10G automotive Ethernet is intended to address a growing challenge in vehicle design: moving increasing volumes of data between sensors, controllers and computing resources without relying on a separate point-to-point connection for every function. As automakers consider zonal architectures and more centralized processing, the network must support higher-speed links alongside a range of lower-speed connections.
The case for higher bandwidth is not just about faster communication. It is also about how vehicle electronics are organized, how sensor data reaches compute resources, and how functions might be distributed across electronic control units (ECUs). A practical design therefore needs to consider the entire network—not only its fastest links.
How automotive networking has evolved
Vehicle networks have developed through technologies designed for different communication needs. Controller Area Network (CAN), introduced in the early 1990s, provided a serial data bus for control applications. FlexRay and MOST later served other requirements, while low-voltage differential signaling (LVDS) and serializer/deserializer (SerDes) links enabled point-to-point connections for cameras and displays.
Ethernet entered automotive networking with a progression of speeds that includes 10 Mb/s, 100 Mb/s, 1 Gb/s, and multi-gigabit options. Automotive Ethernet PHY standards listed in the 2019 material included 100BASE-T1 and 1000BASE-T1, as well as work on 10BASE-T1S and 2.5, 5 and 10 Gb/s BASE-T1. Each speed has a role: a vehicle network may need low-rate links for some endpoints and higher-rate paths for demanding data flows.
Why zonal architectures need 10G automotive Ethernet
Traditional vehicle architectures often organize electronics around individual functions or domains. A zonal approach instead groups connections according to their physical location in the vehicle. Zone controllers can connect local devices to Ethernet switches, which then carry traffic toward central computing resources. This changes the communication pattern: information may need to travel across the vehicle between zones and shared processors.
Consolidating processing resources increases the importance of bandwidth between zone controllers and central computers. The whitepaper identifies links above 10 Gb/s as a requirement for this direction of travel, while noting that zonal vehicles would still use a mix of Ethernet speeds. Higher-speed links would serve demanding paths; 1000BASE-T1, 100BASE-T1 and 10BASE-T1S would remain relevant elsewhere.
For network architects, the implication is that bandwidth planning must account for both capacity and topology. Zonal designs may involve more Ethernet links overall, and traffic from multiple systems can converge on shared switches and compute nodes. Engineers must understand which connections carry the heaviest loads and where redundant paths or systems are required.
Autonomous driving adds data-intensive workloads
Advanced driver-assistance and autonomous-driving systems add pressure to the in-vehicle network. The source describes cameras transmitting uncompressed data and the aggregation of information from additional sensors such as lidar and radar. These workloads can require high-bandwidth connections between sensors, processing units and other parts of the vehicle.
Redundant processing units are also identified as part of the autonomous-driving interconnect picture. Connecting those units requires high-capacity links, while a vehicle data recorder—or “black box”—may need substantial bandwidth and storage capacity to capture raw sensor data. These requirements are distinct, but they all make the design of the data path important: moving information, processing it and retaining it are connected system-level considerations.
The paper also raises live software migration as a potential use case. Transferring processes between compute nodes or ECUs would require low-latency, high-speed connections. This points to a broader role for the network: supporting not only sensor streams, but also the flexible distribution of software and processing tasks.
Standards development and engineering considerations
In 2019, IEEE 802.3 activity was examining the development of electrical PHYs for automotive Ethernet above 10 Gb/s. A study group’s role was to develop project documentation, including a Project Authorization Request (PAR), Criteria for Standards Development (CSD) responses and high-level objectives. The group was not responsible for making detailed technical decisions about the eventual specification.
The materials describe work on project objectives and link-segment feasibility, including asymmetric use cases. They also report the adoption of a 25 Gb/s PHY objective in September 2019, with two inlines and an 11 m distance specified for that objective. These details reflect the status and scope discussed at that time; they should not be read as a statement about subsequent standards or current vehicle deployments.
For engineering teams evaluating network architectures, the wider lesson is to align requirements with the intended application. Relevant questions include:
- Which sensor and computing paths need multi-gigabit bandwidth?
- How will traffic be distributed across zones, switches and central processors?
- Where are redundancy and low latency required?
- Which links can use lower-speed automotive Ethernet PHYs?
- How will the network support data recording and potential software movement between ECUs?
Planning for the next vehicle network
The move toward zonal and centralized vehicle architectures makes Ethernet capacity a system-design issue, not simply a PHY selection. 10G automotive Ethernet can provide high-bandwidth links for consolidated processing, sensor aggregation and communication between computing units, while lower-speed connections serve other parts of the network.
Automotive engineers, integrators and decision-makers should therefore assess expected data flows, network topology, redundancy and latency together. A mix of link speeds—and a clear understanding of how traffic moves between zones and compute resources—will be central to designing vehicle networks capable of supporting increasingly data-intensive functions.