CAN bus, LIN, MOST, FlexRay and automotive Ethernet – what are the sweet spots for in-vehicle communications and how do the different components fit together?
Cars stopped being purely mechanical systems a long time ago. Modern luxury vehicles can have as many as 150 electronic control units (ECUs) that collectively are responsible for much of today’s driving experience. Even base models still contain numerous ECUs – for example, to provide mandatory features such as tire pressure warning, remote calling in the event of an accident, airbag deployment and more.
Modular in design, ECUs provide a raft of safety features such as anti-lock braking and lane departure warning; perform substantial engine management duties; control cabin temperatures and run infotainment systems. And that’s just scratching the surface.
Microprocessors have become such an integral part of the automobile that engineers at Toyota have reportedly joked that a car’s wheels are simply there to stop all of the computers dragging on the ground.
Piping the data between the individual ECUs are various in-car networks.
Introducing the CAN bus
Point-to-point wiring scales poorly as the number of sensors and controllers in a vehicle increases, which has led to the mass adoption of communications protocols such as the Controller Area Network or CAN bus. Replacing multi-wire looms with a CAN bus design can reduce vehicle cabling by 2km, halve the number of connectors and save more than 50 kg in weight.
Developed by Robert BOSCH, the CAN bus made its debut on production vehicles in the early 1990s. The protocol allows for data rates up to 1 Mbps — extended to 5 Mbps through the use of CAN bus Flexible Data-rate (CAN FD) — as well as message prioritization, which help to support real-time systems.
Designed to be lightweight, fast and easy to configure, CAN bus has proven to be a popular digital network not just for cars, but also submarines, military vehicles and even bicycles! Any node can publish/receive messages on the bus and mechanisms in the protocol are provided to guarantee latency time.
Vehicles may have multiple CAN buses, segmenting high-, medium- and low-speed operations such as RPM management, power-locking and interior lighting, with gateway bridges routing data between the various networks.
CAN bus supports communication for on-board diagnostics (OBD-ii), mandated for cars sold in Europe and the US, which has seen the networking protocol become commonplace.
One of the knock-on effects of cars becoming more computerized is that automobile electronics have jumped from representing around 15% of the cost of a vehicle in 1990 to an estimated 35% by 2020. Some analysts predict that the figure could be as high as 50% by 2030, when more advanced configurations such as LiDAR-equipped self-driving cars are expected to become more prevalent.
Networking on a budget
CAN bus is a relatively inexpensive network for car-makers to deploy, but the cost-saving opportunities don’t stop there. The so-called Local Interconnect Network (LIN) pushes down the price-point further still by focusing on applications where the bandwidth and versatility of CAN are not required. Examples here include power window systems and seat controllers. The LIN bus consists of a single master- and up to 15 slave-nodes, providing network speeds of up to 20 kbps.
Founded in the late 1990s, the Local Interconnect Network (LIN) consortium featured five European automakers (BMW, Volkswagen Group, Audi, Volvo Cars, Mercedes-Benz), Mentor Graphics (formerly Volcano Automotive Group) and Freescale (formerly Motorola, now NXP). The first fully implemented version of the new LIN specification was published in November 2002.
Complementing CAN bus, the adoption of LIN in production vehicles represents a broader move towards hierarchical in-car networks. Alongside CAN and LIN — which are too slow to carry video — many of the vehicles on the road today also feature Media Oriented Systems Transport (MOST) – a high-speed multimedia network technology optimized by the automotive industry.
Dubbed an ‘infotainment backbone’, MOST can support up to 64 devices in a ring configuration. The number of multimedia elements in cars has rocketed from a single radio and amplifier to multicomponent high-definition audio-visual systems that integrate a wide range of functions from movie playback to browsing the web and much more besides.
MOST offers 150x the network speed of a conventional CAN bus, plus provision for diagnostics and software update over Ethernet, and consumer connectivity and built-in charging over USB.
However, it’s not just infotainment that’s adding to in-car networks. Drive-by-wire systems — which replace heavy mechanical linkages with lighter-weight electro-mechanical actuators — place high demands on data communications, particularly regarding timing and error correction.
Offering greater synchronization and networking capacity than CAN bus, FlexRay has been tailored to meet the needs of the so-called ‘x-by-wire’ family of applications, namely – throttle-by-wire, brake-by-wire, steer-by-wire, shift-by-wire and park-by-wire.
Cost considerations remain a key factor in the automotive market and it’s likely, for the time being at least, that mixed networking types will persist as automakers seek combinations that offer the best value. Scenarios here include FlexRay for high-end applications, CAN for mainstream powertrain communications and LIN for low-cost body electronics.
There are other options too. New vehicles with state-of-the-art digital cockpit displays and other data hungry features are likely to feature automotive Ethernet (802.3bw) – a more rugged version of the hugely popular standard for general computer networking.
Automotive Ethernet transceivers are designed to withstand taxing driving conditions (vibration, impact, heat, cold, dampness, and the presence of aggressive fluids), while still maintaining low power consumption and system costs.
Big data on the move
For self-driving cars, automotive Ethernet becomes a necessity to manage the massive amounts of sensor data generated by the vehicle (several terabytes per hour of driving) as it monitors its surroundings. Onboard sensor fusion systems digest information from an array of LiDAR (light detection and ranging), radar and multiple camera inputs to support safe navigation.
Open the boot/trunk of an autonomous vehicle today and you’ll find little room for your luggage as the space will be packed with custom computing (CPUs, GPUs, cooling & data-storage), which includes multiple units to provide redundancy in case of critical component failure.