Modelling a distributed real-time system for feedback control in a nuclear fusion experiment

The paper presents a real time distributed application which is used for feedback control in a nuclear fusion experiment. The system employs a number of Texas TMS320C40 digital signal processors mounted on VME crates which communicate by means of dedicated communication ports and reflective memories. Owing to the distributed architecture of the system, a formal analysis was necessary to fully understand its behaviour and derive performance measures. The system has been modelled using interval time Petri nets, an extension of Petri nets which associates time intervals with transitions. Based on this model, an automated analysis tool has been developed and has been used to derive performance parameters, such as system throughput and delays