Symbolic Solver for Live Variable Analysis of High Level Design Languages

This paper presents an efficient binary decision diagram solver for live variable analysis of high level design languages like SpecC. A novel representation of the program as a global structure using efficient data structure called Binary Decision Diagrams is presented. Based on the global structure of the program, a relational framework for live variable analysis is proposed and the solver algorithm is developed based on the relational framework. The algorithm obtains the fixed point of live variable analysis in iterations significantly less than the traditional chaotic iteration approach. The global representation of the program using Binary Decision Diagrams and the ease of doing set theoretic operations in the domain of Binary Decision Diagrams helps the solver algorithm to determine the fixed points of the analysis efficiently. Our experimental evaluations using the solver demonstrate a marked decrease in iterations over traditional approach for fixed point calculations. General Terms Algorithms