“Informatics and Applications” scientific journal
Volume 10, Issue 2, 2016
Content
Abstract and Keywords
About Authors
- A. V Borisov Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
- A. V. Bosov Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
- G. B. Miller Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
Abstract: The Real Time Transfer Protocol (RTP), widely used in Voice over IP (VoIP) technologies for audio and video data transmission, is analyzed to design a mathematical model of VoIP connection. The model attempts to meet basic features of VoIP technologies as well as key features of the real link functioning like the frame delays, losses, etc. The proposed approach is based on the finite-state unobservable hidden Markov model. The unobservable state is assumed to be a finite-dimensional Markov process, whereas the observation is assumed to be a non-Markovian multivariate point process that indicates the heterogeneous frames reception. For the proposed model, the hidden link state optimal filtering problem given the packet/losses stream observations is formulated and its solution is provided. Proposed link model validity and filtering algorithm performance are illustrated by processing of captured real video streams delivered via 3G mobile network by Linphone VoIP services.
Keywords: VoIP technologies; RTP; network link; hidden Markov model; multivariate point process; optimal state filtering
- I. N. Sinitsyn Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
- E. R. Korepanov Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
Abstract: The analytical synthesis theory of continuous and discrete sub- and Pugachev conditionally optimal filters and extrapolators for information processing in linear state stochastic systems (StS) is presented. For Gaussian StS, Liptzer and Shiraev performed the first works for filters and extrapolators synthesis. For non-Gaussian StS, the first works belong to Pugachev and Sinitsyn. Stochastic equatuins for state and observation of continuous and discrete StS are given. Algorithms for continuous normal sub- and conditionally optimal filters and extrapolators are presented. The corresponding algorithms for discrete StS are also given. The developed algorithms are the basis of the software tool "StS-Filter, 2016." The results maybe developed for autocorrelated noises and multiplicative noises.
Keywords: Liptser-Shiraev filter (LSF); Liptser-Shiraev conditions; normal approximation method (NAM) for a posteriori density; normal conditionally optimal Pugachev filter (NPF); stochastic systems (StS); state linear StS; statistical linearization method (SLM)
- I. N. Sinitsyn Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
- V. I. Sinitsyn Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
- E. R. Korepanov Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
Abstract: For nonlinear differential stochastic systems on manifolds (MStS) with Wiener and Poisson noises, the synthesis theory of ellipsoidal suboptimal filters based on ellipsoidal approximation and ellipsoidal linearization methods is developed. Special attention is paid to MStS with additive non-Gaussian noises based on ellipsoidal linearization method. The algorithms are the basis of the experimental software tool "StS-Filter" (version 2016). Accuracy and sensitivity equations are presented. Some generalizations are mentioned.
Keywords: ellipsoidal approximation method (EAM); ellipsoidal linearization method (ELM); orthogonal expansions method (OEM); Poisson noise; stochastic system on manifolds (MStS); suboptimal filter (SOF); Wiener noise
- A. V. Goncharov Moscow Institute of Physics and Technology, 9 Institutskiy Per., Dolgoprudny, Moscow Region 141700, Russian
Federation
- V. V. Strijov A. A. Dorodnicyn Computing Centre, Federal Research Center "Computer Science and Control" of the Russian
Academy of Sciences, 40 Vavilov Str., Moscow 119333, Russian Federation
Abstract: The paper discusses the problem of metric time series analysis and classification. The proposed
classification model uses a matrix of distances between time series which is built with a fixed distance function.
The dimension of this distance matrix is very high and all related calculations are time-consuming. The problem of
reducing computational complexity is solved by selecting reference objects and using them for describing classes.
The model that uses dynamic time warping for building reference objects or centroids is chosen as the basic model.
This paper introduces a function of weights for each centroid that influences calculation of the distance measure.
Time series of different analytic functions and time series of human activity from an accelerometer of a mobile
phone are used as the objects for classification. The properties and the classification result of this model are
investigated and compared with the properties of the basic model.
