Typing messages for free in security protocols

Typing messages for free in security protocols. Rémy Chrétien, Véronique Cortier, Antoine Dallon, and Stéphanie Delaune. ACM Transactions on Computational Logic, 21(1), October 2019.

Download

[PDF] [HTML] 

Abstract

Security properties of cryptographic protocols are typically expressed as reachability or equivalence properties. Secrecy and authentication are examples of reachability properties while privacy properties such as untraceability, vote secrecy, or anonymity are generally expressed as behavioral equivalence in a process algebra that models security protocols.
Our main contribution is to reduce the search space for attacks for reachability as well as equiva- lence properties. Specifically, we show that if there is an attack then there is one that is well-typed. Our result holds for a large class of typing systems, a family of equational theories that encom- passes all standard primitives, and protocols without else branches. For many standard protocols, we deduce that it is sufficient to look for attacks that follow the format of the messages expected in an honest execution, therefore considerably reducing the search space.

BibTeX

@Article{TOCL2019,
  author = 	 {R\'emy Chr\'etien and V\'eronique Cortier and Antoine Dallon and St\'ephanie Delaune},
  title = 	 {Typing messages for free in security protocols},
  journal = 	 {ACM Transactions on Computational Logic},
  year = 	 {2019},
  volume = 	 {21},
  number = 	 {1},
  month = 	 {October},
  doi={10.1145/3343507},
  abstract = {Security properties of cryptographic protocols are typically expressed as reachability or equivalence properties. Secrecy and authentication are examples of reachability properties while privacy properties such as untraceability, vote secrecy, or anonymity are generally expressed as behavioral equivalence in a process algebra that models security protocols.
  \par
Our main contribution is to reduce the search space for attacks for reachability as well as equiva- lence properties. Specifically, we show that if there is an attack then there is one that is well-typed. Our result holds for a large class of typing systems, a family of equational theories that encom- passes all standard primitives, and protocols without else branches. For many standard protocols, we deduce that it is sufficient to look for attacks that follow the format of the messages expected in an honest execution, therefore considerably reducing the search space.},
}