Abstract :
For distributed safety-critical systems, such as avionics and automotive, shared networks represent a bottleneck for timing predictability, a key issue to fulfill certification requirements. To control interferences on such shared resources and guarantee bounded delays, the Time Division Multiple Access (TDMA) protocol is considered as one of the most interesting arbitration protocols due to its deterministic timing behavior and fault-tolerance features. This paper addresses the problem of computing the worst-case end-to-end delay bounds for traffic flows sharing a TDMA-based network using Network Calculus. First, we extend classic timing analysis to integrate the impact of non-preemptive message transmission and various service policies in end-systems, e.g., First In First Out (FIFO), Fixed Priority (FP) and Weighted Round Robin (WRR). Afterwards, the proposed models are refined using Integer Linear Programming (ILP) to obtain tighter end-to-end delay bounds. Finally, this general analysis is illustrated and validated in the case of a TDMA-based Ethernet network for I/O avionics applications. Results show the efficiency of the proposed models to provide stronger guarantees on system schedulability, compared to classic models.
Keywords :
computer network reliability; electronic messaging; fault tolerance; integer programming; interference suppression; linear programming; local area networks; telecommunication traffic; time division multiple access; I-O avionics application; TDMA-based Ethernet network; TDMA-based network timing analysis; arbitration protocol; distributed safety-critical system; fault tolerance feature; guarantee bounded delay; integer linear programming; interference control; network calculus; nonpreemptive message transmission; resource sharing; time division multiple access protocol; traffic flow; Aerospace electronics; Calculus; Computational modeling; Delays; Protocols; Time division multiple access; FIFO; Fixed Priority; Integer Linear Programming; Network Calculus; Scheduling; TDMA; Weighted Round Robin;
Conference_Titel :
Modelling, Analysis & Simulation of Computer and Telecommunication Systems (MASCOTS), 2014 IEEE 22nd International Symposium on