Access Official Technical PDF

Share Summary:LinkedInEmail

Virtual ECU co-simulation brings software models and virtualized microcontroller behavior together to test embedded systems without relying solely on physical control units. Although the approach is rooted in automotive development, its central engineering challenge is relevant across industrial automation: how can teams collaborate on complex systems when models, tools and intellectual property belong to different organizations?

Work by the vECU-MBD Working Group, a cross-industry initiative involving automotive manufacturers, suppliers, semiconductor companies, tool vendors and research institutions, explored this question through proof-of-concept activities. Its findings illustrate both the potential of virtual ECUs and the practical requirements for making multi-party simulation workable.

Why virtual ECU co-simulation requires collaboration

A virtual ECU, or vECU, represents an electronic control unit in a software-based environment. In model-based development (MBD), engineers use models to design, simulate and verify system behavior. A virtual ECU can extend simulation to software running at the binary level: the working group described using microcontroller simulators with Simulink in a Simulated-Processor in the Loop Simulation (SPILS) arrangement.

This makes it possible to examine how software behaves in a virtualized controller and, in some cases, how multiple controllers interact. The working group identified use cases including multi-ECU systems, fault injection and cloud-based co-simulation between different users. Fault injection, for example, can be used to examine binary-level software behavior under hardware or memory fault conditions.

However, a simulation spanning multiple engineering organizations introduces challenges beyond the models themselves. The group identified barriers involving model intellectual property, tool compatibility, modeling skills, model precision and operational guarantees. Addressing these issues calls for cooperation across the supply chain rather than isolated technical solutions.

Co-MBD: sharing results without transferring models

One proposed approach is Collaborative MBD, or Co-MBD. Instead of sending model files to another organization, a model provider makes a model available as a service. The model remains on the provider’s machine, runs in the provider’s environment, and returns results to the model user. The user can therefore work with another party’s model without receiving the model itself.

This arrangement can help a model provider retain confidentiality or protect intellectual property. For the user, it can reduce the need to maintain a model supplied by another company and make use of computing resources provided through a cloud environment. The approach depends on agreed interfaces between models, tools and execution environments. The working group referenced D-EIPF, or Design Electric Integrated Platform, in its Co-MBD concept.

In a representative automotive scenario, a supplier’s engineer could use an OEM’s car model while retaining the supplier’s own ECU model. Each organization prepares and operates its own portion of the virtual test environment, with communication between components enabling the combined simulation.

Connecting virtual test environments with ASAM standards

Interoperability is essential when different organizations use different tools and models. The working group examined how ASAM standards could support cloud-based co-simulation, including ASAM XIL and MCD-1 XCP.

ASAM XIL provides an API standard for communication between test automation tools and test benches. MCD-1 XCP is a bus-independent, master-slave protocol used to connect ECUs with calibration systems. In the proposed virtual test environment, XCP communication and related interfaces form part of the connection between a virtual ECU and a measurement or calibration tool.

The group’s planned proof of concept was structured as a sequence of steps, beginning with assessment of the existing setup and progressing toward cloud migration and a larger vehicle-model scenario. The plan emphasized reusing existing models and tools where possible. This staged approach helps teams separate basic connectivity questions from later challenges involving cloud execution and more complete system models.

Key engineering issues to resolve

The working group’s 2020 planning materials identified several operational concerns that teams need to address when connecting virtual ECUs, calibration tools and cloud environments:

  • Startup and synchronization: Different startup sequences across virtual ECUs and measurement tools could prevent synchronous measurement. Startup should not depend on one fixed order.
  • Network and security controls: PC security mechanisms may restrict network or firewall changes, making it impossible to connect the virtual ECU and measurement tool. Device licensing requirements may also affect the execution environment.
  • Multiple master connections: MCD-1 XCP does not define the behavior of a multi-master connection. If multiple tools connect to the same virtual ECU, command sequences could interfere with measurement. The group identified exclusive control as a possible safeguard and planned to investigate whether multi-master use cases warranted feedback to ASAM.
  • Authentication and disconnection: Teams need to consider Seed&Key support and a way to terminate an XCP connection if a tool remains connected unintentionally.
  • Cloud-based tool access: Tool licensing, simultaneous user access and the location of A2L files require consideration when measurement and calibration tools are hosted in the cloud.

These are not merely configuration details. They affect whether a shared test environment is secure, predictable and usable by engineers from multiple organizations.

Practical implications for automation and embedded teams

The working group’s activities also produced guidance on introducing virtual ECUs and procuring or integrating models, alongside a specification for a CAN bus model used in MBD. Its broader lesson is that virtual validation depends on processes as much as simulation technology. Teams need to define model ownership, interface expectations, execution responsibilities and access controls before expanding collaboration across organizational boundaries.

For industrial automation engineers and system integrators, the same considerations apply when testing embedded controllers or coordinating models across suppliers. Begin with a bounded use case, verify tool-to-model communication, and document how startup, access and disconnection will work. Then extend the environment as the interfaces and operating model are validated.

Conclusion

Virtual ECU co-simulation offers a way to test embedded software and system interactions while keeping models with their owners. Co-MBD formalizes this idea by treating models as services, but successful implementation requires compatible interfaces, clear operational rules and deliberate handling of security and shared access. For technical decision-makers, the practical priority is to validate these foundations in stages before relying on a multi-party virtual environment for broader development work.

Access Official Technical PDF

About The Author