REU Supplement for EAGER: PUF-Locked Circuit Obfuscation for Counterfeit and Piracy Prevention.

The business model of semiconductor industry has changed significantly in last decade. With increasing complexity and cost of modern ICs, a design house has to seek the aid from various external agencies, such as EDA companies, IP vendors, library providers, and fabrication foundries. The active participation of external entities in the design and manufacturing flow has produced numerous hardware security issues. Among all the hardware security problems, piracy is likely to be the most ubiquitous and expensive one. Most leading-edge design houses have outsourced their fabrication to the offshore foundries for the sake of lower labor and manufacturing cost. However, many offshore foundries are hard to be trusted since they may be in a country without consummate enforcement law for IP protection. The economic impacts and security hazards of hardware piracy is not apt to be neglected compared to software, but is even more severe. The loss due to global hardware piracy has now reached the level of billions per month, with a major share in almost all electronic devices. It was reported by the Alliance for Gray Market and Counterfeit Abatement that about 10% of the start-of-the-art technology products available on market are counterfeits. In this project, we are going to develop a theory and a framework of circuit obfuscation for hardware counterfeit and piracy prevention. Obfuscation is a technique to make a design obscure therefore hard to understand. We will leverage the recent theoretical studies on program obfuscation, and build our work on the definition of the best-possible obfuscation. With efficient obfuscators to structurally transform a circuit to an obfuscated one, we can embed any security enhancement circuitry such as a lock or a watermark in the circuit without being discovered. Not understanding a design, an potential attacker could not modify the design to remove the security circuitry or to insert any Trojan. Besides the framework and relevant tools for circuit obfuscation, our project will also work on a large number of real world applications to validate the approach. Intellectual Merit: This project will build a solid theoretical foundation for circuit obfuscation and its application to counterfeit and piracy prevention. Different from all existing approaches that need to operate on the behavioral state transition graphs, our approach only uses structural operations such as retiming, resynthesis, sweep, and conditional stuttering, which are exponentially efficient. We will first establish the fact that the best-possible obfuscation can be achieved by a sequence of structural operations. Based on it, a framework to obfuscate sequential circuits by structural transformation will be developed. In order to prevent counterfeit ICs, a secret key will be embedded into the design. The design will have two different work modes: normal mode and slow mode, depending on whether the initial state matches the key. In slow mode, the IC still functions but becomes much slower compared to normal mode. Since the normal mode contains an extremely small portion of all power-up states, any pirate who starts at a random state will most often end up using the degraded IC, without even suspecting the existence of the key. The hidden key can also be used as a proof of design authorship like watermarking, imposing extra penalty on piracy. Broader Impacts: Hardware is the foundation of modern information technology. One big obstacle for increasing the productivity of hardware design thus decreasing the cost and barrier for adoption o