Improved fault tolerant broadcasts in CAN

It is generally considered that the controller area network (CAN) guarantees atomic broadcast properties through its extensive error detection and signalling mechanisms. However, it is known that these mechanisms may fail, and messages can be delivered in duplicate by some receivers or delivered only by a subset of the receivers. This misbehaviour may be disastrous if the CAN network is used to support replicated applications. In order to prevent such inconsistencies, a set of atomic broadcast protocols is proposed, taking advantage of CAN synchronous properties to minimise its run-time overhead. The paper presents such set of protocols, and demonstrates how they can be used for the development of distributed real-time applications.