It's all connected

18 November 2021

The basics of in-car networking technology

The difference between an FJ Holden or XK Falcon and current cars is enormous. Over the last 60 years, vehicles have steadily become more complex in every area of comparison. Toward the end of the 1980s, conventional wiring proved inadequate for keeping everything connected. At that time, Bosch developed the CAN Bus (Controller Area Network) protocol for IVN (In-Vehicle Networking). Manufacturers first used it in production vehicles in the early 1990s. The need for in-car networking solutions has grown steadily ever since and will continue.

The areas of comparison mentioned above are clearly defined as powertrain, chassis, body, telematics, multimedia, and human machine interface. Each of these domains consists of sensors, switches and actuators, described generically as nodes. In modern automotive systems, many nodes contain logic chips with circuits and/or software designed for the particular function of the node. Some, like wheel speed sensors, gather data while others create data, like a switch (an on/off condition).

Many nodes require data from other nodes to function. An ABS control module, for instance, requires data from wheel speed sensors. Almost all nodes create data that will be required by other nodes both within their own domains but across other domains, too. Complex nodes that alter their function based on data received from other nodes are known as ECUs (Electronic Control Units). An engine management system is a very complex ECU. A temperature sensor is not, although data from such a sensor is vital for the correct functioning of an engine management system.

These days, cars can have 70 or more ECUs, so connecting each of them to all other nodes is impossible. The wiring alone would weigh more than a full compliment of passengers. Apart from weight considerations, copper wiring is one of the most expensive components in a car. So, all of the nodes are connected to data buses. In the fledgling days of IVN this was the CAN Bus and it’s still used in virtually every vehicle. Now, however, CAN and CAN FD (Flexible Data) Bus networks are used alongside a number of other IVN technologies like LIN (Local Interconnect Network), CAN FD, FlexRay, MOST (Media Oriented System Transport) and Ethernet. There are others, but these are the main types.

There are many differences between these networking protocols, but one of the most significant is the speed at which they move data. LIN transfers at 10 – 20 kilobits per second (kb/s), CAN 1 Megabit per second (Mb/s), CAN FD 5 Mb/s, FlexRay 10 Mb/s, Ethernet 100 Mb/s and MOST 25 Mb/s – 150 Mb/s. LIN, CAN, FlexRay and Ethernet all use copper wires to transmit data. CAN uses twisted pair wires in which the electromagnetic fields surrounding one wire cancel those of the other wire. This eliminates electrical ‘noise’. Also, it’s said the twisted form ensures the wires are equal distances from any source of interference. FlexRay and Ethernet also use twisted pair wires. LIN uses just one single wire at battery voltage, so it’s a less expensive system. MOST uses fibre optics, so it’s not susceptible to electrical interference. Not surprisingly, though, it’s the most expensive. 

The data transmitted across an automotive network consists of strings of binary digits, or bits. These are grouped in blocks of fixed length and structure called frames, which differ between IVNs. There are also different frames within a particular IVN technology. LIN, for instance, has different frames described as unconditional, event-triggered, sporadic, diagnostic and user-defined. The last type allows frames to be used in ways that aren’t described in the LIN specification. IVNs allow flexibility in vehicle systems. 

While the data contained in different frames change, certain parts of each frame (within a particular IVN technology) have fixed form and meaning that facilitate the basic operation of the system. For instance, the first bits in a frame will indicate the start of the frame and an identifier that lets the various nodes know if the frame is intended for them. 

The identifier also indicates the importance of a frame. This is necessary because if two nodes on the same network transmit simultaneously the frames can collide with each other, creating a network conflict. If this happens, the frame of greatest importance passes while the other does not. The trailing bits of a frame might contain error-checking bits and others will indicate the end of the frame.

The various IVN systems offer support for the wide range of sophisticated features found in today’s vehicles. Indeed, without IVN  such features would not be possible. Another important advantage provided by IVN is the ability to upgrade or add features to a vehicle model without redesigning the systems involved. IVN systems also have positive effects outside the vehicles to which they’re fitted. For example, reducing air pollution from vehicles has only been possible because of the fine control of engines provided by digital technology. And as digital safety systems like ABS and ESP connected by IVN right the wrongs committed by motorists, the world becomes a safer place.

IVN protocols have been developed for the differing requirements of particular vehicle systems. ABS, ESP, Break-by-Wire, Steer-by-Wire and other safety-critical systems need networks with high data rates and network redundancy built in. The networks for other non-critical systems like power windows, door locks, folding mirrors and the like do not need such high data transfer speeds.

The limited speed and relatively simple structure of LIN technology is well suited to non-critical systems like those mentioned above. Also, the relative inexpensiveness of LIN helps make the addition of such features economically feasible. LIN is designed to be a sub-network to a CAN bus.

A LIN bus is limited to a length of 40 metres and a total of 16 nodes. So, we group nodes in a LIN in clusters defined by their physical proximity to each other. There might be a right-front door cluster containing window switches, a lock actuator, a folding door mirror actuator, a courtesy light, and a door ajar sensor. Each door will have its own cluster of such nodes. These are known as slave nodes and they are all connected to a master node. The master node controls the transmission of data on a LIN bus. Slave nodes cannot transmit data until the master node allows it, which eliminates the bus conflicts described previously.

