In this thesis, the author proposes a circular system development model which considers all the stages in a typical development process for industrial systems. In particular, the present work shows that the use of virtual prototyping at early stages of the system development may reduce the overall design and verification effort by allowing the exploration of the complete system architecture, and uncovering integration issues early on. The modeling techniques of this research are based on VHDL-AMS, yet supporting other modeling languages such as C/C++, SPICE, and Verilog-AMS, together with integrated simulation tools. Contrasting with conventional approaches, it is shown that the proposed methodology is adapted for small-scale Cyber-Physical Systems (CPS) design and verification thanks to the modularity and scalability of the modeling approach. The proposed modeling techniques enable seamlessly the CPS design together with the implementation of their subsystems. In particular, the contribution of this work improves the virtual prototyping approach that has been successfully used during the development of smart electrical sensors and monitoring equipment for high and medium voltage applications. The design of the measurement and self-calibration circuits of a medium voltage current sensor based on the Rogowski coil transducer is presented as an example. The proposed small-scale CPS design methodology based on virtual prototyping, namely VP-based design methodology, uses important theoretical concepts from layered design, component-based design, and platform-based design. These foundations are the basis to build a modeling methodology that provides a vehicle that can be used to improve system verification towards correct-by-design systems. The main contributions of this research are: the re-definition of the system development lifecycle by using a virtual prototyping methodology; the design and implementation of a model library that maximizes the reuse of computational models and their related IP; and a set of VHDL-AMS modeling guidelines established with the purpose of improving the modularity and scalability of virtual prototypes. These elements are key for supporting the introduction of virtual prototyping into industrial companies that can thoroughly profit from this approach, but cannot commit a specific team to the creation, support, and maintenance of computational models and its dedicated infrastructure. Thanks to the progressive nature of the proposed methodology, virtual prototypes can indeed be introduced with relatively low initial effort and enhanced over time. The presented methodology and its infrastructure may grow into a bidirectional communication medium between non-expert system designers (i.e. system architects and virtual integrators) and domain specialists such as mechanical designers, power electrical designers, embedded-electronics designers, and software designers. The proposed design methodology advocates the reduction of the CPS design complexity by the implementation of a meet-in-the-middle approach for system-level modeling. In this direction, the modeling techniques introduced in this work facilitate the architectural design space exploration, critical cross-domain variable analysis (especially important in the component interfaces), and system-level optimization and verification.