Automotive Complexity, Supply Chain Strength Demands Tech Collaboration

Relationships in the automotive ecosystem stretch to deep technical developments as the industry pivots to electrification and autonomy.


The automotive supply chain is becoming more complex and collaborative, changing longstanding relationships between automakers and their suppliers in ways that would have seemed unimaginable even a couple of years ago.

Rather than just developing parts for a tightly defined specification, suppliers are taking an increasingly active role in determining how various technologies are combined, what gets prioritized, and how the overall systems architecture is partitioned. Software companies, OEMs, and Tier 1s and 2s are now working together to integrate car-specific functions with software applications, breaking down silos and replacing them with others in order to reduce the complexity and cost of adding new features and functions. The end result is more chips, more software, much more integration, and a fair amount of confusion surrounding these changes.

“When I first started at Ford, there was really only one chip — one computer module, and it was the engine controller,” said Judy Curran, who spent more than 30 years at the carmaker before joining Ansys as senior chief technologist for automotive. “I worked on the software, and that one controller. Fast forward three decades, there are now 115 million lines of code, 150 controllers, 1,000 chips, and huge change. The industry is well over 100 years old, but the last couple of decades have seen tremendous change, and the skills needed are so much different. When I hired in, there were a small number of computer software engineers and a whole bunch of mechanical engineers. Now, we need a whole bunch more electrical software engineers, and there’s a shortage. The challenges for these engineers in the automotive industry is maybe more significant than in some other industries because of the [technological] change and the culture change.”

The genesis of this upheaval is inextricably tied to the smart phone revolution. “It was when people realized what the phone could do for their life,” Curran said. “That led people to ask why their car was not able to know them and understand what they want. ‘Why do I have all these buttons? Why isn’t it upgradable like the phone is upgradable?’ Then, when Tesla came out and started the whole vehicle based on the software, people realized this is the way of the future. These electrical engineers are taking this historical vehicle and re-architecting it. Most OEMs don’t have a clean sheet. They have an existing business, so there’s a lot of work currently with these 150 modules coming down to these new architectures that are larger, supercomputer-type modules, and managing all that software with the suppliers, but in a totally different relationship.”

For automotive OEMs to adopt new architectures requires a fundamental shift in how they approach their supply chain. Modules cannot be developed individually by multiple Tier 1 and Tier 2 suppliers. Instead, they need to be developed in sync, with an understanding of how each is characterized and how they can be fully integrated. “You can’t have Bosch do one module, Continental do another module, Aptiv do a separate module, then plug them in on the assembly line and think the experience is going to be great,” she said.

This has created disruption throughout the supply chain. “OEMs are competing with each other for dollars, mindshare, and market share,” said David Fritz, vice president of hybrid-physical and virtual systems automotive and mil-aero at Siemens Digital Industries Software. “There’s not a whole lot of cross synchronization between OEMs. However, what is happening is that the Tier 1s are lining up behind OEMs in almost one-to-one relationship with them because that’s their secret to survival.”

For example, Bosch is very committed to the Volkswagen Group brands. “It’s multiple OEMs, but they’re in the same group,” Fritz said. “For Denso, it is Toyota. Then you look at others like Subaru and Mazda. They don’t have enough volume to have that kind of relationship with a Tier 1, and as a result they’re getting further behind.”

NVIDIA, and the ecosystems it plays in, represents another dynamic. When Jensen Huang left Intel to start NVIDIA, Intel wasn’t particularly interested in GPUs. NVIDIA was successful selling its GPUs into the Intel customer base, so Intel developed its own GPU. NVIDIA, in turn, pivoted into the gaming market and put GPUs on a PCIe card. What happened at Nvidia is akin to what is happening in the automotive supplier space. OEMs realized the Tier 1s were not doing what they needed, so the OEMs pulled more development in-house.

“This is happening at every single OEM,” Fritz said. “Every OEM is trying to build up their own SoC team, and is in the process of doing that.”

Even with the various business model configurations, all of these companies are trying to get to the same place. That involves more collaboration in both the design and testing areas.

“One of the big problems here is software integration along the supply chain,” said Larry Lapides, vice president of sales and marketing at Imperas. “As each supplier adds features, they’re going to add software to that. They’re not supplying source code, typically. But how do you test this executable? How do you effectively do an incoming quality inspection. They’ve developed software to spec and they’ve tested to the spec, but whether you’re receiving metal parts or software, you still want to have some sort of an incoming inspection. Doing that in a fast and comprehensive manner can be difficult. And then what happens if a test fails and you don’t have the source code? The interesting thing is that virtual platforms are being used as a vehicle for communication.”

