Automotive Ethernet: Solving complex wiring harnesses and supporting more data

If you've ever had the chance to look at the mess that is a car wiring harness, you know that these things can be massive; hundreds of wires all tied together with cable ties that you hopefully never have to cut to avoid nicking the sensitive shielding inside.

These bundles take up valuable space within the car's chassis, adding unnecessary weight and creating more potential points of failure. But thanks to some new high-bandwidth cabling developed using enterprise-standard network protocols, this tangle of cables can be replaced with a single, thin wire that delivers more bandwidth than all the other wires combined.

This is part of the promise of automotive Ethernet, but its benefits go far beyond that.

Ethernet and the software-defined car era

The era of software-defined vehicles (SDVs) is upon us. With more code running more parts of more cars, the potential to fundamentally change the way a car behaves or even drives itself using code downloaded from the cloud is growing.

But more software running in the car also means more data. Lots and lots of data. All those high-definition sensors that help modern cars understand the world around them generate a flood of information every second. All of that data must be transmitted through the car in endless streams that quickly become rivers.

If that's not enough, you can also use a modern infotainment system that can activate 64 channels of Dolby Atmos audio or high-definition video streaming to multiple rear-seat infotainment displays.

In other words, there's a lot going on under the hood of your next car, all of which requires more bandwidth. Ethernet is the solution.

What is Ethernet?

If you're familiar with the term Ethernet, you probably know it as the cable you use to connect your internet provider's modem to your wireless router (the most common is a Cat 6 Ethernet cable). Or, if you're a serious gamer, you can use it to connect your PC or console directly to your network for the lowest latency possible.

This is common stuff, but it’s not exactly what we’re talking about in-car Ethernet. Formally speaking, Ethernet isn’t actually a single type of cable at all; it’s a networking standard that defines how multiple devices can interact securely and reliably.

Ethernet, by the way, is capitalized because it refers to a technology that Xerox trademarked in the 1970s. Xerox released the trademark so that the technology would eventually become a cross-industry standard, but it remains a proprietary name.

What we usually call an "Ethernet cable". The cables in your home contain multiple pairs of smaller cables twisted together, which is why they are often called "twisted pair". They usually have an RJ45 connector on each end, which looks like an oversized phone jack. The official name for this type of cable is 100BASE-T or 1000BASE-T, which refers to the data transfer speed of 100 or 1,000 megabits (gigabits) per second. If you are using Gigabit Ethernet, it should provide enough bandwidth to transfer a 4K movie in about four minutes.

Even that is too slow when it comes to modern in-car data. Faster speeds require an entirely different kind of cable. "We're talking about a technology called T1," says Amir Bar-Niv. He's vice president of marketing at Marvell Semiconductor, a major international supplier of semiconductors and networking equipment headquartered in California. "Everything we do, 100 megabit, gigabit, multigigabit and 25 gig, is designed for T1."

When Bar-Niv says “25gig,” he means 25G, enough to stream the same 4K movie in eight seconds, and enough bandwidth to make all the feature films you could ever want.

The type of cable here is not only optimized for speed, but also for automotive use. It's thinner, lighter, and cheaper than the mess of cables behind your desk - a pair of wires twisted into a single cable, or even a fiber-optic line, using a new standard connector called T1, providing all the bandwidth needed for next-generation SDV applications.

It wasn't always this way. In fact, in-car Ethernet dates back 15 years. "BMW first introduced the technology in 2008 with the F01 7 Series," says Ian Richards, vice president of automotive at TechInsights, an international company focused on semiconductor analysis. "It was originally used for diagnostic ports and also for 20 Mbps front/rear seat links on infotainment systems." The first applications used 100BASE-T cables, like you'd use at home.

Of course, these evolving cable standards are a critical part of the system, but to actually build a network, you need a place to plug them in. On one end, you have sensors and systems generating data and getting work done. On the other end are the various processors on the car that crunch the data. Keeping everything moving in the right direction requires a series of network switching devices, i.e. switches and routers.

Optimally placing these switches within the car minimizes the amount of wiring required while still enabling all of the devices in the car to communicate with each other. This configuration, often referred to as the network architecture or topology, is where things start to get interesting.

Exit domain control

Traditionally, as a car gains features and options during its life cycle, the bundle of wires collectively known as the wiring harness grows, with more harnesses created with each new feature.

Bar-Neve says this is a legacy of the old way things were connected, which he calls domain architecture.

Think of each system of the car, such as lighting or infotainment, as a separate domain. Each domain is directly controlled by one or more (possibly dozens) of independent microprocessing units, all of which are dispersed throughout the car.

If a car gets some new feature in its mid-cycle update, such as heated washer nozzles or a blind-spot monitoring system, more wires will be added to the bundle to control that feature.

“In a car where there’s almost no sharing between domains at all, each domain needs to have all the cables wrapped around the car,” Bar-Neve said. “The result is, you know, lots of cables, high cost, cable highways.”

