Organic Computing
Organic computing refers to an interdisciplinary research initiative with the explanatory concern is to gain a better understanding of organic structures and the development goal of an organically structured information technology. The research initiative refers to the biological paradigm of self-organized information processing in organic systems. Biological organisms exhibit specific phenomenal features, so-called self -x properties that are cited as quality features an organically structured information technology. The realization of the desired quality characteristics should be based on the concept of self-organization. This basic methodological positioning implies in particular a move away from the algorithmically organized information technology
Among other things - motivated by the challenges to computer science with respect to the design of technical systems that will be present from about 2015 - answers to the problems of the anticipated technology developments elaborated. The progressive miniaturization and increasing the efficiency of micro-and nanoelectronic systems cause future a number of intelligent systems exist that provide their services in dynamically changing operating environments. Over a wide variety of communication systems information is exchanged, it inevitably creates networks of intelligent systems, whose behavior is not completely predictable. The individual components must therefore be able, even in unexpected situations to meaningfully respond, ie technical systems will be able to vote in their behavior with each other and with the environment and must - they adapt and they organize themselves
Self -x properties
Self-organization is in itself not a quality characteristic, because a self-organized system configuration does not necessarily correspond to the technical objectives. Uncontrolled self-organization processes can sometimes have quite dramatic consequences. Self-organized systems should therefore remain manageable. They should not develop behaviors that run counter to the desired functionality and the current requirements of particular human user. The question concerning the technology of the future is not therefore whether self-organized systems arise, but how we will make this.
Needed systems that can adapt to different operating conditions and functional requirements through a sufficient number of degrees of freedom to compensate for the misconduct of components through appropriate measures, taking into account in particular the needs of human users and support a total of people in his life circumstances trustworthy in many ways. There are system architectures required, which are characterized in particular by its robustness and flexibility. Because of the life-like behavior of such systems they are called organic. The future organic computer science systems will comprise a series of so-called self -x properties, they are particularly
- Self-configuring,
- Self- optimizing,
- Self-healing,
- Self-protecting and
- Self-explanatory.
After the presentation of the vision of Organic Computing in a position paper of the GI and ITG / VDE the establishment of the DFG priority program 1183 Organic Computing gave this new research direction substantially boost. In the current 18 research projects 2011 Characteristics of self-organizing systems are studied fundamentally by the year, appropriate system architectures are designed and developed tools that support the design and management of these systems varied.
The area of Organic Computing has strong links with the Autonomic Computing initiative by IBM, however, focuses on mastering the complexity of large server architectures by generating self -x properties. However, Organic Computing observe, in particular the resulting by local interaction behavior of the overall system in which unforeseen global effects can occur. Since not every self-organizing resultant (also called emergent ) behavior is desired, Organic Computing aims to find a form of controlled self-assembly, as is achieved among other things by the generic observer / controller architecture.
Observer / Controller Architecture
In the field of organic, autonomous and autonomous - related systems there are in addition to the observer / controller architecture, a number of other architectural approaches to support self-organizing, adaptive behavior. Foremost among these are particularly
- The MAPE cycle, with the Monitor, Analyze, plan and execute the main steps of an Autonomic describes element and thus a central concept of autonomic computing is
- The operator / controller modules developed by the DFG Collaborative Research Center 614 which have their main use for self -optimizing systems of mechanical engineering,
- Within the framework of the project Organic Fault - Tolerant Control Architecture for Robotic Applications designed Organic Control Units ( OCU ) that allow a self-organized behavior of 6- legged walking robot,
- The SPA architecture (sense, plan and act ) in robotics or the 3- tier architecture (component control, change management and goal management ) from the Software Engineering for Adaptive Systems,
- Agent-based approaches, such as (for example ) or in self-assembling factories (eg ) are used for example in robot soccer, as well as
- Autonomous system-on- chip architectures (eg ).
In the Observer / Controller architecture, the main components of this architecture approaches are merged into a generic concept in which particular aspects of machine learning are considered.
- Various observer / controller architectures
Multilayer
Hierarchic
Distributed
The Observer / Controller architecture observed, analyzed and evaluated with respect to a given objective criteria in a control system regulating the behavior of the monitored systems. This leads to the selection of appropriate measures in order to influence future behavior in the desired direction. The architecture consists of a network of autonomous units ( called production system), supplemented by one or more observer and controller units. For the Observer an appropriate methodology must be developed in order to observe the (global ) system behavior and analyze for the occurrence of emergence effects and evaluate. A Shannon entropy based method for the quantitation of emergence is proposed. The controller is to decide in what form the production system must be manipulated to allow for a controlled self-organized global behavior within the limits and objectives, from an external unit ( the environment) are specified based on the results of the Observer. The controller should thus be able to improve its behavior learning, ie adapt its behavior in particular due to experience of the impact of previous actions. For more information on the generic framework is referred to.
Robustness and flexibility
In addition to the quantification and controllability of emergent behavior, the demand for robustness and adaptivity of an organic system is considered as a major challenge: How to be designed a system so that it may be in its functionality to adapt to changes in the operational environment and react robust at the same time, ie its functionality despite changes in environmental parameters continue to be met? How can system properties such as robustness, flexibility, autonomy and self-organization be quantified? Approaches for enhanced quantitative analysis can be found among others in
Until the implementation of the formulated in the vision of Organic Computing system requirements, it is still a long way. However, in the DFG Priority Programme 1183 Organic Computing already developed important concepts and has been achieved partial results. It is important to supplement this basic research program by other, more application-oriented research projects to test the lessons in the priority program of theoretical and methodological knowledge in specific technical systems and develop.
Current Research
First steps towards organic, self-organizing and adaptive systems are made in the DFG Priority Programme 1183 Organic Computing. Here, a number of issues to be considered, in addition to other aspects include the following priorities: Adaptivity, reconfigurability, emergence of new system properties and self-organization in technical systems.