The DS-Pnet modeling formalism for cyber-physical system development

Abstract

This work presents the DS-Pnet modeling formalism (Dataflow, Signals and Petri nets), designed for the development of cyber-physical systems, combining the characteristics of Petri nets and dataflows to support the modeling of mixed systems containing both reactive parts and data processing operations. Inheriting the features of the parent IOPT Petri net class, including an external interface composed of input and output signals and events, the addition of dataflow operations brings enhanced modeling capabilities to specify mathematical data transformations and graphically express the dependencies between signals. Data-centric systems, that do not require reactive controllers, are designed using pure dataflow models. Component based model composition enables reusing existing components, create libraries of previously tested components and hierarchically decompose complex systems into smaller sub-systems. A precise execution semantics was defined, considering the relationship between dataflow and Petri net nodes, providing an abstraction to define the interface between reactive controllers and input and output signals, including analog sensors and actuators. The new formalism is supported by the IOPT-Flow Web based tool framework, offering tools to design and edit models, simulate model execution on the Web browser, plus model-checking and software/hardware automatic code generation tools to implement controllers running on embedded devices (C,VHDL and JavaScript). A new communication protocol was created to permit the automatic implementation of distributed cyber-physical systems composed of networks of remote components communicating over the Internet. The editor tool connects directly to remote embedded devices running DS-Pnet models and may import remote components into new models, contributing to simplify the creation of distributed cyber-physical applications, where the communication between distributed components is specified just by drawing arcs. Several application examples were designed to validate the proposed formalism and the associated framework, ranging from hardware solutions, industrial applications to distributed software applications.