Marvell Debuts New Multi-Gig Ethernet PHY For Next Generation Cars And Trucks

Marvell MARVELL

As a tech analyst, one of my favorite coverage areas is the multi-industry cadre of companies working to advance and enable next-generation autonomous vehicles. For one, although it has companies at the head of the pack, it’s still a relatively new open playing field. You get to see all sorts of cutting-edge technologies from companies across the tech sector coming together to create something new and impactful. It is truly an all-hands-on-deck business right now.

As our vehicles become increasingly autonomous, on the road to self-driving, nothing is more important than guaranteeing the safety of the passengers, other motorists and civilians. The key to safety is the ability for the car to quickly and securely receive and send vast amounts of data inside the car’s in-vehicle network (IVN). To that end, this past week, Marvell announced the arrival of its IEEE 802.3-based multi-gig automotive Ethernet PHY. 

For those unfamiliar with Marvell, it’s the market leader for IVNs. The company currently has over one thousand engineers working on Ethernet IP, over two thousand Ethernet patents in its portfolio, and developed over 30 automotive IVN products in 2020. Founded in 1995 and located in Santa Clara, CA, the company boasts more than 5,000 workers and pulled in an FY21 revenue of $3 billion. 

It’s all about connection

Marvell’s place in the automotive Ethernet market is pretty indisputable. It’s been on the forefront, delivering the first automotive secured switch to the market, the first multi-gig Ethernet PHY (the predecessor of the just-announced PHY) and the first 1000BASE-T1 PHY. Marvell currently has over one thousand engineers working on Ethernet IP and over two thousand Ethernet patents in its portfolio. It released 30 automotive IVN products in 2020 alone.

Marvell MARVELL

The announcement of the new offering, the 88Q4346, represents the second generation of Marvell’s pre-standard, multi-gig automotive Ethernet PHY solution, one of the industry firsts mentioned above upon its 2018 launch. Marvell says the new generation PHY can deliver as much as 10Gbps throughput data transmission for IVNs, making it one of the first suppliers to offer a triple-speed Ethernet PHY with support for 2.5G, 5G and 10G Base-T1. Additionally, customers can leverage TC10 to engage in partial networking, turning off or minimizing the parts that aren’t being used. 

Moving on, the PHY’s integrated 802.1AE MACsec serves to secure the connections in the IVN. One can imagine the high stakes of a security breach of the systems that keep an autonomous vehicle functioning as it maneuvers a four-lane highway. MACsec works to prevent so-called “Layer 2” security threats, including intrusion, man-in-the-middle, and replay attacks.

The high-speed, low-latency data exchange enabled by multi-gig Ethernet PHYs will be crucial in tomorrow’s vehicles. While Ethernet is already integrated into almost all new cars today, it will take multi-gig connectivity to support complex, data-intensive operations.  (advanced driver assistance systems (ADAS), 5G connectivity and various degrees of autonomous driving). Without this bandwidth, a fully autonomous vehicle simply could not happen. 

The 88Q4246 is also notable because it comes when vehicle architectures are transitioning away from domain-based platforms, where the electronic control units (ECUs) of each domain directly connect to the various sensors, cameras, and actuators. The problem with this approach is that connecting all of the end nodes requires significant amounts of cable—cable that requires a large and expensive cable harness.

The 88Q4346, on the other hand, is what we call a zonal-based platform. According to Marvell, it aggregates signals and protocols in each zone of an automobile, on top of a multi-gig Ethernet backbone. This approach ditches the heavy cable of domain-based architecture for a single pair of lightweight cables, which Marvell says will help shrink power consumption and costs for electric cars. 

5G coming to automotive 

As mentioned earlier, one of the next-generation technologies necessitating multi-gig IVN is the big transition to 5G—the other crucial piece of the connectivity puzzle for tomorrow’s vehicles. With its incredibly high throughput and low latency, the new standard will be a game-changer, enabling vehicles to directly communicate with other cars, pedestrians, infrastructure and the cloud for improved safety. Additionally, it will make software updates and over-the-air application downloads nearly instantaneous. It’s going to be a whole new world. The 88Q4346, according to Marvell, offers the high-speed and low latency required to support 5G networks. 

And it’s scalable too…

Autonomous driving technology is still developing and can be described based on a spectrum of self-sufficiency, from Level 2 to Level 5 (five being fully autonomous and two being the least). This means that any solutions integrated into these advanced vehicles need to be able to scale with the technology. NVIDIA shared that its DRIVE Orin platform utilizes Marvell’s new 10Gbps Ethernet PHY to deliver the multi-gig data transfer speeds, load balancing and security needed to scale across Levels 2-5 of self-driving vehicles. For more info on Orin, see my recent write-up on NVIDIA’s partnership with Plus. 

Wrapping up

Autonomous cars are going to change our world in ways we’re only beginning to fathom. Every year we inch a little closer to that reality, as computing, networks and other technology providers continue to improve and leapfrog competitors’ benchmarks.  There’s a lot of different roles to play in driving the industry forward. While not everyone will get a slice of the pie, I believe Marvell has found its niche. The new PHY looks like its got the right mix of Ethernet connectivity, compute power, security and storage to get the autonomous vehicle efforts a little further down the road. And I can’t wait to see how Inphi plays a role as well.

Note: Moor Insights & Strategy writers and editors may have contributed to this article.