Stealthy Attacks for Partially-Observed Discrete Event Systems

The problem of stealthy attacks for partially-observed discrete event systems is considered. An operator observes the plant through an observation mask that does not allow him to detect the occurrence of certain events (silent events). The observation is corrupted by an intruder who can insert and erase some sensor readings. We construct an attack structure $A$ which guarantees that, when the plant reaches an unsafe state, $A$ can make the operator think that the plant is in a safe state. Based on $A$, we can obtain a stealthy attack structure $A_{s}$ by simply removing from $A$ all the exposing states, all the weakly exposing states and their input and output arcs. The stealthy attack structure $A_{s}$ only contains the possible attacks that can not be detected by the operator.