Testbench Qualification for SystemC‐AMS Timed Data Flow Models
Muhammad Hassan1,2,a, Daniel Große1,2,b, Hoang M. Le2, Thilo Vörtler3,d, Karsten Einwich3,e and Rolf Drechsler1,2,c
1Cyber‐Physical Systems, DFKI GmbH, 28359 Bremen, Germany
amuhammad.hassan@dfki.de
bgrosse@informatik.uni-bremen.de
cdrechsle@informatik.uni-bremen.de
2Institute of Computer Science, University of Bremen, 28359 Bremen, Germany
hle@informatik.uni-bremen.de
3COSEDA Technologies GmbH, Dresden, Germany
cthilo.voertler@coseda-tech.com
ckarsten.einwichg@coseda-tech.com
ABSTRACT
Analog‐Mixed Signal (AMS) circuits have become increasingly important for today’s SoCs. The Timed Data Flow (TDF) model of computation available in SystemC‐AMS offers here a good tradeoff between accuracy and simulation‐speed at the system‐level. One of the main challenges in system‐level verification is the quality of the testbench. In this paper, we present a testbench qualification approach for SystemC-AMS TDF models. Our contribution is twofold: First, we propose specific mutation models for the class of filters implemented as TDF models. This requires to analyze the Laplace transfer function of the filter design. Second, we present the mutationbased qualification approach based on the proposed specific mutations as well as standard behavioral mutations. This allows to find serious quality issues in the testbench. Our experimental results for a real‐world AMS system demonstrate the applicability and efficacy of our approach.
