Abstract: As massively distributed storage becomes the norm in cloud networks, they contend with new vulnerabilities imputed by the presence of data in different, possibly untrusted nodes. In this talk, we consider two such types of vulnerabilities. The first one is the risk posed to data stored at nodes that are untrusted. We show that coding alone can be substituted to encryption, with coded portions of data in trusted nodes acting as keys for coded data in untrusted ones. In general, we may interpret keys as representing the size of the list over which an adversary would need to generate guesses in order to recover the plaintext, leading to a natural connection between list decoding and secrecy. Under such a model, we show that algebraic block maximum distance separable (MDS) codes can be constructed so that lists satisfy certain secrecy criteria, which we define to generalize common perfect secrecy and weak secrecy notions. The second type of vulnerability concerns the risk of passwords’ being guessed over some nodes storing data, as illustrated by recent cloud attacks. In this domain, the use of guesswork as metric shows that the dominant effect on vulnerability is not necessarily from a single node, but that it varies in time according to the number of guesses issued. We also introduce the notion of inscrutability, as the growth rate of the average number of probes that an attacker has to make, one at a time, using his best strategy, until he can correctly guess one or more secret strings from multiple randomly chosen strings.
About the Speaker: Muriel Médard is the Cecil E. Green Professor of the Electrical Engineering and Computer Science Department at MIT. Professor Médard received B.S. degrees in EECS and in Mathematics in 1989, a B.S. degree in Humanities in 1990, a M.S. degree in EE 1991, and a Sc D. degree in EE in 1995, all from MIT. Her research interests are in the areas of network coding and reliable communications, particularly for optical and wireless networks. She was awarded the IEEE Leon K. Kirchmayer Prize (2002), the IEEE Communication Society and Information Theory Society Joint Paper Award (2009), and the IEEE William R. Bennett Prize (2009). She received the 2004 MIT Harold E. Edgerton Faculty Achievement Award. She was named a Gilbreth Lecturer by the NAE in 2007. She is a Fellow of IEEE, and past President of the IEEE Information Theory Society.