Using genetic evolutionary software application testing to verify a DSP SoC

Abstract

A digital signal processor (DSP) system-on-chip (SoC) can be designed using a variety of architectures and techniques. This often presents different verification challenges compared to conventional SoC or processor designs. Verification of such designs should take into account the goals and applications of the DSP, and how they are eventually used. This paper proposes an application based verification methodology and demonstrates this technique on a real-life DSP SoC design. Our technique employs a library of specially devised application functions as test building blocks, followed by a genetic evolutionary test generator to compose these application functions into effective test programs.