Real-time Scheduling and Resource Managementfor Energy Autonomous Sensors

International audience; Energy autonomous sensors are attracting increasing attention because they have potential to provide monitoring for long-time and continuous measurement in many application areas such as transport and health. Energy harvesting can guarantee energy autonomy also called energy neutrality over long period of time if adequate power management strategies are implemented in sensors. As most of them have to execute multiple time critical tasks which access to shared resources in mutual exclusion, classical scheduling strategies should be revisited to provide efficient energy aware solutions. The ED-H scheduler was proved optimal to guarantee energy neutrality of uniprocessor real-time energy harvesting systems when all the tasks are assumedindependent. In this paper, we propose to refine it and we introduce a new scheduling scheme, namely ED-H/DPCP which is adapted to DPCP, the well-known Dynamic Priority Ceiling Protocol to support resource access control. The paper provides the schedulability analysis attached to ED-H/DPCP. We also discuss the implementation and we present integration of ED-H/DPCP within a cobotic application.