We study the security of password protocols against off-line dictionary attacks. In addition to the standard adversary abilities, we also consider further cryptographic advantages given to the adversary when considering the password protocol being instantiated with particular encryption schemes. We work with the applied pi calculus of Abadi and Fournet, in which the (new) adversary abilities are modelled as equations between terms. As case studies, we analyse the Encrypted Password Transmission (EPT) protocol of Halevi and Krawczyk, and the wellknown Encrypted Key Exchange (EKE) of Bellovin and Merritt. Finally, we propose a modification to EKE that prevents a particular attack that arises when ciphertexts are distinguishable from random noise.
|Name||CTIT Technical Reports|
|Publisher||University of Twente, CTIT|