Ensuring Deployment Predictability of Distributed Real-time and Embedded Systems

The dynamic deployment and configuration (D&C) of components in response to environmental changes or system mission mode changes is essential to facilitate runtime resource allocation for component-based distributed real-time and embedded (DRE) systems. This paper provides several contributions to the study of predictable D&C for component-based DRE systems. First, we describe how the predictability of component-based D&C can be affected by application dependency relationships and priorities. Second, we describe how a multigraph algorithm called partial priority inheritance via graph recomposition (PARIGE) can improve D&C predictability. Third, we empirically evaluate the effectiveness of PARIGE on a representative DRE system based on NASA Earth Science Enterprise’s Magnetospheric Multi-Scale (MMS) mission system. The results show that PARIGE can avoid unbounded deployment time priority inversion when component assemblies with different priorities have complex dependencies among each other, thereby significantly improving the responsiveness of mission-critical tasks with higher priorities.