Secure Information Flow by Self-Composition

  • Authors:
  • Gilles Barthe;Pedro R. D'Argenio;Tamara Rezk

  • Affiliations:
  • INRIA Sophia-Antipolis, France;Université de Provence, France;INRIA Sophia-Antipolis, France

  • Venue:
  • CSFW '04 Proceedings of the 17th IEEE workshop on Computer Security Foundations
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

Non-interference is a high-level security propertythat guarantees the absence of illicit information leakagesthrough executing programs.More precisely,non-interference for a program assumes a separation betweensecret inputs and public inputs on the one hand,and secret outputs and public outputs on the other hand,and requires that the value of public outputs does not dependon the value of secret inputs.A common meansto enforce non-interference is to use an informationflow type system.However, such type systemsare inherently imprecise, and reject many secure programs,even for simple programming languages.Thepurpose of this paper is to investigate logical formulationsof non-interference that allow a more precise analysisof programs.It appears that such formulations areoften sound an complete, and also amenable to interactiveor automated verification techniques, such astheorem-proving or model-checking.We illustrate the applicability of our method in severalscenarii, including a simple imperative language, anon-deterministic language, and finally a language withshared mutable data structures.