Constrained Sensor Selection for Discrete Event Systems Modeled by Petri Nets

This paper studies how to place a minimum number of sensors in discrete event systems modeled by partially observed Petri nets while maintaining structural observability. When the sensors are constrained to be associated with specific sets of transitions (which could be the result of physical or geographical constraints), the resulting constrained optimal sensor selection problems are shown to be reducible to the optimal place sensor selection (OPSS) problem introduced in previous work. These reductions establish the central role that the OPSS problem plays in our sensor selection problem formulation. Therefore, in order to obtain a solution with known performance guarantees (that we precisely characterize), we propose in this paper a heuristic method based on a reduction from the OPSS problem to the set cover problem.