Adaptive and dependable real-time systems

Dynamic resource management enables a system to dynamically adapt to changes in the environment (e.g., on the ground or in-flight in case of an airplane) and to changes in resource demands or resource availability (e.g., power, time (scheduling), communication bandwidth, memory) while ensuring real-time constraints. It enables better resource utilization, improved dependability and can be used in power-aware systems. Due to safety concerns, reconfiguration of safety-critical systems is often reduced to selecting system-wide modes out of statically defined scheduling tables, which provides good analyzability and determinism, though impairing the flexibility of resource management.

One of the aspects that is addressed by our research is to investigate how to perform dynamic reconfiguration in safety-critical systems that need to be certified. In particular, we extend time-triggered technologies (as widely deployed in safety-critical applications) with dynamic reconfiguration abilities. In time-triggered systems, all activities such as the transmission of a message or the execution of a task are controlled by a schedule table that denotes the points in time of these preplanned activities.

One approach is to compute new time-triggered schedules at run-time (in predictable time and with predictable results). For example, we extend time-triggered communication protocols in distributed systems (e.g., Time-Sensitive Networking, TTEthernet) and at chip level.

The picture below shows an example of adaptation and dynamic reconfiguration in a time-triggered system, namely active diagnosis for fault recovery. More information is available at “http://ieeexplore.ieee.org/document/6899153/”

Research activities in this focus area include:

  • Coexistence of safety-critical static subsystems and non-safety-critical subsystems with dynamic resource management

  • Extension of time-triggered platforms with support for dynamic reconfiguration

  • Assured reconfiguration with predictable behavior

  • Bounded time for reconfiguration

  • Continuity of service during reconfiguration

  • Consistent switch to new configuration

  • Robust reconfiguration mechanisms