algorithm of Automated Estimation of Human Performance Parameters under Load Tests

Load physiological tests are universal method of estimation individual performance parameters that characterize source of energy of muscular work. The main purpose of load testing is estimation the aerobic and the anaerobic thresholds. The main problem of estimation of the aerobic and the anaerobic thresholds is processing of experimental results, which is held for the most part in manual mode and interpretation of data is often too difficult. It is not always possible to clearly identify important physiological information duo to noise, outliers, a significant dependence on the initial test conditions and individual modes of roll out of physiological processes. Estimations of the aerobic and the anaerobic thresholds obtained by different criteria can vary greatly and therefore the attempts of their association with each other give rise to more ambiguity.
The aim of this work is to develop mathematical algorithm of determination of the aerobic and the anaerobic thresholds based on gas analysis experimental data during tests with increasing load power protocols on different types of ergometers. Developed algorithm based on robust regressions methods, procedures of optimization, confidence interval computation by bootstrapping method. The algorithm uses data of multiple physiological parameters such as oxygen consumption, release carbon dioxide, respiratory coefficient, pulmonary ventilation, heart rate and is implemented as a software module. Comparison of estimates obtained by the developed method with estimates of experts-physiologists.
The software module has been developed and tested on data of endurance sports athletes. Obtained values of aerobic threshold and anaerobic threshold are consistent with expert's estimations. Therefore, the method can be successfully used in everyday practice of laboratory testing of athletes.
Keywords: functional load test, oxygen consumption, aerobic threshold, anaerobic threshold, ergometers, load test protocol, performance level, robust piecewise regression, bootstrapping method

P. 309–314

S. V. Belim, D. Sc. Professor, Head of Department, e-mail: sbelim@mail.ru,
S. B. Larionov, Postgraduate Student, e-mail: me@stas-larionov.ru Omsk State University n. a. F. M. Dostoevski, Omsk, Russian Federation

The Algorithm of Broken Pixel Detection Based on Community Detection in Graphs

The goal of this article is to analyze images that damaged using impulse noise. The first issue that need to be solved is searching of damaged pixels. This issue is causing on first step of noisy image restoring. Later one can restore source values of pixels based on already known positions of damaged pixels using approximation. This article suggests an algorithm of pixels detection, broken by impulse noise using image segmentation, based on community detection in graphs. The image that contains noise is representing as non-oriented, weighted graph. There is an exponential function that depends on color components of neighbour pixels is using to determine weights of edges. The resulting graph is splitting on communities. Damaged pixels are communities that contain only one vertex. Based on this approach one can build an algorithm with linear complexity. Suggested algorithm checking an expediency of combination each pixel of image with it nearest neighbors. Pixel assumed broken if it combination with nearest neighbors is not profitable. The computer experiment was carried out both on artificial images and photos. For each images impulse noise was generating using generator of pseudo-random numbers. Noise level of analyzed images was from 10 % to 70 %. Computer experiment showed that suggested method has some advantages compared to others when noise level is high. When noise level is low, the results of suggested algorithm are comparable to analogs. Suggested method has much greater efficiency compared to others when noise level is high. It should be pointed out that algorithm has lower level of false positives then analogs. Additionally, with increasing of noise
level, count of false positives decreases which distinguishes the proposed algorithm.
Keywords: image analyzing, impulse noise, image restoring, community detection, community

P. 315–320  

To the contents