Towards Efficient Transient Fault Handling in Time-Triggered Systems

Abstract

Transient communication faults in distributed control systems (DCS) are unavoidable but must be handled adequately in order to enforce correct system behaviour. A typical way of handling transient faults is temporal redundancy by means of retransmissions. However, DCS are frequently designed with time-triggered architectures, being scheduled offline and not coping efficiently with retransmissions as these require the pre-allocation of bandwidth that, in the absence of errors, is wasted. In this paper we propose using the Flexible Time-Triggered paradigm to reconcile the Time-Triggered model with on-line scheduling of retransmissions when needed, only, leading to an efficient bandwidth usage. This is confirmed with preliminary experimental results obtained on an FTT-CAN network.