Towards Mixed Criticality Industrial Wireless Sensor-Actuator Network

In industrial Internet of Things, feedback control loops employed over wireless sensor-actuator network (WSAN) for various process monitoring and control applications require real-time communication for stability. In the real world, most complex control systems are, de facto, Mixed-Criticality (MC) system, meaning that all control loops are not equally critical for the system’s correct operation. While the notion of mixed-criticality has been studied widely in CPU scheduling, it still remains largely unexplored for wireless domain. For MC CPU scheduling, the key challenge stems from the uncertainty of worst-case execution times, while the uncertainty in WSAN comes from unpredictable channel conditions and plant dynamics. In this paper, we formulate the MC scheduling problem, formally define the MC semantics for WSAN, and propose MC real-time scheduling in multihop WSAN that allows co-scheduling of the loops for handling dynamic criticality changes. This proposed approach exploits the capture effects of the radios for dynamic resource allocation and reclamation when criticality changes. Then, by exploiting the unused channel capacity of WSAN, we propose a technique to minimize redundancy in high criticality control loop scheduling while preserving the communication reliability and MC constraints, thereby enhancing MC schedulability.