The automotive industry is adopting self-driving cars as the next major evolutionary step in vehicular technology. However, improvements need to be made to data buses inside the car to support the additional data needed for a car to operate autonomously. Ethernet and the Mobile Industry Processor Interface (MIPI) are replacing previous network technologies, allowing for the next generation of autonomous vehicles through faster data rates and lower latency.
For the better part of the last century, automotive makers have been creating new and innovative technologies with the goal of designing fully autonomous vehicles. The concept of a highly autonomous and self-driven vehicle can be dated back to at least the 1950’s with the idea being glorified throughout pop culture in the following decades, such as comic books, TV, and film. However, only in the last 20 years have the technological improvements in the car advanced enough to support such applications. Vehicle to Vehicle (V2V) and Vehicle to Infrastructure (V2I) wireless communication protocols have revolutionized the future of autonomous vehicles, but a fast, reliable, and robust wired backbone is still needed to meet the bandwidth needs of systems and sensors within the car.
For decades each Automotive OEM used proprietary technologies to enable new features in their vehicles, but eventually some technology standards (CAN, Flexray) were developed for the specific needs of the automotive industry and adopted by OEMs. However, the bandwidth capabilities of these networks quickly became inadequate for the growing number of sensors needed to make the self-driving car a reality.
About 10 years ago, the automotive industry looked to other network technologies and architectures to overcome these bandwidth limitations and recognized Ethernet as a potentially viable option. Additionally, MIPI has recently gained the attention of automotive OEMs as well because the solutions are mature, relatively simple to use, and help reduce the number of wires needed to connect components.
Brief History of Automotive Ethernet
Ethernet was first considered for in-vehicle applications around 2008, but it never quite took off during that time. For starts, 100-Mbps twisted-pair Ethernet (100BASE-TX - Fast Ethernet) was installed and used mostly for diagnostic applications and servicing electronic control units (ECUs). It was not used as a data bus during vehicle operation because of poor EMI performance. Additionally, ECU data polling and firmware updates required the transfer of large sets of data, leading to long vehicle service times.
Lastly, Ethernet was not designed to be a lossless network, and Automotive applications require guaranteed bandwidth and latency for safety reasons. However, it was during this time that the Automotive industry would start to recognize the key benefits of Ethernet: reduced cost, proven interoperability, and network ubiquity.
“Automotive Ethernet” - Improvements to Previous Ethernet PHYs
After recognizing the benefits of Ethernet over other comparable technologies, the automotive industry kept a watchful eye on developments within IEEE 802. A few years later, the IEEE 802.1 Working Group started developing Time Sensitive Network (TSN) specifications that would remedy the guaranteed bandwidth and latency concerns. Since the TSN specifications were creating a way to solve the latency issue, the need for an automotive-specific Ethernet PHY could be discussed.
In 2012, ambassadors of the automotive industry attended the IEEE 802.3 Working Group meeting to build consensus to create an Ethernet PHY project specific to the automotive industry’s needs. In the six years since that decisive meeting, there have been three IEEE 802.3 projects specific to the automotive market including 100 Mbps and 1000 Mbps PHY definitions. Both PHYs are specified to operate over 15 meters of a single unshielded twisted-pair cable, half as much cabling as 100BASE-TX and one quarter the cabling needed for 1000BASE-T, leading to weight and cost reductions when installing the vehicle cable harness.
They also have improved EMC requirements to accommodate the harsh RF environment of a motor vehicle. These characteristics combined with the IEEE 802.1 TSN protocols, create a cost competitive solution that is reliable, high bandwidth, and multi-vendor interoperable, allowing for greater network flexibility and scaling, while lowering complexity and cost.
Advantages of MIPI and Its Automotive Use Case
While Ethernet is getting a lot of attention from the automotive industry, it’s not the only technology that’s been recently integrated in the car. Over the last 15 years, the MIPI Alliance has been creating standards specifically for mobile applications, particularly the data interfaces between processors, memory, and displays. The protocols within the MIPI standards are designed for high bandwidth (up to 12 Gbaud) over very short distances (approximately one meter).
While intended for mobile applications, other industries have found an interest in the unique feature set of MIPI PHYs, including the automotive space. Driver assisted vehicle features typically require several sensors located all around the car and these features use video from cameras to determine lane position, foreign object location, etc.
Whatever the application, it is vital that the video feed be the highest resolution possible to accurately detect other vehicles or pedestrians. High resolution requires high bandwidth, especially when dealing with uncompressed video feeds. Since MIPI was designed for high bandwidth video use cases it’s easy to see why automotive OEMs are adopting it into their network architectures.
In 2017, the MIPI Alliance chartered the formation of the Automotive Birds of a Feather Group to investigate future MIPI applications in vehicular networks. Since then the group has become an official MIPI Alliance working group (WG). The Automotive WG is quite young so it’s unclear exactly what deliverables they will create, but it is expected to continue the pursuit of high bandwidth, asymmetric networks with a focus on video or display applications.
How Automotive Ethernet and MIPI are Being Implemented
Ethernet and MIPI are both attractive solutions with different sets of benefits, and therefore are being considered for different use cases within the car’s network architecture. Many of the sensors needed for self-driving applications include some form of high-resolution video recording and analysis. Some cameras have native Ethernet integration, but with MIPI’s focus on short distance video delivery it’s a potential choice for sensors with proximally located processors.
A gateway electronic control unit (ECU) can then be used to connect that processed information to the primary backbone network. Automotive Ethernet is then used as that robust backbone for the entire network within the car as IEEE 802.1 switching specifications make it simple to implement switches within the car to scale the network as necessary without significantly increasing the amount of cabling needed.
Vehicular networks have evolved in the last 10 years with automotive Ethernet and MIPI standing out as the two emerging technologies providing the high bandwidth/low latency advancements needed to implement self-driving features. By replacing the assortment of various networks and technologies with a succinct and overarching Ethernet backbone, Automotive network architects have improved design flexibility while reducing system costs.
MIPI offers a high bandwidth, asymmetric connection to vehicle sensors allowing for more uncompressed high-resolution video streams in driver assist applications. Both technologies offer unique features, cost competitive architectures, and interoperable specifications to support the growing Automotive community.
About the author
Curtis Donahue is the Senior Manager of Ethernet Technologies and manages the Automotive Ethernet Test Group at the University of New Hampshire InterOperability Laboratory (UNH-IOL). His main focus has been the development of test setups for physical layer conformance testing, and their respective test procedures, for High Speed Ethernet and Automotive Ethernet applications. Curtis holds a Bachelor of Science in Electrical Engineering from the University of New Hampshire (UNH), Durham and is currently pursuing his Masters in Electrical Engineering at UNH.