Interlocking obfuscation for anti-tamper hardware

  • Authors:
  • Avinash R. Desai;Michael S. Hsiao;Chao Wang;Leyla Nazhandali;Simin Hall

  • Affiliations:
  • Virginia Tech, Blacksburg, Virginia;Virginia Tech, Blacksburg, Virginia;Virginia Tech, Blacksburg, Virginia;Virginia Tech, Blacksburg, Virginia;Virginia Tech, Blacksburg, Virginia

  • Venue:
  • Proceedings of the Eighth Annual Cyber Security and Information Intelligence Research Workshop
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

Tampering and Reverse Engineering of a chip to extract the hardware Intellectual Property (IP) core or to inject malicious alterations is a major concern. Digital systems susceptible to tampering are of immense concern to defense organizations. First, offshore chip manufacturing allows the design secrets of the IP cores to be transparent to the foundry and other entities along the production chain. Second, small malicious modifications to the design may not be detectable after fabrication without anti-tamper mechanisms. Some techniques have been developed in the past to improve the defense against such attacks but they tend to fall prey to the increasing power of the attacker. We present a new way to protect against tampering by a clever obfuscation of the design, which can be unlocked with a specific, dynamic path traversal. Hence, the functional mode of the controller is hidden with the help of obfuscated states, and the functional mode is made operational only on the formation of a specific interlocked Code-Word during state transition. No comparator is needed as the obfuscation is embedded within the transition function of the state machine itself. A side benefit is that any small alteration will be magnified via the obfuscated design. In other words, an alteration to the design will manifest itself as a large difference in the circuit's functionality. Experimental results on an Advanced Encryption Standard (AES) circuit from the open-source IP-cores suite suggest that the proposed method provides better active defense mechanisms against attacks with nominal (7.8%) area overhead.