Projects / Programmes
Computer Structures and Systems
January 1, 2004
- December 31, 2008
| Code |
Science |
Field |
Subfield |
| 2.07.00 |
Engineering sciences and technologies |
Computer science and informatics |
|
| Code |
Science |
Field |
| P170 |
Natural sciences and mathematics |
Computer science, numerical analysis, systems, control |
information technologies, design automation, computing structures, system-on-chip, embedded systems, testing, optimization methods
Researchers (13)
| no. |
Code |
Name and surname |
Research area |
Role |
Period |
No. of publicationsNo. of publications |
| 1. |
11983 |
PhD Anton Biasizzo |
Computer science and informatics |
Researcher |
2004 - 2008 |
149 |
| 2. |
17040 |
PhD Uroš Kač |
Computer science and informatics |
Researcher |
2004 - 2008 |
35 |
| 3. |
22314 |
PhD Peter Korošec |
Computer science and informatics |
Researcher |
2004 - 2008 |
238 |
| 4. |
10824 |
PhD Barbara Koroušić Seljak |
Computer science and informatics |
Researcher |
2004 - 2008 |
339 |
| 5. |
29530 |
PhD Uroš Legat |
Computer science and informatics |
Junior researcher |
2008 |
24 |
| 6. |
25621 |
Marko Mali |
Computer science and informatics |
Junior researcher |
2005 - 2006 |
23 |
| 7. |
05601 |
PhD Franc Novak |
Computer science and informatics |
Head |
2004 - 2008 |
316 |
| 8. |
18291 |
PhD Gregor Papa |
Computer science and informatics |
Researcher |
2004 - 2008 |
352 |
| 9. |
15671 |
PhD David Podgorelec |
Computer science and informatics |
Researcher |
2004 |
214 |
| 10. |
09862 |
PhD Jurij Šilc |
Computer science and informatics |
Researcher |
2004 - 2008 |
359 |
| 11. |
28488 |
PhD Katerina Tashkova |
Computer science and informatics |
Junior researcher |
2007 - 2008 |
28 |
| 12. |
11972 |
PhD Drago Torkar |
Computer science and informatics |
Researcher |
2004 - 2008 |
91 |
| 13. |
26473 |
MSc Mariusz Jerzy Wegrzyn |
Computer science and informatics |
Junior researcher |
2006 - 2008 |
6 |
Organisations (1)
| no. |
Code |
Research organisation |
City |
Registration number |
No. of publicationsNo. of publications |
| 1. |
0106 |
Jožef Stefan Institute |
Ljubljana |
5051606000 |
90,724 |
Abstract
New silicon technologies allow implementation of complex devices composed of heterogeneous cores, massive memory, communication networks, electro-mechanical sub-assemblies and intelligent sensors, which requires new multidisciplinary design approaches and introduces numerous problems in design and test. The proposed research programme addresses these problems and is primarily focused on the subjects in which we have gained experience in our earlier research work.
A processor core is one of the basic blocks of a modern system-on-chip (SoC), therefore the basic knowledge of processor architectures is imperative for the proposed research programme. Following the progress in this field for a number of years, we concentrate on the problem of task scheduling and resource allocation in multithreaded systems. In the area of embedded systems we address problems of dynamic reconfiguration in real-time. We aim to explore the potential for rapid embedded reconfiguration in terms of predictability, reliability and power consumption searching for adequate technologies, such as real-time operating systems, programming languages, hardware platforms for experimental work and prototyping.
High-level systhesis faces complex combinatorial problems including parameters that are mutually dependent. Consequently, optimization of the design is often subject to analysis of extensive space of possible solutions. A typical example is concurrent resource allocation of the computing structure synthesized for a given target algorithm. Conventional approaches are often non-adequate, therefore we investigate alternative approaches based on metaheuristic algorithms.
System-on-Chip (SoC) design integrates large reusable blocks (i.e. cores) that have been designed and verified in earlier applications in practice. The cores put together in a SoC provide a wide range of functions, like CPUs, DSPs, interfaces, controllers, memories, etc. and normally originate from different core providers which makes the problem of testing rather challenging to the SoC designer. Technical problems (i.e., testability problems due to the limited access of internal points, built-in self-test design, high frequency signal integrity, etc.) as well as problems related to intellectual property issues initiated development of IEEE P1500 (Standard for Embedded Core Test). In collaboration with Laboratoire d'informatique, de robotique et de microelectronique de Montpellier (LIRMM) we explore possibilities of its extension to mixed-signal devices.
We also develop new approaches to functional test and built-in self-test for digital systems as well as mixed-signal test approaches based on oscillation test technique optionally supported by IEEE 1149.4 test infrastructure.
