A Dataflow Framework for DSP Algorithm Refinement

Abstract

Current video compression algorithms are increasingly complicated and difficult to analyze and profile. Design tools and system level languages often prove to be inefficient and incapable of providing complexity analysis as a first step directed toward at the implementation of video compression algorithms. This paper proposes framework that will help to develop a methodology that facilitates the derivation of analytical dataflow models. The framework proposes dataflow models for quantifying the underlying algorithm's memory complexity, related timing considerations, and verification of the correctness of the video compression algorithm.