Dependability (FI, reliability evaluation, safety, fault-tolerance algorithms, diagnosis)

Embedded systems need to deliver an acceptable level of service despite the occurrence of transient and permanent hardware faults, design faults, imprecise specifications, and accidental operational faults. A system must be resilient with respect to unanticipated behavior from the environment of the system or of subsystems. In case such unanticipated behavior occurs, the system should still exhibit some sensible behavior, and not be completely unpredictable. In particular, insufficient support exists for building reliable systems out of unreliable components. In particular, technologies are required for tackling the significant increase of transient failure rates, which results from trends in the semi-conductor industry, such as shrinking geometries, lower power voltages and higher frequencies.

The expertise includes:

  • Fault-tolerance techniques (e.g., error detection, fault isolation, fault masking, transformation of failure modes, agreement protocols for consistent distributed computing base, end-to-end reliability across heterogeneous networks)

  • Analytical dependability evaluation (e.g., formal verification using model checkers SAL and UPPAAL, reliability modeling using Mobius, ...)

  • Experimental evaluation using fault injection (e.g., network fault injection in TTEthernet, chip-level fault injection in FPGAs using dynamic partial reconfiguration, ...)