Data driven nets: a maximally concurrent, procedural, parallel process representation for distributed control systems

A procedural parallel process representation, known as data-driven nets is described. The sequencing mechanism of the data-driven representation is based on the principle of data dependency. Operations are driven into action by the arrival of the required working set of input operands. Execution of DDN processes is side-effect free, and influence in the net representation is transparaent. Data-driven nets have several advantages over many of the existing parallel process representations. These nets are capable of representing parallelism below the statement level, and in addition may be arbitrarily pipelined. Data-driven nets are simpler than other data-flow schema in that no distinction need be made between control and data. A process model for data-driven nets is given and a number of properties of the model are discussed. The operating rules for data-driven nets are completely asynchronous and the nets therefore serve as an excellent low-level process notation for distributed systems.