Keywords: metric classification; weighted dynamic time warping; time series classification; centroid; distance function
- R. V. Isachenko Moscow Institute of Physics and Technology, 9 Institutskiy Per., Dolgoprudny, Moscow Region 141700, Russian
Federation
- V. V. Strijov A. A. Dorodnicyn Computing Centre, Federal Research Center "Computer Science and Control" of the Russian
Academy of Sciences, 40 Vavilov Str., Moscow 119333, Russian Federation
Abstract: This paper is devoted to the problem of multiclass time series classification. It is proposed to align time series in relation to class centroids. Building of centroids and alignment of time series is carried out by the dynamic time warping algorithm. The accuracy of classification depends significantly on the metric used to compute distances between time series. The distance metric learning approach is used to improve classification accuracy.
The metric learning procedure modifies distances between objects to make objects from the same cluster closer and from the different clusters more distant. The distance between time series is measured by the Mahalanobis metric.
The distance metric learning procedure finds the optimal transformation matrix for the Mahalanobis metric. To calculate quality of classification, a computational experiment on synthetic data and real data of human activity recognition was carried out.
Keywords: time series classification; time series alignment; distance metric learning; LMNN algorithm
- A. N. Tyrsin Science and Engineering Center "Reliability and Resource of Large Systems and Machines," Ural Branch of the Russian Academy of Sciences; 54a Studencheskaya Str., Yekaterinburg 620049, Russian Federation
- S. M. Serebryanskii Troitsk Branch of Chelyabinsk State University, 9 S. Rasin Str., Troitsk 457100, Russian Federation
Abstract: The article describes the method of recognition of dependences based on the use of inverse mapping. From a given finite set of models, one chooses the model that best fits the sample data. For each model, the selective dependence corresponding to it is determined by the sample. For the one-dimensional case, each selective dependence is mapped to the same reference model in the form of the straight line equation by means of inverse mapping. For each model, sample data are mapped to the same equation of the straight line with some mistakes. It is suggested to use the minimum of variance of mistakes as the criterion of adequacy of the constructed model of sample of data. In the case of multidimensional dependences, a heuristic method is suggested according to which a set of inverse functions for each of explanatory variables is considered for each model. Approbation of the method by means of statistical modeling by the Monte-Carlo method is carried out.
Keywords: recognition; functional dependence; model; inverse function; sample; variance; approximation
- O. V. Shestakov Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University, 1-52 Leninskiye Gory, GSP-1, Moscow 119991, Russian Federation, Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
Abstract: The thresholding techniques for the wavelet coefficients of the signal and image functions have become a popular denoising tool because of their simplicity, computational efficiency, and possibility to adapt to the functions with different amounts of smoothness in different locations. The paper considers the recently proposed stabilized hard thresholding method which avoids the main disadvantages of the popular soft and hard thresholding techniques.
The statistical properties of this method are studied. The unbiased risk estimate is analyzed in the model with an additive Gaussian noise. Wavelet thresholding risk analysis is an important practical task, because it allows determining the quality of the techniques themselves and the equipment which is being used. The paper proves that under certain conditions, the unbiased risk estimate is strongly consistent and asymptotically normal. These properties allow constructing the asymptotic confidence intervals for the theoretical mean squared risk of the method.
Keywords: wavelets; thresholding; unbiased risk estimate; asymptotic normality; strong consistency
- Ya. M. Agalarov Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
- M. Ya. Agalarov PromsvyazBank OJSC, 10 Smirnovskaya Str., Moscow 109052, Russian Federation
- V. S. Shorgin Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
Abstract: The paper considers the problem of maximizing the average profit per time in the M/G/1 system on the set of access restriction stationary threshold strategies with one "switch point." Profit in the described model is defined as the following measures: service fee, hardware maintenance fee, fine for service delay, fine for unhandled requests, and fine for system idle. The conditions of existence of optimal and finite threshold values are obtained.
