On the Decidability of Model-Checking Information Flow Properties

Deciding Selective Declassification of Petri Nets

“…The result of [3] and its relation to [1] (and therefore also to the present paper) has already been discussed in [1]. The main results of the paper [4], which came to our attention only after [1] had been published, concern the undecidability of several of the security properties described in [8] for pushdown systems. These sets of results are not directly comparable, since pushdown languages and Petri net languages are not comparable either.…”

“…These sets of results are not directly comparable, since pushdown languages and Petri net languages are not comparable either. At the end of [4], we also find a decidability result. This result pertains to a very restricted system model.…”

“…More precisely, it is shown there that a property called "weak non-inference", which falls neither into Mantel's framework [7,8] nor into the NDC/INI/INISD framework of the present paper, and which is undecidable even for finite-state systems, becomes decidable for pushdown systems if one limits both the set of visible actions and the set of secret actions to cardinality 1. Nevertheless, the approaches in the present paper and in [4] lead to the question (which is so far open, to our knowledge), as to which -if any -of the many trace-based transitive security properties of [7,8] are actually decidable for unbounded Petri nets, and to a similar question for the intransitive security policies discussed in section 8.…”

“…The only other decidability results of trace-based security properties for infinitestate systems we know about are described in [1,3,4]. Of these papers, [1] is a direct predecessor of the present one.…”

The Complexity of Intransitive Noninterference

“…Union(s · a, t · a) ; 22 if obs u (s · a) = obs u (t · a) then 23 return compute-witness(s · a, t · a, , ) ; 24 return "secure" add ((s · a, t · a), (s, t), (a, a)) to store; 23 insert (s · a, t · a) into the list P ;…”