With Ethernet, he said, the car can be transformed into a collection of interconnected zones, each containing sensors or devices , which he called agents: “Each zone now needs to have its own Ethernet switch connected to all the agents of that domain in that zone, and then all the traffic is aggregated over a thin Ethernet cable that really builds the backbone of the car.”

That way, multiple in-car systems all communicate over the same network. By sharing, the design of the car can be radically simplified. Instead of dozens of custom chiplets scattered throughout, fewer, more powerful CPUs handle all the digital heavy lifting for the entire car.

Besides simplification, this opens the door to the evolution of car behavior that is the hallmark of SDVs. Bar-Niv uses automotive cameras as an example: “Cameras today use point-to-point connections,” he says, transmitting data directly from the imaging sensor to the system that uses it.

Because of this, the rearview camera only works as a rearview camera as other systems in the car cannot access it.

For example, you can’t share cameras between infotainment and ADAS,” Bar-Neve said.

However, once all of these cameras are communicating via Ethernet (all on the same network), software can be written to use these cameras in interesting new ways, such as turning a rear-view camera into a dash cam that streams footage over the internet. This functionality can then be downloaded via an over-the-air (OTA) update.

This flexibility is at the heart of SDVs, and according to Bar-Niv, a recent survey of automakers predicts that 80% of new vehicles will move to this architecture by 2026, becoming a $5.6 billion industry.

In other words: This isn't just a trend for high-end luxury cars.

And, as this trend continues, the need for more bandwidth will likewise increase. “100 megabits is more than enough for domain infrastructure. Anything below 2.5 GB is unacceptable today,” Bar-Neve said.

Automotive-grade reliability

Much of the core technology here is based on common enterprise network hardware - the same stuff used by banks or businesses. However, bringing it to automotive applications requires a higher level of reliability. If a bank's network hardware fails, the bank may suffer losses until it is fixed. This is not good, but when the car's active safety systems run over the network, any failure could result in loss of life in the worst case scenario.

A key part of preventing such failures is improving the final quality of the components installed in the car. “We measure reliability in DPPM,” Bar-Niv said, which stands for defective parts per million. For consumer networking equipment, DPPM “is in the range of a few hundred to 1,000,” Bar-Niv said.

In the automotive sector, the goal is less than 1 DPPM, or one defective part per million produced.

But even if these parts are built to a higher standard, failures can still occur. That’s where redundancy comes in, like with network switches that use so-called dual-core lockstep processors. Here, each calculation is handled by two processors. The output of each is compared, and if there is a discrepancy, an algorithm can ensure that the entire system continues to operate based on the output of the functional processor.

Improve car safety

If every component in a car, from the high beams to the head-up display, were communicating on the same network, it sounds like a bit of a security risk. It’s true. Previously, if a hacker gained access to a given system on a car (one of the domains we discussed above), it would be unlikely for the hacker to take over other aspects of the car’s systems. Because of the amount of wiring these inefficiencies required, each system domain actually ran on a separate, disconnected network.

In cars optimized to use automotive Ethernet, where everything is consolidated onto a single network, the potential for system-wide access via a single compromised component is greater. Thankfully, the protection is also greater.

"What we need to protect against is what's called a man-in-the-middle attack," Bar-Neve said.

This refers to an attack where a hacker effectively pretends to be an authorized system within the car, connecting with other systems and using them to access data within the car.

One way to thwart such attacks is to use hardware security modules (HSMs). These generate digital keys inside the car, encrypting and digitally signing all data passing through the vehicle. This technology is usually only used in high-end enterprise digital networks, and its presence in the car ensures a similar degree of mobile security.

Costs and Benefits

Between the fancy cabling, switches, and HSMs, not to mention all the software that runs it all, running a car on Ethernet instead of the more basic but redundant domain model that existed before, might sound like an expensive proposition.

“Yes and no,” said TechInsight’s Riches. “Yes, Ethernet is more expensive than the slower existing CAN links in vehicles. However, it is a fundamental building block for the new vehicle architectures needed to enable on-demand features as well as the high-speed communications required for autonomous driving.”

In other words, there's no way around it. Riches said Ethernet is "the only game in town" if you want to do anything that requires a lot of data. But the cost of the network hardware to connect them will be minuscule compared to the cost of other sensors, such as the 3D laser scanners installed on the new Lotus Eletre.

In fact, in some cases, switching from a domain to a zone architecture, and removing all redundant components in the process, can actually save money. “There is also a long-term view that as compute performance is centralized, savings can be achieved at the vehicle level by removing many of the existing discrete ECUs today,” Riches said.

According to Bosch, a Tier 1 supplier to many OEMs, switching to this architecture could reduce the cost of these in-vehicle networks by 10%. Future cars would also require 20% fewer in-vehicle components, such as microprocessors, according to Bosch, with persistent shortages of such components still plaguing the industry.

Therefore, the move to in-vehicle Ethernet could mean that cars will not only be smarter, cheaper, and lighter, but also less difficult to find in the future.