The master node is the connection point to the CAN network that links the cluster to the rest of the vehicle. It contains a schedule that determines the timing of activity among the nodes. This is why LIN technology is described as a deterministic system. That various LIN clusters are connected across a central CAN bus allows things like driver control of the other windows and door locks in a vehicle.

There are a number of CAN protocols defined as ISO standards. Both of the twisted wires in a CAN bus carry the same data but in the opposite way. One wire is called CAN-high the other is CAN-low. In a CAN system (and in many other networking systems) a 0 is described as dominant while a 1 is called recessive and this is an essential feature of CAN. The fact these wires mirror data means any interference is less likely to affect the system, which helps maintain data integrity.

All controllers read all frames transmitted on a CAN bus and each controller determines if the data contained in a frame is relevant by examining the ID section of the frame. Operating in this way means more than one controller can use the data transmitted. In other network systems, messages can be destroyed during a conflict between frames. CAN has non-destructive message arbitration so data is preserved. CAN also has a number of ways to detect errors.

CAN has some characteristics that make it less than ideal for safety critical applications. Standard CAN has good data integrity, but this comes at the expense of speed. CAN FD is much faster, but this is at the expense of data integrity, which is not good for safety critical applications. However CAN FD is well suited to data logging because such systems have to transfer quite a bit of data that’s constantly updated. In doing so, it may not matter much if the odd frame is dropped, depending on the application. CAN XL is a newer version of this technology that can reach speeds of up to 10 Mb/s. This is to keep CAN competitive with other IVN technologies like FlexRay.

FlexRay is very similar to CAN in many fundamental ways, but has a much higher rate than standard CAN and even CAN FD. Now, however with CAN XL the extra speed, or bandwidth, advantage of FlexRay is considerably eroded. Frame collisions are avoided by restricting node transmissions to strict time slots. This is known as a Time Triggered Protocol (TTP) as opposed to the event-driven paradigm of CAN. The maximum length of a FlexRay bus is 24 metres with 22 nodes.

Like CAN, FlexRay is well suited to use as a so-called backbone network to which many other sub-networks are connected. It’s more difficult to implement than CAN and changes to the programming of one node mean all other nodes have to be reprogrammed. This extra complexity was worthwhile because of the increased data transfer rates over CAN FD. FlexRay was created to meet the needs of high-speed control functions in powertrain and X-by-wire systems. FlexRay does have built-in redundancy, but it will be interesting to see what effect the increased capacity of CAN XL will have on it.

Modern vehicles are increasingly demanding. With their high bandwidth, MOST networks are built to cater to that demand. As the acronym suggests, MOST fibre optic networks are designed to handle media and entertainment, like video. This requires enormous bandwidth. Yet demand for MOST is set to be exceeded by Automotive Ethernet. There are several reasons for this.

MOST is a propriety technology that was available from just one source. The automotive industry doesn’t like that because it’s been used to open standards actually developed by car manufacturers themselves. MOST technology has now been licenced to other companies, so it’s available from a number of sources. Regardless, it still suffers from some other difficulties. In networking, topology refers to how nodes are connected on a bus. MOST has a ring topology which means if any node fails, the entire network fails. Although there’s a lot of MOST in current cars, predictions are that it will slide out of favor in the near future.

What will replace it?

Automotive Ethernet is the answer. This is adapted for automotive use from standard Ethernet. Automotive Ethernet has had data transfer rates of 100 Mb/s and 1000 Mb/s (1 Gb/s). While MOST can reach 150 Mb/s the newest versions of Automotive Ethernet will be available in 2.5 Gb/s, 5 Gb/s and 10 Gb/s. That’s right, Gigabits. It’s also a full-duplex technology, meaning nodes can transmit at full speed in both directions and do so simultaneously. This, and other aspects of the system, can mean data rates can be higher again than the quoted base figures.

Apart from the ability to transfer data rapidly, there is no possibility of collision with Automotive Ethernet. That pretty much makes it game over. Connection to Automotive Ethernet is more complex and requires an expensive and complex component called a switch, but hey, 10 Gb/s! There’s also a version of Automotive Ethernet with a reduced capacity of 10 Mb/s. It’s designed to compete with CAN XL. Why would anyone want Automotive Ethernet that’s so slow? Well, it’s been specifically designed to compete with CAN XL. It has a bus structure similar to conventional IVNs and it’s much cheaper than Automotive Ethernet. The notion with Automotive Ethernet is that it should take over all domains.

A new data transfer technology is called SerDes (Serialiser/Deserialiser). It’s designed for connecting cameras, so it’s well suited to critical camera-based technologies starting to appear in cars. It runs at 12 Gb/s, but only in one direction. Still, that’s not bad considering cameras are all about sending images. It’s a point-to-point system.

Words: Paul Tuzson. As featured in Australian Automotive December 2021.