Incremental verification of Co-observability in discrete-event systems

Abstract

Abstract-Existing strategies for verifying co-observability, one of the properties that must be satisfied for synthesizing solutions to decentralized supervisory control problems, require the construction of the complete system model. When the system is composed of many subsystems, these monolithic approaches may be impractical due to the state-space explosion problem. To address this issue, we introduce an incremental verification of co-observability approach. Selected subgroups of the system are evaluated individually, until verification of co-observability is complete. The new method is potentially much more efficient than the monolithic approaches, in particular for systems composed of many subsystems, allowing for some intractable state-space explosion problems to be manageable. Properties of this new strategy are presented, along with a corresponding algorithm and an example.

Hierarchical Interface-Based decentralized Supervisory Control

“…Most of the material in this chapter first appeared in a joint paper with R. Malik, R. Leduc and L. Ricker [LLMR14].…”

“…This is done this way because interfaces represent the behaviour provided by its low level and the information needed to verify that it is co-observable is typically present at the low level but not the high level. To avoid having to repeat this information at the high level, we use the results of [LLMR14] that allow us to treat supervisors as if they are plants once we verify they are coobservable. By treating interfaces as plants at the high level, we allow the high-level supervisor to be more permissive in general as there will typically be fewer strings that can cause the co-observability verification to fail.…”