Management of replicated IEC 61499 applications

Abstract

The EEC-61499 standard defines a framework in which distributed control applications may be developed, and how these are interpreted and executed. With the distributed nature of the control applications, many new issues also need to be taken into account. Most of these arise due to the new error model and failure modes of the distributed hardware on which the distributed application is being executed. On the other hand, the distributed nature of the hardware also allows new capabilities to be explored, such as improved reliability through masking of faults. Fault tolerance is usually achieved by the use of replication. Replication at the software level allows for much smaller granularity, but all replicas must be kept synchronised in order to produce the same output data and events. To achieve replica determinism all replicas must process the same set of inputs in the same order, and the outputs of the replicas must somehow be consolidated. The EEC 61499 framework however does not explicitly handle the replication of software components. In this paper we propose a replication model for IIEC-61499 applications, and discuss which mechanisms and protocols may be used for their support. In short, replica determinism is achieved through the use of distributed synchronised clocks, atomic multicast protocols, as well as timed messages.