Virtual System Development Platforms For Safeguarding Complex Microelectronic Systems

Electronic systems are rapidly becoming more complex. This impacts almost all domains in which electronics are used today — especially industrial applications, medical engineering, communications engineering and, of course, automotive applications.

What’s changed is the addition of huge numbers of sensors and actuators that interact with the environment, the local integration of highly complex algorithms for evaluating the collected data, and the multitude of standards for wired or wireless communication with other systems.

Within the electronic systems themselves, there is also a significantly increase in interactions between the system parts. Various physical domains, such as electrical, thermal, mechanical, must be considered at the same time. Today, the development of high-performance ICs often no longer works without the simultaneous optimization of the package and the circuit board.

Consider a control unit in a vehicle, for example. It can only be designed robustly if you know the details of how and where it will be installed, including the assumed environmental boundary conditions. At the same time, the interaction between hardware and software is becoming increasingly intertwined, with the execution of complex algorithms distributed across available and redundantly designed processor elements. Future updates of the software, and its secure processing on manipulation-protected hardware, must already be planned in the development stage.

Add up all of these developments and there are two clear consequences. First, a lot more effort will have to be expended in the future on verification and validation in order to develop safe and robust products. And second, this protection will have to be supported by virtual methods. For example, the developers of vehicles assume that for the approval of highly automated driving functions, 1 million test kilometers are necessary, of which only about 10% are actually driven on the road. The further verification of the correct implementation, as well as the error-free interaction of the subsystems, will take place in a virtual environment.

Such virtual development environments must therefore meet the following important requirements:

• Joint mapping of hardware and software components
• Integration of different models of the individual system parts
• Resources for handling highly complex systems
• Ability to integrate already existing real system components

Some of these requirements already have been implemented in system development platforms. However, the second requirement for the integration of models from completely different sources demands a fundamentally new approach. No one, for example, will expect that the mechanical model of an engine, the electronic description of the control IC, and the thermal model for the heat dissipation of both components will be modeled in the same tool and described by the same language. Tools adapted to the respective special disciplines are to be used in such cases. These tools can efficiently process the models in the individual areas by exploiting the valid modeling paradigms.

Improved quality begins with the combination of different models and previously separated domains into a comprehensive virtual design platform. The basic characteristic of such a platform is therefore the integrated processing of models from very different sources in a verification environment.

A key goal is to check the interaction of individual system components at a high level of abstraction in the context of the other components, as well as within their environment. Many principles from other areas, such as Model-Based Systems Engineering (MBSE) and Hardware-in-the-Loop-Simulations (HiL), can be integrated into such a virtual development platform. Vital to the success of such a platform is not only the ability to verify the functional system behavior and its parts, but also the mapping of non-functional properties such as power consumption, as well as safety, security and reliability.

Therefore, the focus is not, as in the past, on a common description language, a uniform simulation tool or the modeling of individual components. The core of a virtual design platform instead must be a generic and open interface that enables the integration of different models, languages and modeling principles. It supports the entire development cycle, from the architectural concept to the verification of the correct interaction of the individual system components, to the integration of real hardware components. It has to work at different levels—from the single IP, through the SoC and an ECU, to the whole vehicle and its environment.

FMI (Functional Mockup Interface) represents one such interface. It allows different models to be coupled with very different simulation principles. The uniform interface is used to exchange data between the models and to synchronize the sequential processing of the individual models. For this purpose, a hierarchy of models is defined according to the master-slave principle. FMI is by now a component of many tools in the field of systems engineering, and is supported by languages such as Modelica.

Overall, the topic of virtual prototyping platforms is of fundamental importance for the development and verification of complex microelectronic systems. This topic calls for new principles in the integrated description of heterogeneous multiphysical systems and their standardization.