Full title: Organic Computing with Artificial DNA for Reliable Dynamic Systems based on Semantic Models and Evolutionary Algorithms for Fault Diagnosis and Adaptation

Type: DFG

Webpage link:

Duration: 01/01/2021 to 31/01/2024


Organic Computing leads to significant advantages for complex dynamic systems like reduced development efforts, increased adaptability and robustness. However, for safety-critical systems which have to maintain functionality even in the presence of faults or failures (fail-operational) further properties are necessary. This includes the maintenance of the major core functionality even if non redundant system resources fail, the organic computing run-time environment is harmed or the remaining resources are insufficient to maintain all services. These failure scenarios require semantic knowledge of the system combined with fault-diagnosis and adaption techniques to properly degrade and reconfigure the system. The proposed research project addresses the corresponding research gaps and their interactions based on artificial DNA: (1) semantic description methods for organic computing systems based on artificial DNA, (2) diagnosis techniques with a high level of automation for organic computing systems using artificial DNA, and (3) adaptation techniques for such systems in highly safety-critical applications. Semantic description methods for organic computing systems with artificial DNA are the foundation for higher semantic-based failure detection and adaptation techniques. Diagnosis techniques for organic computing systems with artificial DNA can exploit the semantic descriptions to automatically build diagnosis models. Furthermore, these models can be optimized by evolutionary algorithms to improve their failure detection rates. Adaptation techniques modify the artificial DNA based on the recognized failures and the semantic description to realize the reconfiguration and degradation concepts. Within the project, the developed models and algorithms will be prototypically implemented and evaluated using sample scenarios and failure injection experiments.