Dataflow-like languages for designing dependable real-time control systems

We discuss the use of dataflow-like models for the "in-the-large" design of real-time applications. These models are based on modules which communicate via asynchronous message-passing; each module starts execution upon receipt of a pre-defined set of messages, processes their contents and terminates with the output of messages towards other modules. This model somewhat restricts the freedom of a designer, but is well suited for the description of typical, cyclic control applications, and the very restrictions that it imposes imply more predictable behaviour and hence better verifiability for the design. The main advantage of this approach is that timing considerations can be included in the early stages of design. The data flow graph describes not only the structure of possible communications between modules (as common with in-the-large design notations), but information about the execution sequence. Such a design can be analyzed using well-known tools (like Petri nets), and its timing behaviour can be checked We outline the proposed approach and discuss the features of the model that affect the trade-off between predictability of behaviour and expressive power.