In-Vehicle networking

Enhancements to existing communication buses will help in addressing some aspects of next generation in-vehicle networking. The ongoing evolution of CANbus has led to the introduction of CAN-XL. This already allows 10Mbit/s data rates, with 100Mbit/s to follow soon and the intention for it to eventually be pushed up to 1Gbit in the years ahead. The main basis for data transportation though is going to be multi-Gbit Ethernet. Infrastructure will move from supporting 2.5Gbit/s to 5Gbit/s, then onto 10Gbit/s operation.

Though Ethernet standards clearly have the speeds, there are other factors that need to be considered. With a large percentage of the data passing through in-vehicle networks being safety-critical, low latency and deterministic operation are called for. Standard Ethernet protocols do not support deterministic data flows. As a consequence, time-sensitive networking (TSN) protocols have been developed and applied here. Through these, bounded low latency communication is provided - meaning that any data which is of a safety-critical nature will be guaranteed to arrive on time. The appropriate response can then be actioned: for example, imaging data could flag there being an obstacle on the road ahead, this data would be processed by the vehicle’s on-board computer and then the brakes applied, or an avoidance maneuver made.  

Alongside all this, there is an architectural upheaval underway. The domain-based architectures that define current in-vehicle networks will not be valid for fully autonomous driving. The domain approach to networking has everything arranged in relation to function. As the complexity of automotive systems increases and the network capacity needed has to be expanded, problems are going to emerge though. Retaining a domain-based approach would require a substantial ramp up in the vehicles’ cable harnessing. That would not only add to the overall vehicle weight (which will impact on the range that can be covered before recharging/refueling), but also result in an increase in component costs.

A zonal-based architecture offers much greater scalability. Adopting it will lead to more streamlined in-vehicle networking, with less space being needed for cabling and less financial outlay. Thanks to reductions in the weight involved, it will be possible for vehicles' range to be extended. There will be greater opportunity for integrating redundancy into systems too, in line with functional safety requirements. Also, provision will be put in place so that if one zone is out of action due to damage, another zone can cover it. The higher the level of automation within the car, the more acute the need for a zonal approach will be. It is expected that for the next few years vehicles will feature a combination of both architectures, before moving to a fully zonal-based one by the end of this decade.

Another way in which networking is being streamlined is through use of single pair Ethernet (SPE). Here multiple twisted copper pairs are substituted by a single pair, to save space and reduce weight. This still enables data rates far above those of CAN to be supported over relatively long distances. It must be noted though that this requires more comprehensive electromagnetic compatibility (EMC) testing.

EBV offers an extensive selection of communication ICs for in-vehicle networking. Among the most popular are: