According to the results of the analysis of the state of the level of development of unmanned aerial vehicles with an electric traction unit, the concept of increasing energy efficiency of the electric drive of the traction unit is proposed. It is proposed to use pulsed quasi-resonance transducers with the minimum dynamic switching power loss as the main voltage converter of unmanned aerial vehicles.

The article was designed database (DB) for the existing project "Green School". The task was to collect, store and process information from SWEGON GOLD using web-technologies. The collection of information was performed using the PHP library c URL. To store information, the MYSQL database was used. To reduce the query time, caching and optimization of database queries were performed. On the basis of the obtained data, a conclusion was made on the possibility of optimizing and accelerating work with the database of the web-application.

The factorization of a covariance function or its analog in the frequency domain - spectral factorization - is a much more difficult task than the synthesis of a covariance function or spectrum. When considering the factorization problem, two questions arise. The first is connected with the motivation of why it is necessary to study this task in general, and the second one - with the choice of the most effective procedure for factorization. The necessity of studying factorization algorithms follows from the fact that in many tasks there is no information on the process generation, and the known measured characteristics of the output process: covariance function or spectrum. Regardless of any consideration of the discovery or evaluation, it is desirable to find a model for describing the dynamics of process generation, since it can give a deeper understanding of the physics underlying the task. In the context of detection and evaluation, many of the synthesized receivers contain devices for evaluating process parameters. In many tasks, the implementation of the relevant evaluation algorithms requires knowledge of the generation model. This article is devoted to the development of the software application Trinity-factor for determining the noise signals by the factorization method of the correlation matrix of the multidimensional dynamic system.

The paper considers the main types of DDoS attacks on computer networks, provides recommendations for protection against these types of attacks.

The article describes the necessity of automation of work of the enterprises of the library sphere. An information system was designed and developed that allows automating general operations to optimize the work of library institutions. To do this, an object-oriented development environment was used in a high-level language with the connection of the database from MS Access as a data warehouse. The created software supports user accounts.

In this paper, the relevance of decision support decision systems is considered, and the requirements for the DSS, which will be used by the enterprise engaged in retail trade, are described. The goal was to develop a system that would perform the following functions: calculation of the balance of the goods at the end of the working day; sorting; calculation of total sales for every 2 weeks; allocation of sales, which sales exceeds the average, or is less than average; processing sales information; processing information on the purchase of goods; processing of information about the remnants; forecast sales for future periods. To implement the SPPR, the Microsoft Office Excel 2010 and Visual Basic 2010 environments were used.

At today's level of complexity of electronic and computer technology knowledge of the basics of technical diagnostics becomes mandatory for specialists in the field of development and operation of such technology. The application of methods and means of technical diagnostics is an effective way to ensure high reliability of products, reduces the time for their manufacture and repair. The number and complexity of electronic products manufactured by the industry is growing so rapidly that it is difficult to imagine a clever decision to train specialists for debugging and technical diagnostics that would cope with their tasks, with only manual and intuitive methods for detecting and troubleshooting. The object of the control system may consist of other objects that may have a permanent structure of interconnections. The article is devoted to the analysis of methods and means of technical diagnostics of technical objects. The approaches to the construction of mathematical models of test diagnostics of control systems are determined.

The article presents the latest software paradigm - probabilistic programming. This paradigm allows us to use all the more sophisticated probabilistic models to provide ever more advanced tools for the development and creation of artificial intelligence. Since probabilistic programming should take its place among other software paradigms and have its own specific features, it was the task of identifying them, as described in this article.

Nowadays, local area networks (LANs) are widely used and probably will be used later. But the larger and more complex the network, the more problems and malfunctions in it can arise. Therefore, it is necessary to increase the fault tolerance and reliability of local computer networks, especially in the structures of enterprises, where LAN is a key tool in the work. Network performance is provided by eliminating bottlenecks in the network nodes and collision domains, which reduce the speed of operation and can lead to the interruption of the signal from the network node. In these cases, it may take a long time to restore the LAN operation, which can lead to idle time and financial losses. Also important is the security of the network, which can be improved by installing network filters, and setting the appropriate security policy for each network node. This article deals with the method of correction of a single failure in a local computing network. The matrix of reach is constructed through the adjacency matrix.

Communication security in the transmission of speech messages is ensured by changes in the characteristics of the language in such a way that it becomes indiscriminately and unrecognizable for the intruders who eavesdrop or intercept a closed speech message. The choice of methods for closing speech signals depends on the type of specific use and technical characteristics of the transmission channel. This article analyzes the methods of closing speech messages and examines the types of language closure systems, and their differences. The quality and degrees of secrecy of language closure schemes are considered. Selected type of closing speech signals for further development of the device.

Cellular sensor networks are one of the most promising areas for the development of modern wireless telecommunication technologies. The prospects of using sensor networks can not only replace the cable infrastructure with wireless but also get new functionality due to the miniature nodes, low power consumption, built-in radio interface, sufficient computing power, and relatively low cost. They have already found widespread use in automation systems of information gathering, monitoring and control of the characteristics of various technical and natural objects. The main feature of the use of porous structure is the creation of data transfer systems with a dynamically changing structure, to the basic access points is the ability to connect client devices with built-in logic (Smart Sensors), which can be personal computers, PDAs, mobile phones, etc. ., which become nodes of the network. Devices connected to the network can participate in data transfer, increasing the reliability and performance of the system. The emergence of new standards for wireless telecommunications and the emergence of porous data structure have greatly expanded the scope of their use. This work proposes the use of sensor networks in sports tactical games such as lasterag, paintball, strikeball, hardball and other similar type.

