Static sessional dataflow

Abstract

Sessional dataflow provides a compositional semantics for dataflow computations that can be scheduled at compile-time. The interesting issues arise in enforcing static flow requirements in the composition of actors, ensuring that input and output rates of actors on related channels match, and that cycles in the composition of actors do not introduce deadlock. The former is ensured by flowstates, a form of behavior type that constrains the firing behavior of dataflow actors. The latter is ensured by causalities, a form of constraints that record dependencies in the firing behavior. This article considers an example variant of the sessional dataflow approach for dataflow applications, expressing known ideas from signal processing in a compositional fashion. © 2012 Springer-Verlag Berlin Heidelberg.