Multi-Attacker Protocol Validation

  • Authors:
  • Wihem Arsac;Giampaolo Bella;Xavier Chantry;Luca Compagna

  • Affiliations:
  • SAP Research, Sophia Antipolis, France;Dipartimento di Matematica e Informatica, Università di Catania, Catania, Italy and Software Technology Research Laboratory, De Montfort University, Leicester, UK;SAP Research, Sophia Antipolis, France;SAP Research, Sophia Antipolis, France

  • Venue:
  • Journal of Automated Reasoning
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

Security protocols have been analysed focusing on a variety of properties to withstand the Dolev-Yao attacker. The Multi-Attacker treat model allows each protocol participant to behave maliciously intercepting and forging messages. Each principal may then behave as a Dolev-Yao attacker while neither colluding nor sharing knowledge with anyone else. This feature rules out the applicability of existing equivalence results in the Dolev-Yao model. The analysis of security protocols under the Multi-Attacker threat model brings forward yet more insights, such as retaliation attacks and anticipation attacks, which formalise currently realistic scenarios of principals competing each other for personal profit. They are variously demonstrated on a classical protocol, Needham-Schroeder's, and on a modern deployed protocol, Google's SAML-based single sign-on protocol. The general threat model for security protocols based on set-rewriting that was adopted in AVISPA (Armando et al. 2005) is extended to formalise the Multi-Attacker. The state-of-the-art model checker SATMC (Armando and Compagna, Int J Inf Secur 6(1):3---32, 2007) is then used to automatically validate the protocols under the new threats, so that retaliation and anticipation attacks can automatically be found. The tool support scales up to the Multi-Attacker threat model at a reasonable price both in terms of human interaction effort and of computational time.