Vol 22, No 7 (2016)

The industrial ecology are typical management tasks pollution emission sources, the territory of the group. If you set the standards of environmental safety territory (scalar, vector, or function), the optimization should detect the redundancy of such emissions.

The task is complicated potential functions multiextremal general pollution of the territory as a result of imposition of emission sources, as well as the presence of the infinite parameter — the number of points of the territory, which leads to an infinite number of constraints.

Algorithms for semi-infinite optimization (SIP, Semi-Infinite Programming) reduces the initial formulation of the problem with an infinite number of constraints to the sequence of approximating problems with a finite number of critical constraints. Critical limits are maximizing function limitation from the start, the corresponding point of the next constraint violation. To search for violations of applied stochastic procedure. Approximating the final task it is usual formulation of nonlinear mathematical programming.

However, a theoretical departure from the infinity does not negate choice discrete grid area. The numerical experiment with parameterization by a discrete grid. Range Resolution made by [400 x 200] to [3200 x 1600].

A significant and sustainable impact on the approximation of the discrete nature of the results. Changes in grid discrete nature do not lead to a loss of convergence of approximations. Increase grid discreteness leads to better quality (distributed on most items) approximations.

Introduced and analyzed several surrogate grid discrete nature (approximation dirt overlay function, the objective function weighted approximations). The logic of their trend is well aligned with the variable grid increments.

Selecting an area must first demonstrate sufficient quality (to describe) the characteristics of the overall pollution of functions, but it remains a heuristic procedure.

The article discusses aspects of the implementation of network technology in distributed control systems. The described various approaches of hardware support for network solutions provide higher throughput capacity in the information environment by: exclusion of arbitration procedures, elimination of control information from the data packages, the organization of exchange between the CPU and peripherals based on a unified packages of state and control. The proposed technology, implemented on FPGA, allows to achieve minimal and stable delay of transfer data in on-board networks.

In this paper we describe how to accelerate object group identification and tracking in real time. It is in need to remove non informational part of image backward and to detect highly salient objects. The salient objects are detected by mean-shift segmentation in image pixel feature high dimensional space and by determining the segmented image salience map. The frequency-tuned salient region detection method is in use, which produces segmented image full resolution saliency map displaying salient object boundaries. The method preserves the object boundaries by retaining main frequency content from the segmented image and exploits features of color and luminance. Since salient object detection is not a pure low-level process, next a unified model is in use to incorporate the image low-level features with higher-level knowledge. In the model, a special image matrix in the feature space is decomposed as a low-rank matrix representing background plus sparse matrix indicating the salient regions. To ensure the model validity, a linear transform of the image feature space has to be learned. Next, the higher-level knowledge is fused to compose a prior map, and is treated as a prior term in the objective function. The approach efficiency is demonstrated on sample image with salient objects.

A number of diverse factors in the area of microelectronics industry significantly actualized the developments in the field offault-tolerant combinational circuits design recently. Currently in modern CAD systems there are no means to assess and improve logic masking effects in combinational circuits. In this paper, an attempt was made to create a methodology for the implementation of such software. This paper proposes re-synthesis design flow, providing a logical masking capability for the combinational circuits resistance to a single event upsets without the introduction of substantial redundancy. The details of the method including the development of reliability metrics, the selection of vulnerable sub-schemes, as well as methods of generating reliable circuits are presented. Computational experiments were conducted to demonstrate the high efficiency of the proposed method.

The article is devoted to experimental study of characteristics of the new method of secure interactions organization in multi-network environment. The method is based on the use of HTTPS protocol and proxy server technology. The essence of approach consists in the organization of the protected HTTPS connections between proxy servers for safe transfer of information queries from one private local network in another through a global network while client and server components of distributed system are released from of information protection functions. In this approach client and server programs use HTTP protocol and proxy servers have additional functions of "HTTP - > HTTPS" and "HTTPS - > HTTP" gateways. The main advantage of approach over known (for example, VPN) consists in providing "high-leve"" means of HTTP queries routing and filtration the based on headers of HTTP queries, but not on separate IP packages. The application of approach is limited to situations in which private local networks can be considered as "a trust zone", and the only source of threats is the global network. The experimental study of efficiency of the offered approach in terms of influence ofprotection gateways on the speed of electronic services calls processing is conducted. This research showed that the area of effective application of the approach is made by electronic services with performance time over 1—2 s.

Nowadays the computer adaptive testing plays the main role in massive educational testing. This partly causes the growing interest to computerized forms of testing. In its turn, data transfer between item banks and test systems has become the topic of high actuality this last few years. This paper address questions about the interoperability of educational testing systems. The most popular text and XML-formats of the test questions (including rapidly developed QTI/APIP) were studied. The review of existing converters demonstrated that there is no freely accessed converter, which supports several questions’ types, has adequate Russian interface or allows internationalization, and offers authoring tools for the questions of different types. Consequently, the architectural and functional software requirements on test questions’ converter were determined and listed. As a result, the paper presents the web- APIprototype, which was design and implemented according to collected requirements.

Joint analysis of combinatorial, probabilistic and synergistic approaches to determining the amount of information on the basis of the consideration of integrative codes of discrete system elements, divided into parts according to the values of arbitrary sign, have been carried out in this article. (Integrative code is an individual description for each element of the system and presents a sequence of symbols of any alphabet, the length of which is a function of the total number of elements.) It is shown that, in general, integrative codes are divided into a group and system parts. The group part of code is specified by a set of elements which have identical value of attribute, system part is superstructure over group part and it is result of summing of sets having different characteristic into integrated system.

It was found that information measures of combinatorial, probabilistic and synergistic approaches to determining the amount of information are exactly expressed through the weighted average of the group and system parts of an integrative code. Moreover, if the measures of synergetic approach are expressed through the weighted average of the group and system parts of a code, the measures of combinatorial and probabilistic approaches are functionally linked only with the weighted average of the system parts. Following conclusions are made on this basis. First of all information measures of combinatorial, probabilistic and synergistic approaches are essentially the quantitative characteristics of the structural features of the integrative codes of an elements of a discrete systems. Secondly, the information appearing in a synergetic approach (information about a finite set, as a single entity), is genetically primary in relation to the information measured in combinatorial and probabilistic approaches (removed uncertainty of selection one of the many opportunities).

The article describes main problems and difficulties that are currently available and are likely to be present in the future in the field of development and introduction of new microelectronic technologies in the advanced neuro-computers.

This structure of works represents the operation procedure and composition of the work on introduction of new microelectronics technology in high-performance neurocomputers proposed by the author. Both the operation procedure and scope of work may vary depending on the type of technology, with required, or at least desirable, justification for the exclusion of certain stages of work presented below in this article:

• about classical possible approaches to task solutions;
• work structure for the design and implementation of neural network based algorithms for problem solving;
• the description of spiking neurak network algorithms;
• development of neurochips and neuromorphic computers with use of memristor.