Digital On-Demand Computing Organism: Stability and Robustness
Zusammenfassung der Projektergebnisse
An intrinsic feature of many biological systems is their capabilhies of self-healing, self-adapting, self-configuring, etc., or short, Self-X features. In contrast, today's computing systems hardly feature any of these characteristics even though their complexity has risen to an amount where this very complexity is increasingly hampering maintenance tasks such as optimization, protection, and repair. In the first phase of the project, which was funded within the frame of the SPP 1183 (Organic Computing), we settled the foundation for this new architecture. It consists of three levels, brain, organ and cell. Beginning with the brain level, a software architecture for a robot controller with emphasis on Self-X features was proposed. It closely interacts with an organic middleware at the organ level, featuring an artificial hormone system for reliable and robust self-organization. At the cell level, a novel adaptive and dynamically reconfigurable hardware architecture was designed to implement the Self-X features in an efficient way. Power management co-ordinates the levels for ultra-low power system efficiency. All levels are supplied with monitoring techniques and architectures as a prerequisite for enabling Self-X features. On cell level, the design of reconfigurable organic processing cells was guided by mechanisms found in biological cells. All cells follow the same blueprint and, using an DNA-inspired mechanism, can specialize for an assigned task. The efficient passing and distribution of artifical hormones was a key requirement for the communication infrastructure of these cells. The resulting hardware architecture was prototypically and successfully implemented into a 2x2 grid of basic OPCs. On organ level, a biologically motivated system control resulting in a basic organic middleware was developed, featuring an artificial hormone system for system configuration. Using these artifical hormons, the middleware groups neighbored organic processing cells to so-called virtual organs working on common tasks. By a prototypic simulator, the suitability for using these biologically-inspired methods for task distribution in an adaptive architecture was demonstrated. To enable closed control loops and self-organization found in biology, data from different system levels has, to be collected and made available, e.g. to the middleware to adapt hormone levels. Hence, a distributed and layered monitoring concept was developed featuring data acquisition and processing techniques required for corelation and analysis of monitoring data. Software and hardware monitor prototypes were developed to explore the design space of such monitoring infrastructures and prove the suitability, ofpotential techniques. Efficient energy management is a key issue in biology. Most biological entities are able lo efficiently adapt energy consumption according to current needs such as sleeping, normal activity, or fight and escape situations. Ultra Low-Power processing within this project is inspired by these ideas: closely cooperating with middleware and monitoring it is responsible for energy conservation to increase operational availability and prolongue the life expectancy of the large number of organic processing cells. In stress situations (e.g. lough real-time deadlines), high energy consuming alternatives can be temporarily made available. Based upon initial concept analysis, a simulation environment was implemented and a first prototype evaluated. On brain level, the basic components of an intelligent organic robot-control system were developed. Here, a main necessity is the automatic configuration of the motion control system to the mechanical structure of the robot. Having this, it is possible for the robot to autonomously adapt to e.g. changing tools like a human can do. Furthermore, this is a basis to compensate mechanical problems to a certain amount such as e.g. animals are able to compensate for a stiff leg. Through cooperation and interaction of these parts, the basis of an overall organic system - the DodOrg system - was founded. Individual prototypes were created to explore design spaces and enable early evaluation of chosen design aspects. This was an important first step. Now, the subsequent research steps have to be conducted. Therefore, in a new project the individual parts shall be merged into a combined prototype which then will also enable comparison with traditional non-organic architectures. As a key issue, plasticity, dynamics, and stability with respect to self-adaptation, self-optimization and self-healing shall be investigated.
