Know your network

There is an ever increasing number of electronic control units (ECUs) being packed into modern automobiles, assigned to handle various infotainment, safety, comfort and powertrain functions. In some cases the figure now surpasses 100-120 units. The transfer of data to and from these ECUs is of course critical. Due to this, overall connectivity demands are destined to rise still further, leading to greater reliance on in-vehicle networking implementation.

Currently there are a series of communication protocols employed in vehicles. From the lower to the higher end these include LIN; CAN; Flexray and Ethernet.

The Local Interconnect Network protocol (LIN) can support communication across 40-50 nodes (maximum 16 per network segment) and is a very cost-effective (but low data rate - 20kbit/s) serial bus which is engaged for tasks where latency is not going to cause any problems. For this reason it is mainly targeted at comfort electronics applications (seat positioning, mirror adjustment, climate control, interior lighting adjustment and similar functions).

More recently LIN is applied to diagnostic systems which are not dependent on real-time operation (such as tire pressure monitoring, temperature sensing, etc.). One of the overriding advantages of LIN, in addition to its low initial implementation cost, is its extensibility. Nodes are easily added to the network without the need for the hardware or software to be altered elsewhere in the network. LIN is further architected to cooperate with CAN bus protocol networks.

The Controller Area Network protocol (CAN) is already effectively ubiquitous within the automotive world, this is a long-standing protocol standard that features relatively fast, asynchronous networking with strong reliability due to automatic error detection and dual wire implementation. It is suitable for more time sensitive applications than LIN, such as powertrain, and can easily be implemented across 100 or more nodes.

However, in recent times there have started to be questions raised about its long-term prospects, as the bus loads that are now being placed on the average car are getting ever closer to the 100% mark. The industry is now preparing for the emergence of CAN with Flexible Data-rate (CAN FD), defined by Bosch and General Motors, as the bandwidth available for the original standard starts to reach its limit. CAN FD retains the same architecture as conventional CAN, while increasing the speed substantially.

This new standard benefits from dual bit data rates, increased data lengths and provides support for 8Mbit/s of bandwidth compared to only 1Mbit/s previously possible. The fact that it is compatible with legacy CAN networks means that it offers a more acceptable migration path than the other high speed networking options being considered. Nevertheless, the proliferation of CAN FD is certain to have its difficulties, with many chip suppliers struggling to introduce the highly sophisticated ICs and accompanying software that are essential to back up this new technology. Over the next 12 months it will be crucial that semiconductor devices are introduced having the capability to satisfy the expectations of this networking protocol.

High speed busses

Flexray is a high-speed protocol that can maintain 10Mbit/s and includes safety-critical features, putting it at the vanguard of X-by-wire deployment; where heavy mechanical systems are being displaced by more streamlined electronics to reduce overall vehicle weight and curb fuel consumption. So far Flexray has still only seen fairly restricted use, as it proves to be too expensive for widespread deployment throughout the vehicle. Another hurdle for Flexray is that it offers no possibility to add further modules at a later stage, making its architecture very inflexible. With CAN FD offering nearly the same data throughput, plus the additional benefits of being cheaper and easier to implement, it is not hard to envisage this protocol encroaching on Flexray’s territory.

Additionally, Ethernet is now being pushed hard by leading car brand, BMW. This protocol, which was previously thought of solely in the context of computing and data communication, is showing its value in high bandwidth automotive applications such as external cameras, night vision and multimedia entertainment systems. The first automobiles using this protocol are typically deploying it for service applications such as software updates. With 100Mbit/s data rates achievable (and higher data rates being worked on), it is almost comparable with the very fast Media Oriented Systems Transport (MOST) protocol. As a result Ethernet might begin to compete directly with MOST which, although offering 150Mbit/s, relies on semiconductor technology that has a higher price tag, as it is optical- rather than electronic-based. Ethernet could possibly be a more attractive option in certain scenarios, delivering the desired bandwidth without heavy investment being needed, as well as expending relatively low power levels (thereby not impacting too heavily on the vehicle’s emissions).           

Scaling

An important consideration when assessing the viability of vehicle networking solutions for new cars is the degree to which it can be scaled up. Whether they are founded on disruptive technology or enhancements to already established standards, the key to success of the communication systems implemented into vehicles will be their ability to facilitate flexible expansion. LIN and CAN (plus CAN FD) warrant a much lower cost per node in comparison to MOST or Flexray, Therefore it is advantageous, whenever possible, to base communication on these protocols.

Automobile manufacturers are also increasingly looking to follow a platform approach, allowing the re-use of IP across multiple car models. Luxury cars will have increased need for both Ethernet and LIN based control functions. At the same time, however, the smaller cars will need to appeal to young drivers with lower budgets. Greater use of LIN, to enable features such customisable ambient lighting, will allow car manufacturers to differentiate their models from the competition relatively cheaply. Features of this kind will mean that car owners can express themselves and put their own personal identity into the vehicle they drive.

Moving forward, automobile manufacturers will need to consider how best to implement these different networking technologies; a well thought out structure will be vital. They will have to ensure that there is provision for enough networking capacity. The platform strategy now being favoured means that manufacturers will approach the matter of networking from the ground up, starting with their more basic models, then adding extra complexity as they move through to the top range models.

This presents a considerable challenge, as they try to find an effective way to address the escalating number of nodes. In respect to the LIN network, for example, there are many subsystems; ambient lighting modules, climate control flaps, dashboard switch modules, which are all identical. This leads to configuration issues arising. In response the semiconductor devices specified to control them must incorporate sophisticated functionality. Networks will also need to exhibit increased robustness, in order to cope with the growing threat posed by electro-magnetic interference (EMI) and electro-static discharge (ESD). This again has major implications for the semiconductor companies serving the automotive sector, with the next generation of silicon they produce needing to incorporate more advanced protection mechanisms.

The necessity for greater in-vehicle networking will have an on-going influence on the automotive industry. It is clear that automobile manufacturers and their technology partners will need to continue to innovate so that they are prepared for new communication protocols as they emerge and are also better able to make use of existing protocols.

Author profile: Michael Bender is Product Marketing Manager for the LIN and power control product lines at Melexis. Based at its site in in Germany, he has worked for the company for the last 13 years. Prior to this he was a Technical Marketing Engineer at Thesys, where he worked for a period of 5 years. Michael has an engineering degree from the Ilmenau Technical University.