Abstract

The use cases for CPSs range from industrial automation over automotive to search-and-rescue applications. Nowadays these CPSs work either in static networks, like in production lines, or isolated and mobile, as for example UAVs. The cooperation of mobile CPSs is only possible with very relaxed real-time requirements. For the tight cooperation of mobile CPSs new techniques are needed. The special challenge in such networks is that the communication needs to guarantee hard timing boundaries but also needs to be flexible enough to adapt to changes within the network. In this work we present an architecture that copes with these networks and their challenges. It consists of four main components: a time synchronization, a real-time networking stack, a scheduling algorithm and a management protocol. As cooperation between mobile CPSs requires feedback loops to be closed via the wireless links, the time synchronization needs to be accurate between several CPSs. To be able to time the execution of tasks as accurate as possible we present a sub-microsecond time synchronization. By utilizing our drift compensation, CPSs can make use of low-cost crystal oscillators without loosing timing accuracy. To make use of such an accurate time synchronization, we present a real-time network stack that incorporates not only the scheduler for the communication but also the scheduler for the execution of tasks. Thus, the jitter a feedback loop experiences is kept minimal. To support the adaption to changes in the network we designed all operations in a way that they introduce a minimum amount of jitter. This adaption to changes is one of the key requirements to the scheduling algorithm. As the adaption must happen without harming the timings of running real-time application, a novel scheduling approach is necessary. To fulfill this requirement we introduce a MILP-model to calculate schedules for the presented real-time network stack. As solving MILP-models is computational complex and CPSs often have only limited computational power, we introduce a heuristic to calculate these schedule with far less effort. To disseminate schedules to CPSs, we evaluate the applicability of Concurrent Transmission (CT)-protocols. All previous research on CT was done on similar hardware. We investigate whether the results of this research can be generalized and point out the differences and similarities. Further, we frame the challenges heterogeneous CT networks have to overcome.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call