Architectures and models of high-performance information processing systems

Algorithms and software for remote and distributed processing of integrated circuits topology images on a semiconductor wafer in a multiprocessor system. Effective algorithms and software modules for constructing images of the original topology on a semiconductor wafer have been developed. The software package is intended for use in the development of VLSI topology originals for detecting topology defects that lead to the formation of defects in the process.

[task PR5.1 (3ST/05-222) "To develop and implement the software package for modeling photolithography processes for automatic control systems for VLSI topology originals, focused on high-performance cluster multiprocessor computing systems" (2005-2008, GR No. 2006507) under the program of the Union State "Development and implementation of science-intensive computer technologies based on multiprocessor computing systems in the member states of the Union State" ("Triad")]

The system of preparing topological information for microphotosetting image generators. During the work on the system, the following methods were developed and implemented in software: methods for covering special topological objects with rectangles; method for dividing a set of simple contours into classes containing intersecting contours; method for establishing the inclusion relationship between circuits; information transformation methods (combining intersecting contours; combining intersecting rectangles; highlighting simply-connected and multi-connected topological objects); method for covering arbitrary multi-connected topological objects with rectangles; method for analyzing the obtained coatings for correctness; method for covering the interior regions of a multiply connected topological object, detected during its correctness analysis.

[task 3-07 "To develop the software system for preparing topological information for microphotosetting image generators" (2006-2010, GR No. 2007184) SSTP "Advanced Information and Telecommunication Technologies"]

Technology for transparently launching applications that run under Scientific Linux OS in the Fedora Core operating environment. The developed technology and software make it possible to integrate the resources of the Republican Supercomputer Center for Shared Use into the infrastructure of GRID systems built on Scientific Linux as the main OS. Documentation for installing and configuring application launch services and integrating RSCC resources into the GRID environment has been developed.

PCB topology control system. This system is implemented under Linux and Windows XP in the C++ programming language. The test image of the PCB obtained with the help of the optical system is used as the processing object. To find defects, a reference image obtained from the vector description file of the PC and applied at the stage of its production is also used. The system operation result is files containing information about AE and PCB tracks and their defects on the test PCB in a raster (bmp-files) and vector form (con-files). The system is a single .pcb executable file that integrates all software modules. To work in the Linux environment, you need the X-window graphics server, the QT graphics library version 3.0 or higher. To work in the Windows XP environment, you need Microsoft.NET Framework SDK v.2.0 and software that allows you to work with the QT library. The system performs the following functions: interactive formation of a list of reference corrective elements on the reference PCB, specified in the Gerber format; automatic search for corrective elements on the tested PCB; automatic correction of geometric distortions of the tested PCB raster image by reference points; interactive generation of a list of reference AEs; automatic control of compliance of the TP of the tested PCB with the data of the reference PCB; formation of TP defects lists for the tested PCB; automatic control of PCB track defects; interactive viewing of PCB images specified in raster, Gerber and internal formats.

The experimental software for neural network classification of image areas of the VLSI technological layer. Basic software modules are data preparation, formation and management of a neural network model. The data preparation module prepares the input images of the IS layers and identifies areas in the topology images based on fuzzy clustering, forms the necessary training samples from the input images of the integrated circuit layers for a given neural network model, forms fuzzy sets for models on fuzzy neural networks. Depending on the features of the identified image objects, it is proposed to use a multilayer perceptron with a logistic activation function having a sigmoidal shape, a neocognitron based on a fuzzy neuron or an ensemble of the above-mentioned neural networks. The module for the formation and management of a neural network model allows forming and training the selected neural network model and performing the classification of identified objects. A feature of the developed software is the ability to form complexes of hybrid and hierarchical neural networks via the software. This allows both speeding up the process of constructing the classification neural network and improving the classification accuracy.

[F16-070]

Software package for preparing topological data for specialized technological equipment for controlling topological structures on photomasks. The software is designed for converting and syntactic analysis of topological data from CAD formats into the format of specialized technological equipment produced by VLSI: image generators and automatic control units for semiconductor wafers and photomasks. Supports MEBES, CIF, DXF, ZBA-20 and Gerber formats. The software package is developed on the basis of OpenMP technology for operation on a PC with a 4-core processor. The algorithm for parallel implementation of dividing contours into primitives has been developed.

