CCoE Webinar Series: Stronger Security for Password Authentication

Originally recorded August 28th, 2017 Passwords are an infamous bottleneck of information security: The users choose them badly and then forget them, and the servers store (at best!) a table of password hashes which, in the all-too-common event that the server is hacked, allows the attacker to recover a large fraction of the passwords using the so-called Offline Dictionary Attack. At the same time, we seem to be stuck with passwords because they form the most user-friendly authentication mechanism we know. Our work in the CICI-sponsored project looks at the security vulnerabilities of current password authentication protocols, including Two-Factor authentication protocols, where the user's password is amended by the presence of an Auxiliary Authentication Device, e.g. a cell-phone capable of displaying a short one-time PIN which the user copies onto her terminal in order to authenticate to the server. We show that with modest changes to the authentication infrastructure, involving either the user's client, or the authentication server, or the Auxiliary Device software, we can make password authentication protocols which are as practical as currently used schemes but have much stroger security properties. Most importantly, the schemes we show eliminate the security vulnerability posed by the server storing password hashes, thus eliminating the possibility of the Offline Dictionary Attack in case of server compromise. In other properties, our schemes offer resistance to so-called phishing attacks and, more generally, failures in the Public Key Infrastructure, where the user misidentifies the public key of the authentication server and, which in current password authentication schemes leads to revealing the user's password to the adversary. In this presentation we will present an overview of our work on strengthening password and two-factor schemes, published in NDSS'14, Asiacrypt'14, EuroSP'16, AsiaCCS'16, ACNS'17, ICDCS'17, as well as future directions. Prof. Jarecki's research is in the theory and applications of cryptography, with a focus on the design of efficient cryptographic protocols that meet the requirements of practical applications. His work includes contributions to fault-tolerant and distributed cryptography, privacy-protecting protocols, searchable encryption, and password authentication. Professor Jarecki received his PhD at MIT, was a postdoctoral researcher at Stanford, and has taught at University California at Irvine since 2003. Principal Investigators on the project: Stanislaw Jarecki, UC Irvine, and Nitesh Saxena, University of Alabama at Birmingham.