The Controller Area Network (CAN) is an event-triggered protocol that is widely used in distributed real-time embedded systems. It has been demonstrated that a “Shared-Clock” (S-C) scheduling protocol can be used on top of CAN hardware to implement time-triggered network operations. Previous work in this area has led to the development of five different time-triggered S-C scheduling protocols referred to as: TTC-SCC1, TTC-SCC2, TTC-SCC3, TTC-SCC4 and TTC-SCC5 schedulers. This paper develops mathematical models for assessing message latencies between all communicating nodes in the different S-C scheduling protocols. In particular, the paper provides mathematical equations for estimating Master-to-Slave, Slave-to-Master, and Slave-to-Slave message latencies in all five schedulers. The paper then presents a small case study to allow a practical comparison of the communication behavior in the various S-C schedulers considered. The results show that the communication behavior, especially Slave-to-Slave message delays, can be improved significantly when TTC-SCC3, TTC-SCC4 and TTC-SCC5 scheduler implementations are used. The results also show that even a small selection of S-C scheduler implementations demonstrates a wide range of different patterns of behavior. It is therefore suggested that selection of the most appropriate scheduler will largely depend on requirements of the application for which the scheduler is intended.

scheduler, architecture, Master, Slave, jitter, time-triggered, co-operative, Shared-Clock, Controller Area Network, message latency