- doi:
- 10.1109/MC.2009.145
- issn:
- 0018-9162
Two proposed star architectures for systems based on the Controller Area Network (CAN) protocol aim to improve CAN's error containment and fault tolerance for general distributed embedded systems and safety-critical applications.
Similar to other communication technologies, CAN can also benefit from moving from bus to star topologies. CANcentrate and ReCAN- centrate can provide two levels of dependable communication, the former aimed at increasing robustness for general distributed embedded systems, and the latter aimed at safety-critical applications. Both architectures are fully compatible with CAN and offer its beneficial properties such as electrical robustness, ease of deployment, bounded latency with prioritized medium access control, and low cost. This last aspect is particularly relevant because with the proposed architectures it is pos- sible to reach error-containment and reliability levels in CAN that previously were possible only with more expen- sive technologies such as TTP/C and FlexRay.
Moreover, (Re)CANcentrate retains the advantage of being fully application-independent. This quality makes it flexible to deploy and use, thus it is well suited to sup- port the trend in embedded systems toward adaptability and reconfigurability. However, considering application- dependent information for detecting and isolating more complex error patterns, such as masquerading and bab- bling idiots, enables greater system robustness but at the expense of more complex network configuration. Maintaining a high level of flexibility in this approach is a challenge that we’ll explore next in our quest for robust and adaptable distributed embedded systems that exhibit graceful degradation and that pave the way for cheaper fault-tolerant mechanisms.