Expert System Simulation of Hardware

Abstract

Hardware lint tools based on static analysis techniques only check for structural, coding and consistency problems in RTL code. They have the limitation that they do not perform behavioral analysis and have a predefined rule-base that is not enhanced by users to capture human expertise. We present RTL-CLIPS, an RTL simulation environment based on CLIPS, an open-source expert system shell. In our approach, we do a semantically equivalent translation of RTL code into CLIPS facts. The simulator can be used by an heuristic rulebase to perform dynamic analysis of RTL code. Our method can better approximate design behavior by augmenting the pattern-matching capability of the CLIPS expert system with an environment for RTL simulation. We demonstrate our approach by analyzing ITC99 benchmark circuits. We discuss how our approach supports the reverse engineering of design behavior from RTL code.