| | |
Summary: Noninterference and the Most Powerful
Probabilistic Adversary
Alessandro Aldini 1? and Alessandra Di Pierro 2
1 Istituto STI, University of Urbino \Carlo Bo", Italy
2 Dipartimento di Informatica, University of Pisa, Italy
Abstract. Probabilistic noninterference extends the classical possibilis-
tic notion introduced by Goguen and Meseguer in order to capture the
information leakage caused by adversaries that set up probabilistic covert
channels. In this setting we investigate how to evaluate the observational
power of an adversary to the purpose of establishing the maximal secu-
rity degree of a given system. We introduce three classes of probabilistic
adversaries, which represent the di erent observational power of an ad-
versary, and then we establish properties for each such classes which state
the complexity of e ectively computing the most powerful adversary.
1 Introduction
Noninterference is widely studied in the security community as a property which
formally speci es the absence of illegal information
ow. In the recent literature
various probabilistic variants of this notion have been proposed which allow for
the speci cation and analysis of probabilistic and approximate security proper-
ties (see, e.g., [17, 15, 11, 16, 2, 9]). In particular, the approximate approach aims
|