Imperas already collaborates with the big three EDA companies and their SystemC simulators, but this collaboration is accelerating due to Imperas’ RISC-V models, Lapides said. “RISC-V is in the process of contributing to the restructuring of that supply chain, as well in automotive, because we’re getting to the point that we are near end-of-life on the traditional automotive architectures that are the old workhorses. They are looking old.”

Some of those RISC-V designs will be in the form of chiplets, which automotive OEM see as a way to streamline integration and customization. “The OEMs are saying, ‘If I’m going to go to 3nm, which is $75 million to $100 million for a mask set, plus a huge development team, where am I going to get those people? That’s not the biggest pool of talent in the world,” said Fritz. “‘How do I do that?’ Chiplets. So now they’re saying, ‘I can have these companies, maybe even startups, developing a chiplet.’ It’s more cost-effective for them, because those chiplets can be sold to many customers and across multiple market segments and get the volume up. But they don’t have to worry about packaging. They don’t have to worry about inventory. All they do is get the wafer, package a few, do individual tests, and call it done. But if you’re an OEM, then what you’re saying is, ‘Now I don’t have to build such a big team. I need a packaging team, a test team, and another team that’s going to help us to decide which combination of chiplets has the right operation for our workload? Our workload in automotive is determined by which ECUs we are going to consolidate. How much compute do we need for AD? How much for ADAS? Do we need a hypervisor?’ There are many variables there.”

Many of those involve multi-disciplinary technology issues, as well.

“There’s a renaissance going on with electric vehicles,” said Chris Mueth, digital twin program manager at Keysight Technologies. “That’s where everything’s moving to. But an automobile is a system of systems. The drive train itself is a system. The autonomous driving system that controls the automobile in a lot of ways and keeps us out of trouble — and where things are probably going in the future — is a system itself. These systems can have a number of different systems in them, along with a lot of sophisticated components. Even the radar module is a stacked assembly. It has a digital content, RF content, antennas. It’s mechanically and electrically tied together. It’s a multi-disciplinary design effort, as is a lot of the autonomous driving system itself.”

For decades, aerospace defense companies used model-based systems engineering (MBSE) approaches. Automobile companies followed, although not with technology that was as advanced as in aerospace. “With the complexity of systems today, a systems-level approach needs to happen here,” Mueth said. “There is a hierarchy, starting with systems of systems, the systems and the components. But around that you have all the different engineering disciplines, including mechanical, multi-physics, RF, and digital, and all of these have to come together in a collaborative effort. To help with that, there is a process and data management function within the engineering lifecycle. Whether you’re in the concept phase, the design phase, prototyping, test validation, in the lab, or you’re doing integrated testing, all of that has to be tied together because you need to take the test data and feed it back to the virtual domain and validate things. Everything has to be validated to the nth degree in a high-reliability application.”

Fig. 1: Multiple dimensions of complexity in automotive. Source: Keysight

Fig. 1: Multiple dimensions of complexity in automotive. Source: Keysight

However, Mueth doesn’t believe this is being done very well in the industry today. “Product lifecycle management systems, which are supposed to do the job, weren’t really designed for the special needs of electrical engineering. There’s specialized data that’s produced. There are models that need to be created. They’re not adequately addressing the issues. If you’re on the CAE side of things, there are some synergies with PLM systems, because a lot of the mechanical CAD designs end up being the assembly prints in the long run. But on the electrical side, it’s almost non-existent. It’s a real disconnect. Then, it’s not hard to understand that people need to simulate as much as they can early on before they start building expensive prototypes. You want to do that shift left function and get high confidence in your design as early as you can, and not wait until you’re doing prototype verification to find out if you have issues.”

Component-level workflows are still being developed for automotive. And while it’s possible to shift more left in the design flow with faster tools and more compute resources in the data center, the next generation of issues that need to be solved are a bit different.

The lifecycle management needs to be shared with different teams,” Mueth said. “Maybe you’re a team designing an autonomous driving vehicle. You’ve got to work with your test colleagues, your marketing folks, different people who operate within an engineering lifecycle. And you have to collaborate and share data, and you’ve got to have configuration management, which is a big challenge.”

All of that must then fit into a V diagram, and by extension, into the CAE and EDA realms where the actual design is done.

Adding to the challenges, as well as the confusion, is the growing importance of software applications within a vehicle. This is particularly true for smart phone applications that consumers expect to see within their vehicle.

