Computational forensic techniques for intellectual property protection

Computational forensic engineering (CFE) aims to identify the entity that created a particular intellectual property (IP). Specifically, our goal is to identify the synthesis tool or compiler which was used to produce a specific design or program. Rather than relying on watermarking content or designs, the generic CFE methodology analyzes the statistics of certain features of a given IP and quantizes the likelihood that a well known source has created it. In this paper, we describe the generic methodology of CFE and present a set of techniques that, given a set of compilation tools, identify the one used to generate a particular hardware/software design. The generic CFE approach has four phases: 1) feature and statistics data collection; 2) feature extraction; 3) entity clustering; and 4) validation. In addition to IP protection, the developed CFE paradigm can have other potential applications: optimization algorithm selection and tuning, benchmark selection, and source-verification for mobile code.