Projects / Programmes
Application of evolutionary models to assure fault-tolerance in embedded systems
| Code |
Science |
Field |
Subfield |
| 2.07.00 |
Engineering sciences and technologies |
Computer science and informatics |
|
| Code |
Science |
Field |
| P176 |
Natural sciences and mathematics |
Artificial intelligence |
| P170 |
Natural sciences and mathematics |
Computer science, numerical analysis, systems, control |
evolutionary computing, embedded systems, artificial intelligence, fault-tolerant control systems, data-mining, fault detection
Researchers (2)
| no. |
Code |
Name and surname |
Research area |
Role |
Period |
No. of publicationsNo. of publications |
| 1. |
08067 |
PhD Ivan Rozman |
Computer science and informatics |
Head |
2004 - 2006 |
822 |
| 2. |
20199 |
PhD Matej Šprogar |
Computer science and informatics |
Researcher |
2004 - 2006 |
105 |
Organisations (1)
Abstract
Embedded systems must be dependable, especially in case of errors and faults, as they're used in time- and safety-critical environments. The ever-growing demands for functionality of embedded systems can only be satisfied by increased complexity in both hardware and software, what increases the probability of faults and failures in system behavior. We need a control mechanism that monitors the working system; it should detect and signal any fault and trigger according action. The observed faults can be handled by redundancy or by reconfiguration of existing system; the latter is, if applicable, often also less expensive.
The project aims at research and development of evolutionary computer models and tools that aid the design of fault-tolerant embedded systems. Existing designs all have some problems that can be solved by applying special machine-learning techniques; this way we can build suitable control software, which will replace the expensive (redundant) hardware. The main idea is to evolve monitoring modules, that work in parallel and monitor the behavior of the original system. These modules will signal and take corresponding actions if a fault is detected. Moreover, they could also replace the failed component and provide the needed response. They are simple software modules that simulate the system behavior and will be automatically generated based on the collected data from the running system. The real system that includes such software control modules is likely to be more fault-tolerant. Successful modules could even be implemented in hardware. The project depends on the data-mining and evolutionary computation techniques (collecting data and search for information) and requires knowledge in design and implementation of appropriate software and hardware control modules.
