Connecting the connected car
The electronics-based content of vehicles has grown exponentially over the past ten years. For example, advanced driver assistance systems (ADAS) such as blind-spot detection, adaptive cruise control, and automatic braking systems require significant compute resources and reliable communication to associated vehicle electronic control units (ECUs). In addition, infotainment systems today offer comprehensive navigation features, entertainment capabilities, and seamless integration to smartphone digital assistants. Moving forward, automotive manufacturers are also busy grappling with vehicle-to-vehicle (V2V) and vehicle-infrastructure (V2X) connectivity essential for semi- and fully autonomous vehicles.
The modern vehicle has been compared to a data centre on wheels, highlighting the need for reliable and robust data networking throughout the vehicle. However, unlike an enterprise data centre, a vehicle's networking environment is exposed to the hazards of electrostatic discharge (ESD), electromagnetic interference (EMI), and many mechanical challenges such as shock, ingress, and vibration.
This blog investigates the growing importance of in-vehicle networking, the market demand for higher bandwidth, and some network protocols and architectures addressing automotive engineers' concerns.
In-vehicle networking: How did it become so complex?
We have all experienced the dazzling array of features and driver aids installed in modern vehicles. Some of these, such as blind-spot detection and emergency braking, make our journeys safer. In addition, navigation and real-time traffic information enhance our driving experience by removing driver stress and fatigue associated with traffic jams and other unplanned delays.
Behind the scenes, all these comprehensive features rely on a host of electronic control units (ECUs), most of which are networked together. Also, ADAS functions require sensors such as video cameras, radar modules, and wireless connectivity to operate, and these all need to be connected to their respective ECUs. Thus, automotive in-vehicle networks are a vital aspect of any vehicle today, for which link robustness and reliability are paramount. As new features are introduced, more dependency and scope of the networks deployed in the vehicle grow, presenting automotive system architects with design challenges. For example, camera functions need to stream high data rate video in real-time, potentially introducing latency concerns. In addition, interconnecting ADAS functions, such as emergency braking to the braking hydraulics, require a high quality of service.
Then there are the challenges associated with the operating environment of any vehicle. Extremes of temperature and humidity are commonplace, as is electrostatic discharge, something sensitive electronic equipment needs to be protected against. Susceptibility to electromagnetic inference (EMI) also requires compliance against international standards.
As feature-rich ADAS, infotainment and comfort functions evolve, automotive networking standards need to advance to meet the stringent demands and expectations of the automotive electrical and electronic engineering teams.
Automotive networking protocols
Several popular networking protocols are already well-established in the automotive domain, and they continue to evolve to keep pace with the increasing demands of the industry. For example, CAN (controller area network) was initially conceived for industrial automation applications but quickly found widespread adoption as the need for automotive networking surged. LIN (local interconnect network) devices often work through CAN networks using a CAN-LIN master node for simple low-data rate tasks, such as actuator control and switch monitoring.
The increase in high-speed video processing and real-time applications has driven the need for higher in-vehicle network bandwidth. So, it is perhaps no surprise that Ethernet became a networking candidate. A mature, reliable, and extensively deployed network protocol for enterprise IT applications offers many features automotive system architects require.
To accommodate the automotive domain, adaptations and extensions to both CAN and Ethernet have resulted in optimised protocols that fully comply and are maintained by international standard committees.
When selecting the network protocol, engineers initially need to consider several vital criteria; bandwidth capability, latency, and whether a symmetrical or asymmetrical communication is required. Then there are factors such as how immune the networking method is to EMI and ESD, the physical cabling constraints such as cable size and weight, and the ease of interfacing to other ECUs and functions.
Architectural decisions
Today, a typical vehicle is estimated to have 80 ECUs installed, upwards of 100 sensors, and upwards of 50 kg of wiring harnesses. Moreover, the trend towards autonomous vehicles will only increase the need for more ECUs, sensors, and cables. Aware of this impending complexity, the automotive industry took early steps to ease the networking and computational challenges by consolidating multiple ECUs into a single ECU based around a zonal approach. A zonal electrical/electronic (E/E) architecture expands on the concept of a domain controller approach implemented using a CAN network by adding a high-speed network backbone to interconnect every vehicle function. See Figure 1.
Figure 1 - Implementing a zonal architecture with functional domains and a high-speed network backbone (source Avnet)
For example, the driver's door can be identified as a zone. There are switches for door mirrors, window controls and associated motors, interior theme lighting, blind-spot vision sensors, mirror lights, and door-open warning lights. Zone-based compute resources and access to the high-speed backbone are optimised according to each function by the zonal gateway. Within this zone, door mirror controls are used infrequently, but the blind-spot detection function generates warning information in real-time.
Even though Ethernet is gaining significant adoption for automotive applications, there is a widely held view in the industry that CAN and Ethernet both have their merits and will likely co-exist for the foreseeable future.
The CAN networking protocol is widely used for automotive applications and has evolved considerably over the years. The CAN-FD (flexible data rate) specification is perhaps the most deployed, with reliable data rates of typically 2 Mb/s. Although technically capable of using up to 8 Mb/s, network link reliability suffered in real life, prompting several vendors to consider signal improvement (CAN SIC) technologies. With the signal improvement capabilities, CAN FD is now a reality up to 5 Mb/s. Further, CAN specification advancements are advancing, with CAN XL promising data rates up to 10 Mb/s.
Another CAN enhancement is the partial networking (PN) specification. Aiming to significantly reduce the number of ECUs operating at full power even though they are not always required, CAN PN allows ECNs to be woken up from a low-power sleep mode, saving power and vehicle emissions.
Automotive Ethernet slightly differs from standard Ethernet in that it uses a lighter and thinner single unshielded twisted pair (UTP) of conductors for two-way communication. Standard Ethernet uses a thicker and heavier cable of four UTPs with dedicated transmit and receive pairs. Identified as Ethernet standard IEEE 802.3bp 100Base-T1 and conforming to automotive applications' stringent EMI and EMC requirements, it supports data rates up to 100 Mb/s. In addition, the Ethernet 10Base-T1S specification also supports 10 Mb/s bandwidth, which is the same as CAN XL. Both the 100Base-T1 and 10Base-T1S specifications have a maximum cable drop distance of 15 meters.
The road ahead
The zonal approach to in-vehicle networking is a pragmatic step to advancing the development of our vehicles. With a combination of CAN and Ethernet's complementary network capabilities, the automotive industry looks set to focus on getting the best out of each in the coming years. In addition, the zonal architecture, and the use of single UTP cables assist in keeping cable harness weights and dimensions to a minimum, further contributing to overall vehicle weight reductions. Lowering a vehicle's weight reduces the amount of energy required to move it, but also, for electric vehicles, helps increase its range.
To learn more about the most appropriate, secure and cost-effective in-vehicle networking technologies to fulfil the needs of tomorrow’s cars, join Avnet Silica and our leading supplier partners, Marvell Technology, Microchip Technology Inc., NXP Semiconductors, onsemi and STMicroelectronics for our webinar series: In-Vehicle Networking and Connectivity.
Follow Avnet Silica on LinkedIn
About Author
Michael Pochmann
Michael Pochmann is Segment Manager Automotive EMEA at Avnet Silica. With over 20 years’ experience,...
silica content library/articles/connecting-the-connected-car
Connecting the connected car | Avnet Silica
