The Controller Area Network (CAN) is a serial communications protocol which efficiently supports distributed re-altime control with a high level of reliability. Its domain of application ranges from high speed networks to low cost multiplex wiring [1].
In automotive electronics, control units and sensors are connected using CAN with bitrates up to 1 Mbps. Application in such a harsh environment with widely varying power supplies along with increasing complexity of the control code push CAN reliability to the limit.
There are two message processing engines: interrupt and polling. Interrupts are the preferred method due to the lower message latency on the receiving node, however, polling is often easier to implement.
In this paper, the communication reliability is being investigated using the DCOV process by deriving the transfer function for probability of missing a message as a function of variability of the message timing. The transfer function has been derived by Monte Carlo simulation of two module one-way communication was carried out using MAT-LAB.
This concept was then expanded by deriving the probability transfer function in order to investigate the reliability impact of system complexity with an increased number of CAN messages. The results help to gain understanding of how system wide communication network reliability depends on module variability (hardware and software), which is essential for understanding limitations of the robustness of the overall automotive system. It should be emphasized that this approach is meant to provide rapid high level insight of what can be expected on the system level and it is not envisioned to be a high accuracy simulation tool. Also, data integrity and error recovery procedures are beyond the scope of this paper.