The industry area of embedded and cyber-physical systems is one of the largest; examples are: automotive, avionics, robotics, railway, and production industry. Model-based engineering, esp. component and connector (C&C) models to describe logical architectures, is a common approach to handle the large complexity of embedded systems. Components encapsulate software features; hierarchical decomposition of components enables formulating logical architectures in a top-down approach. Connectors in C&C models describe the information exchange via typed ports. Current development of complex C&C-based embedded systems in industry mostly involves the following steps: (1) formulating functional and extra-functional requirements as text; (2) creating SysML design models of the architecture; (3) developing complete logical model in Simulink; and (4) system implementation in C/C++ satisfying all extra-functional properties. This current development process has the following disadvantages: (a) SysML models are not formalized; (b) check between informal SysML architecture design against the Simulink model is done manually, and thus, error-prone and time-consuming; (c) refactoring of Simulink models needs manual effort in updating the design model; and (d) in most tools extra-functional properties are modeled as comments, and thus, consistencies between them are checked manually. This thesis aims to improve the software development process of large and complex C&C models for embedded systems by providing model-based methodologies to develop, understand, validate and maintain these C&C models. Prototype implementations and an industrial case study show promising results in improving the model-based development process of embedded systems in industry.weiterlesen