Fault diagnosis in complex embedded systems

Locating faults through model-based diagnosis

Product failures do occur, either during testing or normal operation. With the growing complexity of embedded systems, fast, intelligent diagnosis techniques are increasingly important. Broken components must be quickly located so they can be repaired or replaced. However, traditional diagnosis is time-consuming and relies heavily on expert knowledge. 

Model-based diagnosis uses behavioural reasoning based on a system hardware model and actual logging information. By reasoning backwards from anomalies to possible causes, the method automatically points to the component that most likely caused the erroneous behaviour. Improved and applied to electronic components of high-tech embedded systems, model-based diagnosis techniques demonstrate significant reductions in diagnostic effort and time. They allow effective and efficient field services that considerably reduce overall system down-time.

Fault diagnosis in complex embedded systems

Method details


Method description

  1. Identify components and possible faults
  2. Model components’ behaviour, or generate a model automatically, e.g. from a VHDL description
  3. Log system behaviour and detect anomalous behaviour
  4. Use model-based diagnosis to locate most likely faulty components

Research project

Partners

Industrial partner:

  • ASML

Academic partner:

  • Delft University of Technology

Contact

Business manager:

Frans Beenker

Telephone:

+31 (0)88-866 5420

Email:

frans.beenker@tno.nl