A randomized graph-based scheme for software watermarking

Abstract

The insertion of watermarks into proprietary objects is a well-known means of discouraging piracy. It works by embedding into the object some (often surreptitious) data meant to disclose the authorship/ownership of the object. Some promising graph-based watermarking schemes to protect the intellectual property of software have been suggested in the literature, and recent efforts have been endeavored to improve their resilience to attacks. Among the pursued attributes of software watermarking solutions is the one referred to as "diversity", which is the ability to encode the intended information in many distinct forms, making it harder for an attacker to find and remove it. We introduce a graph-based scheme which achieves a high level of diversity through randomization, while admitting an efficient, linear-time implementation nonetheless.