Prof. Dr. Rüdiger Ehlers

Abstract:

We consider the synthesis of reactive systems that are robust against intermittent violations of their environment assumptions. Such assumptions are needed to allow many systems that work in a larger context to fulfill their tasks. Yet, due to glitches in hardware or exceptional operating conditions, these assumptions do not always hold in the field. Manually constructed systems often exhibit error-resilience and can continue to work correctly in such cases. With the development cycles of reactive systems becoming shorter, and thus reactive synthesis becoming an increasingly suitable alternative to the manual design of such systems, automatically synthesized systems are also expected to feature such resilience.

The framework for achieving this goal that we present in this paper builds on generalized reactivity(1) synthesis, a synthesis approach that is well-known to be scalable enough for many practical applications. We show how, starting from a specification that is supported by this synthesis approach, we can modify it in order to use a standard generalized reactivity(1) synthesis procedure to find error-resilient systems. As an added benefit, this approach allows exploring the possible trade-offs in error resilience that a system designer has to make, and to give the designer a list of all Pareto-optimal implementations.

This talk is based on joint work with Ufuk Topcu.