AWS for Industries

In the News: Amazon Web Services’ Edge to Cloud Approach to Drive a New Era of Software-Defined Vehicles

This article originally appeared on Frost & Sullivan

stock image of connected car

The pandemic has changed how mobility will be perceived over this decade and beyond. Industry stakeholders are course-correcting; channeling investments away from traditional hardware-focused approaches into a software-centric future, enabling new on-demand services and new revenue streams. This will lay the foundation for a new generation of software-driven solutions that will facilitate the auto industry’s long-term evolution and survival.

To this end, automakers are embarking on a journey of structural reorganization rooted in digital transformation. This revolution will see automakers introduce novel information and operation technologies that enable the development of next-generation CASE strategies. Software will be the foundational aspect of digital transformation and software-defined vehicles (SDVs) will be the gateway to the future of mobility. It will be how automakers implement new business models, open up revenue streams, and offer differentiated customer experiences. As driving experiences change along with customer expectations, features once considered premium will become standard. Incumbent automakers that fail to cater to changing customer demands stand to lose market share to new entrants that offer convenient, cost-effective alternatives. While vehicle architecture will be simplified for new electric vehicles, an increased number of sensors will require new ways to maximize the ROI.

The Need for Software-Defined Vehicles – Benefits and Current Challenges

Until very recently, few automotive insiders could foresee the scale of the software revolution that the industry now faces. Where hardware was once the primary focus of innovation, now it is software that attracts attention and investment. CASE initiatives are all set to shape the industry with Frost & Sullivan projecting OEM spend on software infrastructure across the connected and autonomous pillars to reach around $2,300 million by 2030.

But will SDVs be possible if the complex E/E architectures of the current generation of vehicles remain unchanged? What about the talent needed to develop associated technologies? How will cloud help accelerate the journey of SDV? Without the proper expertise, development cycles will end up being time-consuming and resource-intensive.

Beyond unlocking new safety, comfort and convenience features, SDVs offer several advantages over their hardware-defined predecessors. Today, over-the-air updates (OTA) offer limited software upgrades, help address few software issues and troubleshooting diagnostic trouble codes require a dealership visit. With an SDV, car manufacturers will have seamless development and deployment of new software services including infotainment, security, as well as alter functional capabilities, such as powertrain, driving dynamics, range extension for EV’s, etc., of the vehicle. This highlights the need for the industry to break out of a silo mindset that sees the vehicle and the cloud as discrete and, instead, envision an edge to cloud approach to the vehicle. This edge to cloud approach will be the key premise and driving force underpinning SDV development. In a software-defined future, automakers will need trusted technology partners that can provide the tools and talent necessary to explore innovative business models and capitalize on emerging revenue opportunities.

As the industry works to embrace such transformation, it faces major challenges including:

  • The significant amount of legacy technology—for example, existing E/E architectures and ECUs functioning in silos— that cannot be scaled up.
  • The lack of data ingestion connected with hardware consolidation and the ability to provide normalized and secured data access that will allow developers to build new microservices.
  • The need to implement a container approach for cross-domain updates, services and new feature sets with the objective of not just developing containers in the cloud but also deploying them inside the vehicle.
  • The challenge of fusing modern CI/CD tools and frameworks with the existing “V” development model to ensure that agility can be brought to in-vehicle software experiences.

Expected Software-Defined Vehicles outcomes based on AWS’s approach

Seamless development and deployment

  • Enables simplified architecture that would decouple h/w from s/w and allows the vehicle to seamlessly interact with the cloud, The whole cycle of architecture development and software development around the vehicle is considerably reduced. This represents notable advantages in terms of time-to-market as well the amount of resources an automaker has to expend on in-vehicle architecture.
  • Challenges such as OTA integration and implementation can be addressed with the coming together of CI/CD and modern automotive DevOps culture. This will create major benefits by deploying containers- including software upgrades, addressing the bugs and new applications – in a much more agile atmosphere. The entire development process of how software is written, tested and deployed in vehicles will change significantly.  

ML at the edge

  • Holistic approach of data ingestion, storage, ML models and deploying inferences back to vehicle and creating data insights and ML experiences in the vehicle.

These two outcomes cannot be realized in silos because there is a need to achieve as much decoupling as possible from the hardware and the software, irrespective of the vehicle make/model, across suppliers and tier-1s to receive the maximum benefit from a generalized automotive development platform in the years to come.

Role of AWS and its partners in driving software-defined solutions for the future

AWS distributed architecture approach from edge to cloud is holistic and can be viewed as a three-pronged approach.

