Detecting algorithms using dynamic analysis

Abstract

Detecting a given algorithm in a program without access to its source code can be valuable in many tasks ranging from intellectual property management to verifying the program's security properties. Unfortunately, approaches based on decompiling or reverse-engineering the program suffer from prohibitive costlyness as well as theoretical limitations. Instead we base our work on examining the program's internal dynamic behavior and trying to find in it tell-tale signs of the given algorithm using various pattern matching and statistical analysis techniques.