Sensor Graphs for Discrete Event Modeling Applied to Formal Verification of PLCs

This paper introduces Sensor Graphs, a discrete event modeling language directed at physical systems with binary and identity sensors (e.g., RFID). The aim of Sensor Graphs is to simplify the modeling of the plant/process that is to be controlled by a discrete controller, for example a programmable logic controller (PLC); thereby making formal verification and other model-based formal methods more applicable for PLC programmers. The formal syntax and semantics of Sensor Graphs are defined and a compact graphical representation is presented. The language is exemplified by modeling a conveyor module and a lab process. For comparison, the latter is also modeled using Statecharts and Net Condition/Event systems. A controller, modeled as a discrete state equation, can be composed with a Sensor Graph of the process in order to form a model of the closed-loop system. It is demonstrated how requirements on such a closed-loop system, based on a PLC program and a Sensor Graph process model, can be formally verified using the model checker Cadence SMV.