Temporal data matching in component based real time systems

We consider embedded systems structured as a graph of communicating components [1]. In this model, we define a data matching property based upon the interactions of the data flow paths passing through common components. If a component uses inputs which indirectly depend on the same component output, these inputs have to depend on a same execution step of the producer component, even when several independent paths link these two components, and when the components have different timing characteristics (esp. different periods). We analyze the system architecture to detect situations that can cause data matching problems. Depending on the system needs, we propose several approaches to manage data matching.

[1]  Jane W.-S. Liu,et al.  Maintaining Temporal Consistency: Pessimistic vs. Optimitic Concurrency Control , 1995, IEEE Trans. Knowl. Data Eng..

[2]  Sang Hyuk Son,et al.  Real-Time Databases and Data Services , 2004, Real-Time Systems.

[3]  Song Han,et al.  A deferrable scheduling algorithm for real-time transactions maintaining data freshness , 2005, 26th IEEE International Real-Time Systems Symposium (RTSS'05).

[4]  Friedemann Mattern,et al.  Virtual Time and Global States of Distributed Systems , 2002 .

[5]  Krithi Ramamritham,et al.  Mutual Consistency in Real-Time Databases , 2006, 2006 27th IEEE International Real-Time Systems Symposium (RTSS'06).

[6]  Johan Fredriksson,et al.  Evaluation of component technologies with respect to industrial requirements , 2004, Proceedings. 30th Euromicro Conference, 2004..

[7]  Leslie Lamport,et al.  Time, clocks, and the ordering of events in a distributed system , 1978, CACM.