Distributed Real-Time Dataflow: An Execution Paradigm for Image Processing and Anti-Submarine Warfare Applications

The central thesis of this project is that real-time scheduling theory can be combined with dataflow methodologies to bound latency and memory utilization of distributed signal processing applications, such as those found in anti-submarine warfare and image processing. To this end, we propose a new real-time dataflow paradigm that is based on the Navy's Processing Graph Method (PGM) [5], which is similar to the dataflow methodology employed by Ptolemy [1, 4] of the University of California at Berkeley and the Rapid Prototyping of Application Specific Signal Processors (RASSP) project, funded by the Advanced Research Projects Agency (ARPA). While our dataflow methodology itself is not novel, our application of real-time scheduling theory to the model is