Projektbezogene Publikationen (Auswahl)
- A Monitoring Infrastructure for the Digital on-demand Computing Organism (DodOrg). In Hermann de Meer and James P. G. Sterbenz, editors, Self-Organising Systems: First International Workshop (IWSOS2006), LNCS4124, page 258. Springer Verlag, Berlin- Heidelberg, 2006. ISBN 3-540-37658-5
Rainer Buchty and Wolfgang Karl
- A Network Agent for Diagnosis and Analysis of Real-time Ethernet Networks. In Seongsoo Hong and Wayne Wolf, editors, International Conference on Compilers, Architecture, and Synthesis for Embedded Systems (CASES 2006), pages 65-73. ACM Press, New York, 2006. ISBN 1-59593-543-6
Hans-Peler Löb, Rainer Buchty, and Wolfgang Karl
- Automatic Configuration of the Dynamic Model for Common Industrial Robots. In C. Hochberger and R. Liskowsky, editors. Proceedings of the GI Jahrestagung, GI-Edition Lecture Notes in Informatics (LNI), pages 137-144, Dresden, Germany, October 2006
M. Wenz and H. Wörn
- Automatic Data Locality Optimization through Self-Optimization. In Hermann de Meer and James P. G. Sterbenz, editors, Self-Organising Systems: First International Workshop (IWSOS2006), LNCS4124, pages 187-201. Springer Verlag, Berlin-Heidelberg, 2006. ISBN 3-540-37658-5
Rainer Buchty, Jie Tao, and Wolfgang Karl
- Entwicklung einer selbstkonfigurierenden und selbstorganisierenden Bewegungssteuerung für Industrieroboter. In G. Brandenburg, W. Schumacher, R. D. Schraft, and K. Bender, editors, Tagungsband SPS/IPC/Drives 2006, pages 619-627, Nürnberg, Germany, November 2006
M. Wenz and H. Wörn
- New 2-Dimensional Partial Dynamic Reconfiguration Techniques for Real-Time Adaptive Microelectronic Circuits. In Jürgen Becker, Andreas Herkersdorf, Amar Mukherjee, and Asim Smailagic, editors, IEEE Computer Society Annual Symposium on VLSI (ISVLSI 2006). IEEE Computer Society, 2006. ISBN 0-7695-2533-4
Michael Hübner, Christian Schuck, Matthias Kühnle, and Jürgen Becker
- Rule-based Generation of Motion Control Software for General Serial-Link Robot Manipulators. In Proceedings of the 8th International Workshop on Computer Science and Information Technologies (CSIT), pages 16-21, Karlsruhe, Germany, September 2006
M. Wenz and H. Wörn
- artNoC - A Novel multi-functional Router Architecture for Organic Computing. In 17th International Conference On Field Programmable Logic and Applications, August 2007
Christian Schuck, Stefan Lamparth, and Jürgen Becker
- Automatische Konfiguration der Bewegungssteuerung von seriellen Kinematiken. In J. Gausemeier, F. Rammig, W. Schäfer, A. Trächtler, and J. Wallaschek, editors, Entwurf mechatronischer Systeme, volume 210 of HNI-Verlagsschriftenreihe, pages 57-68, Paderborn, Germany, 2007
M. Wenz and H. Wörn
- Computing Symbolic-Algebraic Models of Industrial Manipulator Arms in a Fully Automatic Way. In K. Schilling, editor, Proceedings of the 13th IASTED International Conference on Robotics and Applications (RA 2007), pages 334-339, Würzburg, Germany, August 2007. ACTA Press
M. Wenz and H. Wörn
- Exploitation of Run-Time Partial Reconfiguration for Dynamic Power Management in Xilinx Spartan III-based Systems. In ReCoSoC'07 Proceedings, June 2007
K. Paulsson, M. Hübner, S. Bayar, and J. Becker
- Reliablity of an Artificial Hormone System with. In Parallel and Distributed Computing and Systems, Cambridge, Massachusetts, USA, November 19-21 200
Uwe Brinkschulte, Alexander von Renteln
- Solving the Inverse Kinematics Problem Symbolically by Means of Knowledge-Based and Linear Algebra-Based Methods. In Proceedings of the 12th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA 2007), pages 1346-1353, Patras, Greece, September 2007. IEEE Press
M. Wenz and H. Wörn
- Towards an Artificial Hormone System for Self-Organizing Real-Time Task Allocation. In 5th IFIP Workshop on Software Technologies for Future Embedded & Ubiquitous Systems (SEUS 2007), 2007
Uwe Brinkschulte, Mathias Pacher, and Alexander von Renteln
