Intellectual property metering(6)

Abstract. We have developed the first hardware and software (intellectual property) metering scheme that enables reliable low overhead proofs for the number of manufactured parts and copied programs. The key idea is to make each design slightly different d

Inc., an Oregon start-up company, has proposed an approach for integrated circuit identifi-cation from random threshold mismatches in an array of addressable MOSFETs. The tech-nique leverages on process discrepancies unavoidably formed during fabrication. This analog technique can be used in tracking semiconductor dies, authentication and intellec-tual property (IP) tagging. In a recent report of this method’s measured performance[23], for a 0.35um poly CMOS, for generating 112 ID bits, 132 blocks area used, each with the area of 252x93um. The IDs proposed by SiidTech are not deterministic and these IDs can not be deterministically compacted. Also, due to the birthday paradox, there is still a small probability that two IDs generated randomly have the same value. Component applica-tions enables the user to trace a particular die on a wafer and store this information for future usages. There are several advantages of our scheme over the Siid scheme. We have been able to obtain more than 1.3E12 distinct solutions even in our smallest test cases which have only 15 registers (Our number of solutions will go exponentially high by using a few more registers). Furthermore, our IDs are deterministic and therefore they can be used to contain a defined signature to be used in many cryptographic schemes.

From a more global point of view, intellectual property protection schemes, such as hardware metering, are traditionally treated within computational security. Cryptography is the study of techniques related to aspects of information security such as confidentiality, data integrity and entity authentication [34]. Computational security has even more broad scope and includes privacy protection, password protection, denial of service, and content usage measuring.

Modern cryptography started with introduction of one-way trapdoor function-based public-key cryptographical communication protocols by Diffie and Hellman [11]. In 1978, Rivest, Shamir and Adleman discovered the first practical sound public key encryp-tion and signature scheme based on exponential computational difficulty of factoring numbers which are the products of two large prime numbers. A number of excellent cryp-tographical textbooks are available [35, 34, 25].

Intellectual property protection of audio and video artifacts and hardware and soft-ware components and systems recently attracted a great deal of attention. For example, the Virtual Socket Interface Alliance has been making progress on standardizing SoC design and IP protection [38].

Multimedia watermarking schemes are utilizing minute alternations of audio or video so that the signature is embedded while human perceived quality of artifact is fully pre-served. For survey, see [18], Protecting design (hardware and software) IP is a broad and complex issue.

One method to enable design IP protection is based on the constraint manipulation. The basic idea is to impose additional author-specific constraints on the original IP speci-fication during its creation and/or synthesis. The copyright detector checks whether a syn-thesized IP block satisfies the author-specific constraints. The strength of the proof of authorship is proportional to the likelihood that an arbitrary synthesis tool incidentally sat-isfies all the added constraints [19, 32]. Similarly, to protect legal users of the IP, finger-prints are added to the IP as extra constraints [5]. Finally, copy detection techniques for VLSI CAD applications have been developed to find and prove improper use of the design IP [7,20]. These techniques are effective for authentication. However, since they make each design unique, it becomes ill-suited for mass-production and cannot be applied, at least not directly, and without significant modification, to hardware metering.