Input decomposition by clusterization for symbolic execution

Abstract

The clusterization approach to a symbolic execution input decomposition was evaluated in the paper. The study uses the Klee engine to demonstrate the effects of input clustering for the console chess game written in the C language. The clusterization algorithm is adapted to the Klee by isolating a chosen cluster from the program code. The effect of different cluster sizes on the coverage, error detection, and execution time is evaluated. The obtained result shows benefits of the clusterization as a decomposition approach and may be used in further research to improve symbolic execution performance.