Section: New Results

Unconditional Soundness (Objective 2)

Participants : Hubert Comon-Lundh, Guillaume Scerri.

Hubert Comon-Lundh, Véronique Cortier and Guillaume Scerri had shown in a 2012 CCS paper how one could drop one of the assumptions of computational soundness results. However, the proofs remain very complicated and there are still assumptions such as the absence of key cycles, or no dynamic corruption... that are still necessary for all these results.

Gergei Bana and Hubert Comon-Lundh investigated a completely different approach to formal security proofs in a 2012 POST paper, which does not make any such assumptions. The idea can be stated in a nutshell: whereas all existing formal models state the attacker's abilities, they propose to formally state what the attacker cannot do.

This makes a big difference, since the soundness need only to be proved formula by formula and only the very necessary assumptions are used for such formulas (for instance, no absence of key cycles is needed). This does not need to be proved again when a primitive is added.

Once the general setting is fixed, the question was how practical is the method. We studied the complexity of the consistency proofs in this setting and showed that we can complete such proofs in Polynomial Time for a wide class of axioms in

• H. Comon-Lundh, V. Cortier and G. Scerri. Tractable inference systems: an extension with a deducibility predicate. In CADE'13, LNAI 7898, pages 91-108. Springer, 2013

The development of a prototype implementation is under development. We expect to complete experiments on a number of protocols.