Maintaining Data Freshness in Distributed Cyber-Physical Systems

Scheduling real-time update transactions to maintain data freshness is an important issue in real-time database system design, especially for data-intensive embedded applications. Despite years of active study, almost all the past work assumes that the jitter of sensor update transactions can be ignored. But actually, this assumption is not practical in real Cyber-Physical Systems (CPSs), and the existing methods cannot be applied in the design of distributed CPSs. In this paper, we propose, to our best knowledge, the first approach to maintain temporal consistency of real-time data objects for dynamic-priority scheduling when transmission delays are considered. We first propose a method called JB-EDF which calculates periods and deadlines for update transactions with significantly lower workload than that by existing approaches. Next, in order to handle more update transactions, we further propose an enhanced solution called JB-EDF* which use JB-EDF as a building block. JB-EDF is more suitable for small scale applications, while JB-EDF* is more attractive for large scale applications. Finally, extensive simulation experiments are conducted. The results demonstrate that JB-EDF and JB-EDF outperform existing solutions in terms of both acceptance ratio and system workload, and the proposed methods are also time-efficient.