  • First is the vehicle itself which would require a generalized platform offering of in-vehicle compute and integrated with managed IoT services.
  • Second is the edge of the network itself; where AWS has integrated with telecommunication service providers to extend the compute storage at the edge of the network. This will enable ultra-low latency compute, storage and enablement of IoT services.
  • Cloud itself becomes the third piece where a lot of flexibility scalability and on-demand services such as IoT/ML services. In addition, edge services also help customers to run the cloud services/instantiations outside the AWS infrastructure.

There are several services that AWS can offer but follows an ecosystem “first” approach where they with a wide set of partners and ecosystems players to deploy an edge to cloud approach.

One such example is Continental Automotive Edge (CAEdge), deployed on AWS is a virtual workbench offering toolchains to develop supply and maintain software-intensive system functions. The goal here is to provide OEMs and associated partners with a development environment for software-intensive vehicle architectures. This allows efficient development/test and secure OTA roll out directly to vehicles.

On the other hand, AWS is also a partner in the industry-wide collaboration through SOAFEE, a cloud-native architecture for mixed-criticality automotive applications. It includes an open-source reference implementation to enable commercial and non-commercial offerings.

AWS defines the architectural considerations for an SDV across four key areas. These are set to evolve as AWS and its partners push forward on developing software-defined solutions.

  • Data Democratization with Simplifying Developer Experience: AWS and BlackBerry QNX co-developed an Intelligent Vehicle Data Platform, Blackberry IVY. This platform enables automakers to share insights with a massive developer community, where developers can build ML-powered capabilities and contextually aware experiences without requiring specialized automotive skills. In addition, customers can also utilize the rest of the AWS portfolio, to offer deeper business and operations analytics and the ability to deepen and augment data lakes.
  • Automotive DevOps: Automotive DevOps play an important role in edge to cloud. Flexible management of CI/CD, along with mixed critical container orchestration, environmental parity and application-level networking come together to support an agile, modern automotive DevOps environment. This approach will set the immediate foundation for SDV as E/E architecture gets updated in next-generation vehicles.
  • Hardware Architecture and Design: Consolidated E/E architecture with high-performance compute (HPC) and enhanced networking capabilities will enable HW and SW decoupling. This will help to reduce wiring costs and reduce vehicle weight. With holistic edge to cloud architecture, ARM-based in-vehicle CPUs like NXP S32g to ARM-based cloud compute instance from AWS will offer environmental parity to simplify developer flow. Graviton 2 is AWS’s ARM-based, CPU-based instance. It offers 30% performance improvement and also significant cost optimization which is the foundation for environmental parity.
  • Hardware and Software Abstraction Layers: Hardware and software abstraction play a big part in SDVs. They come with open source services as well as some of the services that tier-1s offer that enable the decoupling. Apache TVM and Greengrass V2.0 are the open-source services provided by AWS. The idea behind the software abstraction and orchestration techniques is to run containers at the edge and run AL/ML instances at the edge. An integrated approach using Greengrass and SageMaker Neo allows customers to run AL/ML instances in a much more compact way.

AWS is bringing all this together by providing its tier-1 partners with a way to build their solutions for which they can use Greengrass, a general-purpose and foundational compute service. Customers across manufacturing, IoT services and automotive can seamlessly connect their edge devices to any AWS/third-party services.

The whole idea behind this approach is it offers flexibility and homogeneous heavy lifting. With innovation in its DNA, AWS has become integral to customers’ strategic journey to reduce time to market. As exemplified by NXP, which is already integrated with Greengrass, AWS is vested in an ecosystem-based approach. A combination of NXP S32G and AWS makes it possible for customers to deploy cloud-native applications and also have ML use cases.

Conclusion – What does all this mean for the auto industry?

The shift in E/E architecture and the related vehicle evolution demands a closer partnership between car manufacturers, suppliers and tech vendors. AWS, along with its partners, is creating the much required comprehensive reference platform that spans the vehicle to the cloud, while reducing the complexity of software development and system integration. This offers vehicle manufacturers the prospect of improving life cycle management and developing revenue-generating features that they can offer to customers, all of which result in deeper, more connected relationships with customers.

The SDV will create opportunities for both consumers and OEMs, many of which haven’t even been conceived yet.  Frost & Sullivan believes that by 2025-2026, vehicle programs will have some form of software-defined use cases and will continue to expand, propelling new business models and data monetization opportunities. AWS offers secure, scalable and cost-optimized services for the software-defined future. The idea is to have a single platform that goes across different domains, different makes and different models. This will allow OEMs to invest in strategic platform/use cases, empowering them to think about pushing the boundaries on experiential customer services.