Digital Twin And The Implications For Semiconductor Suppliers

The road to full autonomy requires connecting the virtual and physical with real-world feedback.

popularity

Automotive is a prime example of one industry where advanced semiconductors are forcing change across all aspects of vehicle development. Accelerated investment is fueling this industry wide evolution, driven by new business models, environmental regulations, and the evolution toward fully autonomous, software-defined vehicles.

The demand for fully and semi-autonomous systems has a particularly significant impact on methodologies and flows for system-level verification and testing. To achieve adequate testing of the overall state space, OEMs realize the need to virtualize all phases of the product lifecycle, and this includes emphasizing testing with a virtual representation of the vehicle, its systems, and the surrounding environment. The virtual representation is often referred to as the digital twin.

If you have looked into the concept of a digital twin before, you may have ended that investigation scratching your head as to its meaning, as both industry practitioners and vendors aren’t aligned on its true definition.

At Siemens, our definition of a modern digital twin is more comprehensive than traditional viewpoints. In a nutshell, our belief is that there exists only one dynamic digital twin of a product, and that digital twin represents the product as it traverses the numerous design phases and models of physical behavior across the entire development lifecycle.

Key digital twin characteristics include:

  • A highly accurate virtual representation of the physical product
  • A virtual model acted upon across the lifecycle to simulate, predict, and optimize the product
  • A virtual model that takes on multiple representations throughout the development lifecycle
  • A virtual model that evolves and needs to be managed throughout the lifecycle
  • A closed-loop lifecycle enabling bi-directional connectivity and feedback between the physical and virtual worlds

Fig. 1: The phases of a modern digital twin implementation.

Digital twin for semiconductor suppliers

At first glance, one might conclude that semiconductor companies have been using digital twins of ICs for years, and that isn’t far from the truth. The complexities and cost of failure (i.e., silicon bugs resulting in multiple tapeouts) are simply too high to forego thorough design and verification in a virtual environment.

The good news is that current best practices have already implemented aspects of the digital twin core tenets. The bad news is that new challenges have emerged, and fundamental gaps have been identified, forcing semiconductor providers to re-evaluate their development processes and activities.

The road to full autonomy requires a digital twin that virtually models systems and connects the virtual and physical worlds. This shift in thinking has new implications and challenges for semiconductor suppliers, including:

  • The creation and delivery of models
  • Tighter coupling of verification technologies among integrators and suppliers
  • Feedback of real-world data
  • Traceability across the traditional V-model

Let’s take a closer look at the new challenges suppliers face.

Model creation and delivery

Successfully designing and verifying a vehicle in a virtual environment is dependent on the accuracy of the design models available. The industry is already seeing demand for accurate models of these ICs, and the creation and delivery of these models is quickly becoming a supplier function, as manufacturers are unable to accurately model functions implemented within complex semiconductors.

Tighter coupling of verification technologies

Companies always strive to run fast and run accurate. Often, these are viewed as opposing forces, but that doesn’t have to be the case. Existing verification technology allows for OEMs and suppliers to work collaboratively throughout their respective development lifecycles.

Automating real-world feedback

At this time, the automated capture and feedback of real-world data isn’t common practice, especially data captured at the IC interfaces. SoC complexity, configurability, and the migration to software defined platforms make testing the complete state space across all applications extremely challenging. Therefore, it is imperative that data captured in the real-world is fed back into the development lifecycle. This capability serves both reactive and predictive scenarios.

Fig. 2: Feedback loop from physical to virtual worlds.

Establishing traceability within the digital thread

Traceability is a core tenant of the digital thread, and fundamentally, there are two types of traceability that semiconductor vendors must consider.

Internal traceability comprises the traceability across lifecycle activities and the data generated at each lifecycle phase.

External traceability is the traceability between suppliers and integrators.

Up to this point, veteran automotive suppliers have managed traceability through manual processes and/or home-grown infrastructure. On the other hand, companies new to automotive and safety-critical flows may have not been required to demonstrate any traceability on their previous projects. In fact, Siemens has engaged directly with semiconductor suppliers to perform lifecycle analysis, and insights gleaned during these engagements highlight that those existing processes are incurring substantial overhead.

Fig. 3: Traceability from requirements to implementation and verification.

Processes have not scaled to meet the demands for both internal and external traceability. Companies who invest now in a modern digital thread will be well positioned to succeed as IC complexity rises among compressed market windows.

Conclusion

As the automotive industry evolves, so must the semiconductor suppliers that propel it. Disruption to the traditional supply chain model and a path to full autonomy have real implications in how suppliers build and deliver silicon. Companies who start the migration to a modern digital twin paradigm will be well positioned to tackle tomorrow’s challenges today.

Fortunately, the journey to a modern digital twin is not an all or nothing decision. Companies can realize the value of a digital transformation in a piecemeal approach with minor incremental investment. Siemens Digital Industries Software provides a comprehensive suite of solutions to help companies succeed as they move forward on the road to tomorrow’s electronics-centric vehicles.

If you’d like to learn in more detail about why a digital twin is important to your business, the value a digital twin offers, and the specifics of its impacts on semiconductor suppliers, please read our full paper, Peeling the onion of an Automotive IC Digital Twin.



Leave a Reply


(Note: This name will be displayed publicly)