Section: New Results

Monitoring attacker knowledge with information flow analysis

Participants : Thomas Jensen, Frédéric Besson.

Motivated by the problem of stateless web tracking (fingerprinting) we have investigated a novel approach to hybrid information flow monitoring by tracking the knowledge that an attacker can learn about secrets during a program execution. We have proposed a general framework for combining static and dynamic information flow analysis, based on a precise representation of attacker knowledge. This hybrid analysis computes a precise description of what an attacker learns about the initial configuration (and in particular the secret part of it) by observing a specific output. An interesting feature of this knowledge-based information flow analysis is that it can be used to improve other information flow control mechanisms, such as no-sensitive upgrade. The whole framework is accompanied by a formalisation of the theory in the Coq proof assistant [18].