The automotive industry has been dabbling with the idea of Ethernet-based in-car networking for a while. To have a switched network and a communication based on the ISO OSI (Open Systems Interconnection) layering model has long promised the needed flexibility, scalability, and support of higher data rates.
The breakthrough in the industry though has come with an Ethernet physical layer that allows for the in-car transmission of 100-Mbps Ethernet packets over cost-efficient UTSP (unshielded twisted single pair) cables. For future applications, the IEEE 802.3 working group has decided to investigate the development of an automotive-suitable Gbit Ethernet and, therefore, placed a final stepping stone for the success of Ethernet-based communication in automotive.
Software and electronics are the innovation drivers in the automotive industry. A high-end car can easily contain more than a hundred ECUs (electronic control units) and the amount of flashable software in BMWs broke the GByte threshold back in 2008. Considering that ever more traditionally mechanical functions in cars are supported by, if not replaced with, electronics, the trend to more software and electronics will continue. A functional and reliable in-car network is therefore a must.
The car industry has been aware of this and generated networking solutions for their actual needs. Among a multitude of solutions deployed in the past the industry has lately been converging on technologies like LIN (local interconnect network), CAN (controller area network), FlexRay, and MOST (media-oriented system transport). Each of these technologies is ideal for a specific use case, but not only is there no migration path from one to the next, each technology is based on a fundamentally different communication principle. In addition, none has been designed to follow the ISO OSI layer model nor allow for a switched network. This is different for Ethernet-based communication.
Layering and switching
As is visualized in Figure 1, Ethernet-based communication follows the ISO OSI layering model and thus allows for reuse and exchangeability on the different protocol levels. Thus it is possible to use 100-base TX Ethernet, UTSP Ethernet (also known as OPEN Alliance BroadR Reach) and APIX 2 connections in one communication for the physical layer, while maintaining the same MAC protocol, or to include a WLAN (Wi-fi local area network) link, and keeping the same protocols from the IP (Internet protocol) layer up.
The use case is not limited, so it does not matter whether the application is diagnostics, driver assistance, infotainment, or something else. This means that Ethernet-based communication is flexible in terms of applications, speed grades, and in terms of requirements that are brought into the car from the outside world.
The traditional in-car networking technologies provide a communications bus. As is depicted in the left and middle of Figure 2, this means that all attached units share the available bandwidth and that adding new units thus affects all existing ones. Modern Ethernet-based networks function via switches (see right side of Figure 2). This means that the available bandwidth is not necessarily shared, especially as the topology is not predefined by the technology but can be chosen to suit the specific situation best.
Additional flexibility is given because—owing to the layering—different links can have different speed grades (meaning scalability) and adding or changing units does not automatically affect the whole network, but primarily the unit(s) they directly connect to. This means that Ethernet-based communication supports fundamental concepts for a future-proof in-car network.
Ecosystem for Ethernet-based communication
Next to being a conceptually suitable solution, Ethernet-based communication in automotive has to have a functioning ecosystem. This means that there needs to be an incentive to use the technology, and there have to be multiple vendors as well as a market for enabling and supporting technologies. Additionally the technology has to fulfill the two basic requirements of every successful technology: technical functionality and the possibility of future development (see Figure 3).
Cars are physically challenging environments, in which a multitude of very different applications are used in a limited space. Next to having automotive qualified parts available (ICs, connectors, cables), the networking solution needs to pass emission and immunity requirements. For Ethernet-based communication over UTSP cabling, these requirements can be fulfilled. QoS (Quality of Service) can be realized with the help of the IEEE AVB (audio video bridging) standard(s). Security can be supported by IEEE 802.1Q VLANs (virtual local area networks), IP addressing concepts, and the usual encryption and other protection methods applied in Ethernet/IP networks.
Incentives and multisourcing
There are two main incentives to adopt a new technology: cost reduction and/or the enabling of new, desirable functionality. For UTSP Ethernet-based communication in cars, both are givens. The solution is cost-efficient in comparison to technologies that need shielded cabling (like LVDS, USB, 100base-TX) or optical cabling (like MOST), and it is also a cost-efficient solution for new applications requiring 100-Mbps bandwidth.
The auto industry is actually not quite as averse to a single source monopolist as might be expected. It seems to be easier to influence just one manufacturer into providing very specific functionality than having a whole standardization consortium to submit to these requests. Nevertheless, a lot more confidence is put into markets with multiple suppliers, as it promises a self-enhancing cycle of cutting-edge products at attractive prices, without the risk of one manufacturer dropping out. To have multiple suppliers requires that the used technology is accessible to different vendors.
The only protocol level on which sourcing is at all something to discuss for Ethernet-based communication in automotive is the physical layer, because of the specific immunity and emission requirements. On all other layers, state-of-the-art solutions from multiple vendors are available. In case of the UTSP Ethernet technology used, the specification is available to all interested parties at the OPEN Alliance and there are two vendors (Broadcom and NXP) with respective products on their roadmaps.
Supporting technologies and the future
For Ethernet-based communication in cars, a number of Tier 1s have built up the know-how needed to offer products to the car manufacturers. Likewise, there are tool vendors and test houses that can offer respective solutions. All AUTOSAR versions starting from 4.0 are supporting Ethernet-based communications, and software houses are implementing these. These examples show a healthily developing infrastructure of supporting technologies.
There is no limit to the data rates needed in cars in the future. Already today there are specific applications in cars for which the transmitted data rate exceeds 100 Mbps. These links are generally not integrated in the communication network but represent isolated P2P (peer-to-peer) links. The fact that they exist shows that it is only a question of time until Gbps-capable links need to be fully integrated into the in-car network.
To have a respective technology available in time, the IEEE 802.3 decided to investigate an automotive-suitable Gbps Ethernet physical layer. With this decision, the final element for facilitating Ethernet-based communication in automotive to be successful has been provided.
For further information on automotive Ethernet, visit:
Dr. Kirsten Matheus of BMW wrote this article for Automotive Engineering.