Backward Reconfiguration Management for Modular Avionic Reconfigurable Systems

A major asset of modern avionic systems is the dynamic reconfiguration in response of component faults or unusual external events. The system-level reconfigurations often show the "act-before-sense" feature: a component or a part action would directly be triggered by an unusual event before the reconfiguration management unit, and each component follows its own reconfiguration rules and exerts different effect on the system. Such type of reconfiguration is called as backward reconfiguration. This paper focuses on the issue of system reconfiguration management, and proposes a novel architecture, a development workflow, and a modeling methodology aimed the challenges and imperfections for backward reconfiguration systems. We describe the practical solutions that support the affirmations discussed regarding the management of reconfigurable systems based on the backward reconfiguration assumption, which include: 1) reconfiguration representation model formulation; 2) degradation detection; and 3) system evaluation and validation. The solutions enable the coverage of reconfiguration actions, the straightforward detections of the current state, and provide information to evaluate the reconfiguration designs. An application study of an avionic computational system is presented to demonstrate the proposed workflows and validate the framework.