Data-Based Approach to Designing Self-Adaptive CPS Software Architecture Models

Cyber-physical systems (CPS) are tight integration of embedded computers and physical devices, which has a wide applications in many areas such as process industry, smart energy, medical care, and national defense. However, it is a challenging task to design CPS software that meets both functional and performance requirements, since various physical devices and software in CSPs are interconnected and complex in structures and behaviors. CPS that controls the operation of physical devices is usually running in a dynamical environment. The environmental parameters will affect the structures and behaviors of CPS. This paper proposes a data-based adaptive software structure model design method. In this method, the software architecture model of CPS is constructed by the hierarchical combination of unit modules. The multi-level formal models for CPS software are based on formalisms of Petri net and temporal logic, in order to precisely specify CPS software architecture model, properties, and refine the relation between different levels. The adaptive evolution of CPS is realized by taking advantage of formal semantics, aspect-oriented method, and data analysis algorithms, which abstracts the function of environmental factors into aspect model and obtains a comprehensive CPS model and basic model. The formal method based on Petri nets and temporal logic provides mathematical expression and analysis means for CPS model. Theoretical analysis and experiments show that the designed method is feasible and efficient.