Transaction Parameterized Dataflow: A Model for Context-Dependent Streaming Applications

Abstract

Abstract-Static dataflow programming models are well suited to the development of embedded many-core systems. However, complex signal and media processing applications often display dynamic behavior that do not fit the classical static restrictions. We propose Transaction Parameterized Dataflow (TPDF), a new model of computation combining integer parameters-to express dynamic rates-and a new type of control actor-to allow topology changes and time constraints enforcement. We present static analyses for liveness and bounded memory usage. We also introduce a static scheduling heuristic to map TPDF to massively parallel embedded platforms. We validate the model and associated methods using a cognitive radio application, demonstrating significant buffer size and performance improvements compared to state of the art models including Cyclo-Static Dataflow (CSDF).