Modelling and Design of Distributed Real-Time Control Applications

The Controller Area Network (CAN) constitutes a good low level base for distributed real–time control systems. The priority mechanism allows for tailoring by choice of scheduling policy. It is however unfortunate that the error handling policy of CAN is fixed, this limits the solution space for the designer. Clock synchronization should be added to the basic CAN system in order to obtain a global system clock. Designing a distributed system around CAN requires the identification of the appropriate level of system decentralization and a suitable system organization. These design aspects are presented and the use of CAN as a basis for distributed control systems is discussed.