The method and the algorithm for calculating the lower bound for the optimal threshold and corresponding value of maximal profit per time are proposed. The auxiliary problem of maximizing the system profit, averaged by number of handled requests on the set of the considered threshold strategies, is solved. The necessary and sufficient conditions of existence of solution of the auxiliary problem are found. The method and algorithm for its solution are proposed.
Keywords: queuing system; threshold strategy; optimization
- A. A. Kudryavtsev Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M. V. Lomonosov
Moscow State University, 1-52 Leninskiye Gory, GSP-1, Moscow 119991, Russian Federation
- S. I. Palionnaia Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M. V. Lomonosov
Moscow State University, 1-52 Leninskiye Gory, GSP-1, Moscow 119991, Russian Federation
Abstract: This work is devoted to the study of the parabolic distribution of parameters in the Bayesian recurrent
model of reliability growth of complex modifiable information systems. In the reliability theory, the reliability of the
system depends on the ratio of parameters which are interpreted as indexes of "defectiveness" and "efficiency" of
the tool correcting the deficiencies in the system. In the framework of Bayesian models, it is assumed that only the
information about the a priori distributions of the system's parameters is given. In this work, the average marginal
system reliability is calculated for the a priori parabolic distribution of the parameters. The numerical results for the
model examples are obtained.
Keywords: modifiable information system; reliability theory; Bayesian approach; parabolic distribution
- A. S. Markov Faculty of Computational Mathematics and Cybernetics, M. V. Lomonosov Moscow State University, 1-52 Leninskiye Gory, GSP-1, Moscow 119991, Russian Federation
- M. M. Monakhov Faculty of Computational Mathematics and Cybernetics, M. V. Lomonosov Moscow State University, 1-52 Leninskiye Gory, GSP-1, Moscow 119991, Russian Federation
- V. V. Ulyanov Faculty of Computational Mathematics and Cybernetics, M. V. Lomonosov Moscow State University, 1-52 Leninskiye Gory, GSP-1, Moscow 119991, Russian Federation
Abstract: Generalized Cornish-Fisher expansions are constructed for quantiles of sample mean for a sample of random size in terms of quantiles for the Laplace distribution and Student's t-test. In recent years, the interest in Cornish-Fisher expansions grew significantly in the context of research on risk management. The widespread risk measure Value at Risk, or VaR, is, in fact, the quantile of the loss function. The authors use the general transfer theorem that makes it possible to obtain asymptotic expansions for the distribution functions of statistics based on samples of random size by asymptotic expansions for the distribution function of the random sample size and asymptotic expansions for the distribution functions of statistics based on nonrandom samples. A computational experiment was performed to illustrate the obtained Cornish-Fisher expansions.
Keywords: quantiles; generalized Cornish-Fisher expansions; random size sample; Laplace distribution
- V. G. Ushakov Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University, 1-52 Leninskiye Gory, Moscow 119991, GSP-1, Russian Federation, Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
Abstract: The time-dependent process in the single server vacation model with hyperexponential input stream is analyzed. The Laplace transform (with respect to an arbitrary point in time) of the joint distribution of server state, queue size, and elapsed time in that state is obtained. The author restricts themselves to a system with exhaustive service (the queue must be empty when the server starts a vacation). The queueing systems with vacations have been well studied because of their applications in modeling the computer networks, communication, and manufacturing systems. For example, in many digital systems, the processor is multiplexed among a number of jobs and, hence, is not available all the time to handle one job type. Besides such an application, theoretical interest in vacation models has arousen with respect to their relationship with polling models.
