Modeling architectures of cyber-physical systems

Abstract

Cyber-physical systems (CPS) in automotive or robotics industry comprise many different specific features, e.g., trajectory planning, lane correction, battery management or engine control, requiring a steady interaction with their environment over sensors and actuators. Assembling all these different features is one of the key challenges in the development of such complex systems. Component and connector (C&C) models are widely used for the design and development of CPS to represent features and their logical interaction. An advantage of C&C models is that complex features can be hierarchically decomposed into subfeatures, developed and managed by different domain experts. In this paper, we present the textual modeling family MontiCAR, Modeling and Testing of Cyber-Physical Architectures. It is based on the C&C paradigm and increases development efficiency of CPS by incorporating (i) component and connector arrays, (ii) name and index based autoconnections, (iii) a strict type system with unit and accuracy support, as well as (iv) an advanced Math language supporting BLAS operations and matrix classifications. Arrays and their autoconnection modes allow an efficient way of modeling redundant components such as front and rear park sensors or an LED matrix system containing hundreds of single dimmable lights. The strict type system and matrix classification provide means for integrated static verification of C&C architectures at compile time minimizing bug-fixing related costs. The capabilities and benefits of the proposed language family are demonstrated by a running example of a parking assistance system.