[IT_2-05]

The software complex for image processing and analysis for technical vision systems in the microelectronic industry (SC IPA). Algorithms for preparing and improving images of topological layers, algorithms for identifying and classifying topology objects and their structural components, algorithms for constructing layer topology description, algorithms for analyzing the topology of a layer and determining its defects. A new image stitching algorithm, which doubles the speed of constructing a complete layer image by using the algorithm for finding key points and describing SIFT descriptors, which is optimal for the considered examples of topologies, both in speed and inaccuracy. SC IPA is designed for processing and analysis of VLSI topological layers images and preparation for topology restoration, which reduces the labor intensity of VLSI designers.

[MEKB 23]

The technology for parallel modeling and verification of digital electronic devices and verification of the VLSI topology based on a system of user rules. Software tools for the analysis and parallelization of test tasks and their remote execution on the computing resources of a computing cluster or in a grid environment have been developed on the basis of the technology. The developed technology and software can be used to replace expensive foreign analogs and can reduce costs and time for developing a VLSI topology due to a multiple reduction in the time spent on VLSI modeling and topology verification processes. The verification time for the process of verifying the topology based on user rules on a test device was reduced by more than 5 times.

[IK704]

Methodology for analyzing and verifying the program for autonomous flight of a highly dynamic unmanned aerial vehicle (UAV) with predictive control based on predictive models with regard to the possibility of its implementation, taking into account the dynamic characteristics of the unmanned aerial vehicle and the parameters of the inertial navigation system and the atmosphere state along the route. The developed model, the methodology for analyzing and verifying the program and algorithms for calculating predictive motion models are the basis for the implementation of experimental software for analyzing the feasibility of an autonomous controlled flight of a highly dynamic UAV and assessing the accuracy of its performance under various conditions of the atmospheric state, initial launch conditions and the scatter of model dynamic and structural characteristics and parameters.

Algorithms for a strapdown inertial navigation system (SINS) with Rodrigues-Hamilton parameters for an unmanned aerial vehicle (UAV) based on solid-state accelerometers and solid-state angular velocity sensors. The analysis of the errors’ influence in the computational part of the algorithms for the calculated orientation of the UAV and its location has been performed.

An experimental software that implements SINS algorithms and the software model for navigation calculations.

The software complex for modeling the operation of the GFOS, which allows simulating the operation of an on-board camera located on a UAV moving along a given trajectory at a given speed and orientation over a given territory at a given height. Modeling takes into account the relief and uses satellite imagery coverage of the area. Lighting, fog and other atmosphere characteristics that affect the formation terrain image via the onboard camera are simulated, the effects caused by UAV airframe oscillations and vibrations are simulated as well.

[FM7-10, Start5]

The complex of algorithms for processing and analyzing images of layer-by-layer representation of microelectronic devices: the algorithm for extracting contours, segmentation algorithms that do not use a threshold for processing gray-scale images of digital integrated circuits layers. The boundaries are selected and the found contours are approximated with straight line segments sequentially by means of the Hough transformation. At the next stage, the image is covered with a set of elementary regions. Next, the found elementary regions are merged and, ultimately, the image is replaced with a vector set of regions. This approach can handle highly noisy and distorted grayscale images.

The automated system of layer-by-layer topology reconstruction from a color digital photographic image of an integrated circuit crystal.

[task 05.06 "To develop the automated system for layer-by-layer topology restoration from a color digital photographic image of an integrated circuit crystal" (2001–2005, GR No. 2003360]

Computer control system for printed circuit boards.

[task 06.01 "To develop the computer system for monitoring printed circuit boards" (2001-2005, GR No. 20011531) of the SSTP "Information Technologies"]

Software tools for classification of topology defects detected as a result of automatic control of the semiconductor wafers topology.

[task 3-01 "To develop the software for the classification of topology defects detected as a result of automatic control of the semiconductor wafers topology" (2006–2010, GR No. 20065363)]

The system for digital image processing of elements samples and assemblies of electronic products.

[task 01.06 "To develop the system for digital image processing of elements samples and assemblies of electronic products" of the SSTP "Informatics" (1999-2000, GR No. 20002282)]

Hardware and software complex for combining matrix ICs on a silicon wafer.

[R&D "To develop the hardware and software complex for combining matrix ICs on a silicon wafer" as part of the main task 07.02 "To develop the technology for combining matrix ICs into a system on a silicon wafer" (2001-2005, GR No. 20015042)]

[«Image processing in IC design and production.pdf» (monograph)]

The software complex for controlling critical dimensions control equipment based on machine vision systems (SC CCDCE).

[IIT16/1-03]

An experimental sample of a neural network system for monitoring the state and behavior of spacecraft subsystems based on telemetric data for a ground command-measuring complex.

[6MSG/13-222]

[IK403, F11OB-071]