“An application like Apple CarPlay is different from other components in a vehicle, where others are trying to collaborate as OEMs pull it together,” said Simon Rance, director, product management, data & IP management at Keysight EDA. “The user experience plays a big role in the outcome of that design and application. That’s where there needs to be tighter collaboration between those OEMs that are involved in that system, not just Apple with the CarPlay app and its capabilities and functions. How does it interface with Bluetooth? How does it interface with sensors and sensor data, for example? These are where vendors are looking to take these capabilities, or solutions like CarPlay, to the next level. They’re not just bringing in the primary application of wanting to play Spotify or wanting to see Waze or Maps. They’re also pulling in the head-up displays coming in some of these newer vehicles. They’re displaying that data and user experience, so there definitely has to be a lot more data sharing, for the design aspect as well as the prototyping and testing/validation of it. We’ve all become accustomed to it outside of the vehicle with our phones. We want that to be exactly the same inside the vehicle.”

If it’s too difficult for the driver to see the display or interact with it, they’re not going to be happy, especially if there’s a delay or a lag, like with a map or navigation. “That compounded problem was addressed with Apple CarPlay,” Rance said. “They have the app that takes all of this and gives you that nice overall experience, with visual and sound. Under the hood, it’s interfacing to all these other components that are built by other OEMs, and it comes down to how they communicate and how well they work together. What happens under a certain system environment? How do they react. Which one takes over? Let’s say there’s going to be a collision. How does that car interface quickly go from showing you the next Spotify track to assisted braking, and sensors going off and things like that? We’re seeing more data sharing and tighter collaboration between OEMs around those aspects that require an overall user experience.”

Making connections
Putting this in perspective, design chains are overlapping and shifting, and that still to be sorted out. “OEMs need to coordinate, and they do this partly using standard APIs,” said Frank Schirrmeister, vice president solutions and business development at Arteris. “Even a car that is 2.5 years old has its own map system. But then it can use Apple CarPlay to connect to external sources, which take over most of the function on the display. That needs to be coordinated, tested, with APIs that allow people to program to it.”

It also requires a prioritization of functions. “For instance, when I put my car in reverse, it immediately puts the backup camera on screen and overrides everything,” Schirrmeister said. “Another situation might be that I want my medical device potentially to be connected so that the heart rate is visible, and if the heart rate goes to zero, the car had better stop. This is value chain collaboration, because a Tier 1 supplier might sell this, but they might sell something that is related, or it may have a dependency on other OEMs in the design chain. This just makes the design chain much more fun than it already is. You want to get the economy of scale, because otherwise and OEM might do it themself.”

Collaboration is no stranger to the supply chain, but the amount and complexity of that collaboration is growing. “Collaboration between, for example, Bosch and NXP or ST — that’s been there for a long time,” Imperas’ Lapides said. “But now, in addition to the pure SoC design, we’re starting to see more collaboration in a different sense. We’ve seen it on the software side, say with AUTOSAR (AUTomotive Open System Architecture), but this is different. AUTOSAR was a neutral platform that everybody could use, but there’s still differentiation that’s needed. So how do you collaborate up and down the supply chain on software development and test?”

That still is not entirely clear. “We’ve had these traditional collaborations in automotive where we get OEM cross-synchronization,” Lapides said. “Traditionally that’s been around AUTOSAR, maybe around embedded Linux, and certainly around overall SoC design. But now we’re seeing much more collaboration in the software area, outside of AUTOSAR, outside of the OS. We’re seeing more collaboration getting down to the processor side and what the processor can do. Those things are really interesting — especially in automotive, where AI is going out to the edge with sensors. It’s going in the control and synchronization area for automotive. That means there are now two layers. Plus, as V2X starts happening, you’re going to have data center AI contributing to this, as well. With whatever term we give it, there’s a lot of interesting expansion of collaboration going on in automotive. Automotive is the leading edge of the semiconductor industry right now in so many different ways, because it’s leading edge on the software side, too. The things that are going on here in terms of software collaboration, in terms of system collaboration between OEMs and their supply chain, are going to have echoes in other industries.”

Related Reading
Why Auto Ecosystem Relationships Are Changing
As ecosystem partners bring their unique technologies and expertise to bear, end consumers will be the ones to benefit. But none of this will happen quickly.
Building Better Cars Faster
OEMs want to accelerate the design process, but the availability of tools needed to make that happen will take time to develop.


Esther soria says:

Can you imagine involve vendors were forced to comply with a simply safety standard when they’re attempting to integrate millions of components and all the interaction between the software in there and product! The ISO afety standards call out the documentation of all the interactions of all components and software interaction with all of them! Any application that if it fails that can cause bodily harm and or a Catastrophic event is deemed a safety application! Yet having that standard available we still events that cause deaths example the Tesla issue! . The standard calls out the documentation of any interaction between components that if one component fails how it domino affects the whole system! .Yet we still have Automotive manufacturers that have no clue on how to Comply with that requirement. Can you imagine involve vendors were forced to comply with a simply safety standard when they’re attempting to integrate millions of components and all the interaction between the software in there and product!

Leave a Reply

(Note: This name will be displayed publicly)