Keywords: hyperexponential input stream; working vacations; single server; queue
- Yu. S. Khokhlov Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M.V. Lomonosov Moscow State University, 1-52 Leninskiye Gory, Moscow 119991, GSP-1, Russian Federation
Abstract: Since the beginning of the 1990s, many empirical studies of real telecoomunication systems traffic have been conducted. It was found that traffic has some specific properties, which are different from common voice traffic, namely, it has the properties of self-similarity and long-range dependence and the distribution of loading size from one source has heavy tails. Some new models have been constructed, where these features were captured. Brownian fractional motion and ŕ-stable Levy motion are the well-known examples. But both of these models do not have all of the above properties. More complicated models have been proposed using some combination of these ones. In particular, the authors have proposed a variant of univariate fractional Levy motion. This paper considers a multivariate analog of fractional Levy motion. This process is multivariate fractional Brownian motion with random change of time, where random change of time is Levy motion with one-sided stable distributions.
The properties of this process are investigated and it is proven that it is self-similar and has stationary increments.
Next, it is shown that the coordinates of one-dimensional sections of this process have the distributions which are not stable. But asymptotic of tails for these distributions is the same as for the stable ones. This model is applied to analyze heterogeneous traffic and to get a lower asymptotic bound of the probability of overflow of at least one buffer. There are other possible applications.
Keywords: fractional Brownian motion; ŕ-stable subordinator; self-similar processes; buffer overflow probability
- V. A. Minin Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
- I. M. Zatsman Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
- V. A. Havanskov Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
- S. K. Shubnikov Institute of Informatics Problems, Federal Research Center “Computer Sciences and Control” of the Russian
Academy of Sciences, 44-2 Vavilov Str.,Moscow 119333, Russian Federation
Abstract: The paper discusses the information relationship between science and technology and the methods for indicator assessment of transfer processes (transfer) of knowledge from different fields of research in the area of technological development. The proposed methods are designed to determine the values of the indicator of intensity of citation of scientific papers in the descriptions of the inventions patented in Russia by domestic and foreign applicants. A similar approach can be used to obtain indirect estimates of innovation potential of scientific research. The indicator values of intensity were calculated both in general and with the distribution by country of applicants. The paper presents the results of determining the values of the indicator. Full-text descriptions of inventions on class G06 of the International Patent Classification (Data Processing; Computing; Score) published by Rospatent in 2000-2012 were used as the source of information. The use of information resources of Rospatent was due to the fact that they are in the electronic form, i. e., available for automated processing. The result is the values of the indicator of intensity of citation of scientific works patented in the Russian Federation, divided into groups of domestic, foreign, andjoint inventions. This specification allowed to estimate the activity of international technological cooperation andjoint patenting in information and computer technologies (ICT) in Russia, as well as to determine the themes of cooperation in this area.
Keywords: citation of scientific papers; intensity of citation linkages between science and technologies; information technology; international patent classification; calculation of values of the indicator of intensity of citation
- L. A. Meykhanadzhyan Peoples' Friendship University of Russia, 6 Miklukho-Maklaya Str., Moscow 117198, Russian Federation
Abstract: Consideration is given to the M/G/1/(r - 1) queueing system with LIFO (last in, first out) preemptive generalized probabilistic priority policy. It is assumed that customer's service time becomes known upon its arrival at the system and at any time instant remaining service times of all customers present in the system are available. On arrival of a customer at a nonempty system, its service time is compared to the (remaining) service time of the customer in service and one of the following events occurs: both customers leave the system at once, one of the customers leaves the system (the other occupies the server), or both customers stay in the system (one occupies the server, the other - one place in the queue). Those customers which stay in the system acquire new service time according to a known distribution, which can depend on their initial service times. Arriving customers which find the queue full, leave the system and have no influence on it. Analytical expressions forthe computation of the joint stationary distribution of the number of customers in the system and the remaining service time of the customer in the server, of the busy period and the stationary sojourn time (in terms of Laplace-Stieltjes transform) are proposed.
Keywords: queueing system; special discipline; LIFO; probabilistic priority
|
Phone of the Center: +7 (499) 135-62-60
RUS