Improving flexibility and fault-management in CAN-based “Shared-Clock” architectures

The Controller Area Network (CAN) protocol is widely employed in distributed control systems in a number of sectors (including automotive, industrial, medical and satellite systems). In many such sectors, predictable behavior is a key consideration: such a characteristic is usually associated with time-triggered (TT) system architectures rather than with event-triggered solutions which are more common in CAN systems. Previous studies have illustrated how TT “Shared-Clock” (S-C) algorithms can be used in conjunction with commercial-off-the-shelf (COTS) microcontrollers for creating reliable distributed control systems at low cost. In such studies, it has generally been assumed that S-C designs will be implemented using a bus topology on a CAN network. Such designs met cost requirements but certain flexibility/reliability issues predominantly presented by their bus topology have made their use controversial in safety–critical applications. In this paper, we demonstrate that the use of a Shared-Clock protocol with a novel CAN-based star topology synthesized from COTS components can improve the flexibility and reliability of CAN-based S-C designs without significantly increasing costs.

► We identify scheduling flexibility issues in time-triggered Shared-Clock designs.
► We also identify reliability issues in such designs due to their CANbus topology.
► We propose a novel CANstar topology and a novel Shared-Clock algorithm.
► Improvements in flexibility and reliability are shown through case studies.