Brian Vogel, UC Berkeley
vogel@eecs.berkeley.edu
Heterochronous dataflow (HDF) is an extension to synchronous dataflow (SDF) that provides increased expressiveness while still retaining many of the desirable properties of SDF. In particular, bounded memory usage and deadlock are decidable in HDF.
HDF fits elegantly into the *charts formalism for composing hierarchical finite state machines (FSMs) with various concurrency models. HDF combined with hierarchical FSMs is a natural extension of the *charts semantics for composing SDF with FSMs.
This poster will present the semantics of the HDF model of computation and the *charts semantics for combining HDF with hierarchical FSMs. A Ptolemy II implementation of HDF with hierarchical FSMs is presented.