This paper presents a new developed method that selects a corpus of a concatenated synthesis of a recorded speech. The purpose of such selection is to attribute greater phonetic and prosodic variability to synthetic speech. In this procedure, four steps are measured: prosodic annotation, grapheme-phoneme conversion, representation in characteristic vectors and selection itself.

The problem of blind source separation (BSS) refers to recovering original audio signals, called source signals, from the mixed signals, called convolutive mixtures or observation signals, in a reverberant environment. The mixture is a function of a sequence of original speech signals mixed in a reverberant room or ambient. The objective is to separate mixed signals to obtain the original signals without degradation and without prior information of the features of the sources, such as the locations, the spectral nature or the mixing method. The strategy used to achieve this objective is to use multiband schemes. Multiband schemes work at a lower sampling rate, have less computational cost and a quicker convergence than the full-band approach. For more challenging scenarios, an unequal-passbands scheme with oversampled decimation is proposed for blind source separation application; the greater number of bands, the more efficient the separation. The results are compared to the currently best performing method and three standard methods and show a significant improvement in the computational saving and convergence speed.

Speech is generated from the juxtaposition of automatically selected segments of parts restricted in a earlier recorded and partitioned bank in concatenated synthesis. The synthetic speech obtained through this technique presents naturalness influenced by the effectiveness of the segmentation tools. The act of such tools is regularly favored by a mixed method causing from an HMM model through a refinement method of segmentation. This refinement has been successfully accomplished through the use of techniques based on neural networks. In this work, it is proposed a set of networks whose performance is superior to the other topologies of neural networks presented in the literature. In this case, the training of the networks is done by a distribution of the training set based on the borders of phonemes with similarities between them.

The prosody target of concatenated speech synthesis systems is obtained by using modeling algorithms trained in an offline procedure. These algorithms are developed with reference to linguistic characteristics (for example, syllabic separation, tonicity and division into phonological and classification of phrases) of the sentences contained in a recorded speech bank. In this work, a procedure is presented for automatic classification and division of sentences into phonological phrases. The classification is based on rules that result in a hit percentage of 92.86%. Moreover, it can help in the process of predicting pauses, leading to an adequate establishment of concatenated speech synthesis and, consequently, to a better naturalness of synthetic speech.

In recognition, synthesis and speech conversion systems, a text corpus is required to obtain the speech bank taken as a reference in the development of such systems. The selection of this corpus is often performed automatically through the solution of an optimization problem to maximize a function that represents phonetic / prosodic coverage. The state-of-the-art technique for solving such a problem is called additional greedy. When compared to other approaches presented in the literature, additional greedy exhibits low computational complexity; however, it performs poorer results in the optimization process. In order to improve this performance, this paper proposes a procedure for selecting the corpus based on two steps. First, additional greedy algorithms are applied in the selection of a smaller corpus from a larger initial corpus. In the second step, a greedy genetic technique selects a new set of sentences using the resulting corpus of the first step. In this way, a better compromise relationship between complexity and performance is established in the optimization process. To our knowledge, the greedy genetic selection technique has not been used before for this purpose.

This paper proposes a performance evaluation methodology as well as compares some visual quality metrics for perception threshold characterization in images degraded by Gaussian additive noise. Both non-perceptual (SNR and PSNR) and perceptual metrics (MSSIM, IFC, FV and C4) are evaluated. The subjective evaluation experiment that serves as a basis for the presented results is performed with approximately 120 observers, considering a test set with 60 images, divided into five groups. The data analysis seeks to estimate a detection threshold value for each of the metrics evaluated as well as the quality obtained from the corresponding estimate.

This article presents a new strategy for implementing artificial bandwidth extension (ABWE) systems for speech signals applied to telephony in packet networks. In the literature, several approaches are proposed; however, there is not yet an adequate procedure that can satisfactorily represent speech segments with concentrated energy at high frequencies. In order to improve the performance of these systems, a new strategy based on classification is discussed here. The results obtained are evaluated and the quality enhancement of the reconstructed speech signals confirms the effectiveness of the proposed strategy.

The article deals with the transmission of information in the form of a voice. The relevance and perspective of the topic in modern network needs is indicated. The methods and their language coding parameters are considered. The principles of the creation of the method of encoding СЕLP are resulted. The linear auto regression model of the process of formation of speech signals is presented and it is indicated that the autoregressive model of the speech signal describes it with a rather high degree of accuracy and allows the use of advanced mathematical apparatus of linear prediction. The advantages of the CELP encoding method are announced.

Telecommunication systems are widely used in almost all branches of human activity. With long-term operations, telecommunications equipment can be damaged, which leads to undesirable information loss. The process of diagnosing the critical states of a telecommunication system is quite laborious and requires an analysis of many factors and parameters in order to accurately identify and eliminate the problem. To speed up this process, neural network technologies can be used. In this article, the main components of the neural network and its applications for the diagnosis of critical states of telecommunication systems are considered.

In the paper, the analysis of the restoration of the thermal field configuration by discrete values ​​is carried out using methods of bilinear interpolation and rescaling of Lanczos. An analysis of the mean square deviation and spectral characteristics shows that the minimum hardware does not allow the field to be adequately restored due to the impossibility of registering the high-frequency component. The advantages and disadvantages of alternative methods of representation of a thermal field on the basis of pyrometric sensors are considered.

The moisture content of grain and other agricultural products is taken into account during delivery and admission to grain processing enterprises and grain storages, since the net weight of the product and, consequently, the actual cost depend on it. Increasing the moisture content of the products also leads to non-production costs during transportation. One of the most important characteristics of grain is moisture. Conditionally, the results of moisture measurement can be divided into two components: commercial and technological. This article reveals the main problems of measuring the moisture content of the grain in the stream. Description of the main methods for measuring the moisture content of bulk materials is presented. The analysis of the methods of measuring the humidity and found optimal. The existing tools for grain analysis in the flow are presented.