It's a matter of time before the auto industry has to ditch the CAN bus. The transition to electric vehicles is a great excuse to do it now.
There is a disruption happening in automotive electronics system architecture. It has been going on for nearly a decade and is now picking up speed. It is primarily happening among BEV startups since they do not have historical restriction or ingrained favorite designs and can start with a clean sheet for their electronics architectures.
Tesla has shown the advantages of starting with mostly a clean sheet electronics design using software-centric and OTA-based system architecture. I think everyone is understanding the advantages of remote software updates and the OEMs are investing to get OTA software deployed.
They need to do much more and essentially need to throw away their current playbook using ECU and CAN-based networks. Instead, they need to move to an electronics system architecture that is based on Ethernet and so-called service oriented architecture (SOA)—the sooner, the better. You could also call this a clean-sheet system architecture for automotive electronics. This will take many years, but any new model should use the clean-sheet architecture approach.
This column will explore this topic and why I think it is important that the major auto OEMs get on this system architecture change bandwagon. The next table is a summary of the topics to be discussed.
What is clean-sheet architecture?
Clean-sheet architecture means using the best new technology available and ignoring older system architectures and their restrictions and disadvantages versus newer technologies. This is always a difficult decision because you will not use the experience base in systems, design, supply-chains, testing, maintenance, people expertise and other valuable infrastructures.
You have to develop new infrastructures for the clean-sheet architecture to replace the existing support structure, which will take considerable time. Hence, the clean-sheet architecture development will primarily happen for new auto model designs. With the switch-over to BEVs and many new designs, this is a good time for OEMs to move to better electronics system architectures.
There is good agreement on what characteristics a new automotive system architecture should include. It has to be a software-centric system architecture at all levels that is organized around domain ECUs. This means a hierarchy of re-usable software platforms that are connected by application programming interfaces (APIs)—and used where possible in every ECU. The API connected software platform is synonymous with the SOA.
SOA is heavily used in the IT and cloud industries with major advantages. It is the basis for Amazon AWS, Azure and just about everyone else using cloud platforms. Many of the SOA advantages apply to automotive electronics systems.
Ethernet is the choice for new bus architecture in automotive electronics. There is a large variety of Ethernet versions in terms of speed and physical characteristics with more on the way. These choices should meet all automotive requirements and usually with multiple benefits ranging from speed and weight to cost — if not now, then in the near future.
A major advantage of Ethernet is better characteristics for implementing cybersecurity. With built-in cybersecurity and OTA updates being required in all future auto architectures, Ethernet is the choice.
CAN-bus architecture problems
The controller area network (CAN) bus has been a work horse for the auto industry for about 30 years, but its capabilities are now greatly surpassed by a family of Ethernet versions. There have been many CAN improvements with CAN-FD providing significant advantages. CAN-FD was released in 2012 and is a good short-term fix but not a long-term solution. CAN-FD is not good enough to retain CAN’s customary dominant design wins in future system architectures.
CAN-bus speed is falling behind Ethernet. FlexRay can be substituted, but the experience base and future improvement cannot compete with Ethernet momentum. Cybersecurity weaknesses have been mentioned and will be a growing drawback.
The main advantage of the CAN-bus is the tremendous experience base and support structure from parts to people. These advantages will decline over the next few years and will only delay the inevitable.
SOA & Ethernet advantages
SOA defines how software platforms and components communicate with each other by using APIs. A major advantage of SOA is reusable and interoperable software platforms via service interfaces. These advantages have created a large SOA ecosystem in the high-tech industry.
SOA advantages have sparked growth in chips, systems and software platform and has generated rapid expansion in cloud and SaaS platforms. The auto industry is already seeing the benefits from these SOA-based SaaS and cloud platforms.
AI development is also relying on future SOA improvement and technology advances. AI segments such as machine learning and neural network will increase their impact on automotive system architecture. Autonomous vehicles are especially reliant on AI and SOA in their high-performance computer systems.
SOA-Ethernet innovation potential
The SOA and Ethernet systems are getting tremendous investment and resulting innovation from multiple industries. This is already resulting in much more innovation for the auto industry compared to what happened when much of auto systems were proprietary to automotive.
Multi-industry investments in SOA-Ethernet systems will greatly benefit auto system architectures including software, SaaS and cloud platforms. Similar advances and innovation will be realized by the compute chips, memory chips, sensors and other technology used in auto electronic systems.
OEMs using clean-sheet architecture
The companies using clean-sheet architecture based on SOA and Ethernet are limited to BEV startups. They have followed Tesla’s lead, and several are surpassing what Tesla has done because they started later with more choices of Ethernet versions and larger ecosystems for SOA.
Autonomous vehicle startups are also using SOA-Ethernet system architectures — at least for their AV electronics systems. The question is when will the traditional OEMs follow suit and at least do test cases if they need convincing of the benefits.
Whether I call it clean-sheet system architecture or SOA & Ethernet-based system architecture, it is only a question of time until it becomes dominant. The advantages of SOA-Ethernet are too great, and it makes sense to make the conversion ASAP.
The major auto OEMs have had multiple opportunities to move to SOA-Ethernet system architecture with BEVs since they require major powertrain system architecture changes. I guess the fly-wheel momentum effect of traditional designs makes it difficult to change—especially with the urgency to compete with Tesla and other BEV startups.
Maybe the OEMs are busy introducing SOA-Ethernet architectures in their development programs and we will know in a year or two. In the meantime, the BEV startups are building system architecture advantages that will make them stronger competitors.
Comparing the SOA-Ethernet disruption in the auto industry to the iPhone disruption in the smartphone industry may be a stretch. I think it is in the same class of disruption but will take longer to play out. At least the executives in major OEMs understand the impact of the iPhone and hopefully they will pay more attention to the potential disruption of the SOA-Ethernet architecture and the advantages for the companies that get on this bandwagon.
This article was originally published on EE Times.
Egil Juliussen has over 35 years’ experience in the high-tech and automotive industries. Most recently he was director of research at the automotive technology group of IHS Markit. His latest research was focused on autonomous vehicles and mobility-as-a-service. He was co-founder of Telematics Research Group, which was acquired by iSuppli (IHS acquired iSuppli in 2010); before that he co-founded Future Computing and Computer Industry Almanac. Previously, Dr. Juliussen was with Texas Instruments where he was a strategic and product planner for microprocessors and PCs. He is the author of over 700 papers, reports and conference presentations. He received B.S., M.S., and Ph.D. degrees in electrical engineering from Purdue University, and is a member of SAE and IEEE.