Programming support for time-sensitive adaptation in cyberphysical systems

Abstract

Cyberphysical systems (CPS) integrate embedded sensors, actuators, and computing elements for controlling physical processes. Due to the intimate interactions with the surrounding environment, CPS software must continuously adapt to changing conditions. Enacting adaptation decisions is often subject to strict time requirements to ensure control stability, while CPS software must operate within the tight resource constraints that characterize CPS platforms. Developers are typically left without dedicated programming support to cope with these aspects. This results in either to neglect functional or timing issues that may potentially arise or to invest significant efforts to implement hand-crafted solutions. We provide programming constructs that allow developers to simplify the specification of adaptive processing and to rely on well-defined time semantics. Our evaluation shows that using these constructs simplifies implementations while reducing developers' effort, at the price of a modest memory and processing overhead.