Significance for science
The research was concerned primarily with the design and test automation of computing structures and systems. We are concentrating particularly on metaheuristic approach to engineering design and logistics problems. Metaheuristics approach to combinatorial optimization problems is a rapidly growing research topic. This is due to the importance of combinatorial optimization problems for the scientific as well as the industrial world. In our research we have mainly aimed to solving problems from computer design and logistics. These problems can be intractable because of the wide solution space as well as multiple constraints and objectives. We have developed efficient self-adapting evolutionary and ant-stigmergy-based optimization algorithms. These approaches were used for solving various combinatorial and numerical optimization problems. Evolutionary algorithms were tested on constrained numerical optimization problems. Multiple ant-colonies approach can be successfully used to solve mesh-partitioning problems that arise in mechanical, civil, automobile, and aerospace engineering. The multilevel ant-stigmergy approach was applied in solving discrete numerical optimization problems. We proposed a novel general approach to the transformation of a multi-parameter optimization problem into finding the minimum path problem. We have also developed a differential ant-stigmergy approach suitable for solving discrete as well as continuous numerical optimization problems. Using the evolutionary optimization methods we have been solving the following problems: location of the shortest path, graph partitioning, balancing of healthy meals (multiconstrained knapsack problem), concurrent operation scheduling and functional unit allocation in the process of integrated-circuit design, optimization of the universal motor's geometry, and test pattern generator structure design. Some of the research results have been published in the acknowledged international journal Parallel Computing. In the field of electronic test we explored conditions for the oscillation based test of switched-capacitor biquad filter stages. Relations between parameters describing the filter circuit in continuous time domain and the coefficients of the discrete time transfer function were studied and used to determine necessary conditions for establishing sustained oscillations in the tested circuit. Reconfiguration scheme based on the transformation of the biquad filter stage to a quadratic oscillator was derived and evaluated in Spice simulations. In collaboration with Electronic ceramics department we developed an approach to fault diagnosis of a piezoresistive ceramic pressure sensor based on a finite-element model. We also addressed the security problem of systems incorporating IEEE Std. 1149.1 infrastructure. We developed a security scheme based on a locking mechanism that prevents unauthorized users to access system via standard test access port. We have developed an approach for functional testing of processor cores suitable for built-in self-test. We generate a test sequence that allows arbitrary situations that might occur in practice and consequently detects faults that only appear in a particular sequence of events. This is accomplished by using a test sequence that explores the functionality of each individual instruction and is composed in such a way that it forms a sensitive path, which can be executed more than once, each time with a different input pattern. Experimental case studies performed on Xilinx PicoBlaze and MicroBlaze procesor core confirmed the efficiency of the approach. Within the 6 FP project ARFLEX, we investigated the possibility to use vision sensors for robot control. The objective was to radically innovate industrial robots where these technologies have not yet find full applications. We developed, tested and the vision-based embedded system for closed-loop robot control which will enable a 3D trajectory tracking with high pr
Significance for the country
We have developed efficient optimization algorithms. These approaches were used in several real-world applications. In collaboration with Domel d.d., Železniki, we reduced the production costs of an electro-motor and optimized an aerodynamic power of dry vacuum cleaner impeller. Within the project "The role of Luka Koper in logistic support of the Slovenian Armed Forces and allies" in the frame of the Target research programme (CRP MIR) "Science for Peace and Security 2006-2010" we developed software components for optimization of logistic procedures of equipment and material transportation. Within the project "Secure infrastructure for implementing command and control" in the frame of Technology program (TP MIR) "Technology for Peace and Security 2006-2012" we optimized the structure of an absorber used in telecommunication systems. We applied evolutionary approach also in other problem domains. In collaboration with AET Tolmin we studied rheological properties of ceramic-paraffin suspensions and developed a method for dynamic viscosity prediction with regard to their composition using artificial neural networks. For Alfastreet gaming systems we have developed a secure data storage unit based on new ferroelectric semiconductor memory devices. The storage unit has been designed with the emphasis on data security, system availability and system reliability. High system availability was achieved by the use of recent non-volatile memory technologies based on Magneto-resistive effect (MRAM) and the implementation of Error Correction Code (ECC) technique in a separate hardware core. For the purpose of system testability we explored the possibility of implementation of IEEE Std 1149.1 and IEEE Std 1500 test infrastructure. In an experimental case study we estimated the resources required for local processing of test results within a IEEE 1500 test wrapper. Special attention was paid to the security aspects of the design. Since IEEE Std 1149.1 is known to be vulnerable to hackers, an extension of IEEE Std 1149.1 locking mechanism was developed. In the research area of computer-based menu planning, collaborating with the Department of Environmental Sciences (JSI), CINDI Slovenia and Slovene Society for Clinical Nutrition we performed in the following activities: • We upgraded the web application Optijed with a module for planning nutrition for children, adolescents and patients with special nutritional needs. In addition, we extended the food composition database with new data and described data by using the standardized LanguaL method; • We finished the the project named “Nutrition for special needs”, in which we analysed and optimized menus for people working in extreme conditions; • In the project named “Slovene Web-based Food Composition Tables”, we associated with the European network of Excellence on Food Composition Databank systems EuroFIR (European Food Information Resource Network FP6-CT-2005-513944). In the project named ”Enhancing Elementary Mathematics Pedagogy Competencies by e-Learning” that is part of a large ESS project “Evaluation of the Slovene Education”, we developed a web application for e-learning grammar school mathematics. The decision making system, which we formalized as a combinatorial optimization problem of graph colouring, is based upon a metaheuristic optimization method. In scope of multiconference Information Society we organized three biennial international conferences on “Bioinspired Optimization Methods and their Applications - BIOMA 2008”. Areas of interest at the conference included genetic algorithms, evolution strategies, evolutionary programming, genetic programming, ant colony optimization, particle swarm optimization and related bioinspired methods, and their applications in science, engineering and business. Members of our research programme actively participated in the international Program committee and Organizing committee of the conference.
Most important scientific results
Final report,
complete report on dLib.si
Most important socioeconomically and culturally relevant results
Final report,
complete report on dLib.si
