How SOAFEE Enables Next-Generation Connected Vehicles

Back in the day, cars were admired for the horsepower that their engines delivered or other mechanical attributes. With modern vehicles, on the other hand, software reigns supreme in delivering an array of functions and features for greater safety and a more engaging driver and passenger experience. Ensuring that these systems operate as intended typically involves hours of testing on the road or with virtual prototypes.

Thankfully, software development, testing, and validation are about to become more efficient and effective. It’s all due to a new, industry-led collaborative project designed to produce a cloud-native architecture supported by open-source reference implementations for an array of automotive applications.

The Scalable Open Architecture for Embedded Edge (SOAFEE) project, of which Synopsys is a voting member, is defined by automakers, semiconductor suppliers, open-source and independent software vendors, and cloud technology leaders. The effort builds on technologies that define standard boot and security requirements for the Arm architecture, while adding a cloud-native development and deployment framework along with functional safety, security, and real-time capabilities for automotive workloads.

The shift to software-defined vehicles makes this an ideal time to bring the capacity, scalability, and elasticity of the cloud to automotive application development and testing. Virtual prototyping, in particular, stands to benefit in a cloud environment, which reduces the time- and cost-intensive burden of actual driving tests on roadways. This approach can also open OEMs to additional revenue opportunities as they minimize their investments in non-differentiating middleware. Read on to learn more about how the SOAFEE project aims to accelerate and advance the creation of the next generation of smart, connected vehicles.

The Drive for More Flexibility and Capacity for Virtual Prototyping

In the face of stringent functional safety requirements, performing exhaustive testing of automotive systems has become a complex, costly, and time-consuming endeavor. Traditionally, vehicle developers have tested automotive applications in the field. Because this requires a working vehicle, the testing has happened late in the development cycle. It is also time-consuming and costly, since it requires people to drive the car, interpret the data collected, adjust the software, and so on. Virtual prototyping, on the other hand, supports automotive model-based design methodologies. As a result, automotive semiconductor, tier one, and OEM companies can deploy a more efficient development environment for early hardware/software development, integration, and testing.

There are, however, some serious scalability demands on virtual prototyping hardware. This is where cloud technologies can help. Cloud-based virtual prototyping takes advantage of the virtually unlimited capacity of the cloud, providing the flexibility to scale up or down based on the evolving demands of the design project at any given moment.

Virtual testing in public, private, or hybrid clouds using virtual prototypes can be done earlier in the cycle with fewer resources and cost. Based on abstract software simulation models for SoC and hardware systems, virtual prototypes can be developed at the same time as RTL hardware design, shifting the whole process left for faster time-to-market. What’s more, a continuous integration/continuous deployment (CI/CD) approach can benefit from cloud-based virtual prototyping. With CI, software development teams can address incompatibility issues early on as they merge their code changes into a shared mainline code repository to test functionality, performance, safety, and security. Paired with CI is CD, in which any code build that successfully passes all automated tests is deployed into production. Applying virtual prototypes in a CI/CD setup increases productivity, as it provides for timely dispersion of builds to all the team’s programmers and systems.

Cloud-based virtual prototyping enables fast development and testing of software before deployment through over-the-air (OTA) updates. Simultaneous software testing can be spread across many cloud instances, expanding test coverage, lowering its costs, and enabling detection of errors earlier in the cycle.

A Cloud-Based Software Solution Stack

SOAFEE aims to prescribe a cloud edge development, testing, and validation framework and define how it would work. The result will be an automotive central compute solution stack. Application workloads such as those for autonomous driving would run on top of the solution stack.

According to the SOAFEE project, the bottom level of this stack will consist of standards-based firmware and security interfaces, providing system integrators and software developers a consistent platform for seamless secure boot and system bring-up across compliant hardware. Different components in this framework will re-use existing open standards. The SOAFEE architecture will build on top of these standards and specifications with a reference framework that will standardize critical, non-differentiating middle layers. These middle layers include operating systems, hardware abstraction layers, the hypervisor, and container runtime layers. Because of its open-source components, any middleware or software providers will be able to plug into the solution stack, or even provide the elements of the stack.

What the SOAFEE architecture brings to OEMs is the ability to substantially reduce their middleware investment so they can generate more revenue from their differentiating software. Tesla, for example, sells monthly subscription services for various features, such as its Premium Connectivity offering with live traffic visualization, video streaming, satellite-view maps, and more. Similarly, BMW is monetizing software-driven capabilities such as heated seats and steering wheels via monthly and annual subscriptions.

Developing a custom middleware stack to support these functions and features can require a huge investment in time and money. At the same time, the middleware is not a layer that can be monetized; rather, it’s an enabler of the differentiating software. Being able to tap into an open architecture lets developers and designers focus on what makes their automotive designs unique.