Publicly detectable watermarking for intellectual property authentication in VLSI design

Abstract

Highlighted with the newly released intellectual property (IP) protection white paper by VSI Alliance, the protection of virtual components or IPs in very large scale integration (VLSI) design has received a great deal of attention recently. Digital signature/watermark is one of the most promising solutions among the known protection mechanisms. It provides desirable proof of authorship without rendering the IP useless. However, it makes the watermark detection, which is as important as watermarking, an NP-hard problem. In fact, the tradeoff between hard-to-attack and easy-to-detect and the lack of efficient detection schemes are the major obstacles for digital signatures to thrive. In this paper, the authors propose a new watermarking method which allows the watermark to be publicly detected without losing its strength and security. The basic idea is to create a cryptographically strong pseudo-random watermark, embed it into the original problem as a special (which the authors call mutual exclusive) constraint, and make it public. The authors combine data-integrity technique and the unique characteristics in the design of VLSI IPs such that adversaries will not gain any advantage from the public watermarking for forgery. This new technique is compatible with the existing constraint-based watermarking/fingerprinting techniques. The resulting public-private watermark maintains the strength of a watermark and provides easy detectability with little design overhead. The authors build the mathematical framework for this approach based on the concept of mutual exclusive constraints. They use popular VLSI CAD problems, namely technology mapping, partitioning, graph coloring, FPGA design, and Boolean satisfiability, to demonstrate the public watermark's easy detectability, high credibility, low design overhead, and robustness.