Software Theft Detection Through Program Identification

Abstract

Prior to the general availability of high speed Internet, the spread of piratedsoftware required the transfer of a physical copy like a disk, which limited therate at which illegal software could be distributed. The low transfer raterestricted software piracy to levels which producers found acceptable becausethe associated losses could be absorbed. Large scale cases of piracy were rareand when they did occur the legal system provided suitable retribution. However,recent advances in computer technology have made the need for a physical copyobsolete. Piracy is now a widespread, decentralized problem in which millions ofindividuals take part. Without technical means of identifying pirated software,the protection afforded by the legal system is no longer easy to enforce or costeffective.The research in this dissertation addresses the threat of software piracythrough the exploration of two techniques: software watermarking and softwarebirthmarking. Neither of these techniques can be used to prevent software theftentirely. Instead, they are used to detect occurrences of theft after the fact.One of the limiting factors of the protection provided by the legal system isthat it cannot be used to identify an incidence of piracy. Softwarewatermarking and birthmarking fill this gap, thus providing complimentaryprotection to the established legal protection. In this research, we analyze thestate of the art in both software watermarking and birthmarking and we propose anovel scheme in each of the areas which make significant improvements overexisting techniques.