Design of concurrent software based on problem concurrency

Abstract

Two main approaches to the design of concurrent real-time software exist. One approach looks at how input data is transformed into output. The transformations essentially become the tasks. The other approach models the concurrency of the software on concurrency inherent in the problem. One example of this approach, entity-life modeling, starts by partitioning the events in the problem into concurrent threads, where each thread consists of a sequence of events with a minimum interval between each. Such a thread model is minimal if there is a time when all the threads occur simultaneously. A thread can often be seen as the “life” of some entity in the problem. Essentially, each thread gives rise to a task in the software. As an example, three different thread models of a flexible manufacturing system (FMS) are given. The FMS is essentially about resource contention. This can be made explicit by choosing a thread model where the entities contend for simultaneous, exclusive access to multiple resources.