Detecting malicious modifications of data in third-party intellectual property cores

Globalization of the system-on-chip (SoC) design flow has created opportunities for rogue elements in the intellectual property (IP) vendor companies to insert malicious circuits (a.k.a. hardware Trojans) into their IPs. We propose to formally verify third party IPs (3PIPs) for unauthorized corruption of critical data such as secret key. Our approach develops properties to identify corruption of critical registers. Furthermore, we describe two attacks where computations can be performed on corrupted data without corrupting the critical register. We develop additional properties to detect such attacks. We validate our technique using Trojans in 8051 and RISC processors and AES designs from Trust-Hub.