A Physical Zero-Knowledge Proof and Unclonable Sensors for Nuclear Warhead Verification

Abstract

Future nuclear arms-control agreements may call for reductions of the total number of nuclear weapons and warheads in the world arsenals. Such agreements would require new trusted verification mechanisms to confirm that items presented to inspectors are nuclear warheads and not spoofs. Proliferation and national security concerns require, however, that inspectors gain no warhead design information through this process. To address this paradox, considerable efforts have been directed towards the development of "information barriers." These barriers consist of automated measurement systems that process sensitive information but only display the results of internal analysis in a binary valid/invalid manner. These systems are, by their nature, at risk of electronic tampering and snooping – and their trusted implementation has so far proved extremely difficult to realize.

This thesis takes radically different directions to address this challenge. It demonstrates new approaches to information protection and trusted instruments for nuclear warhead verification that are based on the cryptographic concepts of zero-knowledge proofs and physically unclonable functions.

To the author's knowledge, this thesis provides the first demonstration of a zero-knowledge physical measurement technique. Using fast neutron differential radiography and superheated emulsion detectors, such a technique can show two objects have identical geometry and opacity to 14-MeV neutrons without revealing what these properties are. This zero-knowledge feature no longer holds when the objects compared are significantly different. Such a technique could form the basis of a template-matching verification system that could confirm the authenticity of nuclear weapons without sharing any secret design information.

The thesis then introduces and demonstrates key elements of an optical physical unclonable function sensitive to neutrons and based on superheated emulsions. Such sensors are unique objects that cannot be cloned or simulated. The data they produce are a function of both their internal disordered structure and the physical quantity they measure. Due to their sensitivity against any structural variation, including through neutron irradiation, it is possible to show that they – or the data they have recorded -- have not been tampered with. Such sensors could be used by adversarial parties in sensitive facilities without the risk of being compromised.