PIs: Matthew Blaze and Sampath Kannan University of Pennsylvania
The Isotropic Cryptography project introduces and lays the theoretical and engineering foundations for information-theoretically secure communication over open channels that meet certain requirements. More specifically, the project identifies and is concerned with {em isotropic channels,} in which eavesdroppers monitoring the channel are unable to identify the sources and destinations of messages with certainty. Such channels appear to have very interesting and promising security properties. Even a partially isotropic channel (in which the adversary can identify a message source but only with some bounded certainty) can provide information-theoretic confidentiality.
The project's experimental work focuses on designing and analyzing practical protocols that exploit the isotropic properties of conventional communication channels (such as local Ethernet, near-field radio and mobile wireless networking). Can (partially or fully) isotropic channels be realized in practice over existing network technologies? The project also lays a theoretical foundation for isotropic cryptography, studying what assumptions are necessary to implement protocols for various well-known and new cryptographic primitives and what the most efficient protocols are for various primitives. Finally, the project examines the relationship between traditional anonymity networks and Mix networks on the one hand and isotropic channels on the other. Can isotropy be used to provide anonymity and can anonymous networks simulate isotropic channels? Can isotropic approaches be made practical and efficient?
