Vehicle networking opportunities

Deres Eshete from On Semiconductor explores the potential for more advanced in-vehicle networking technology
There is a wealth of functionality now being incorporated into automobiles, for comfort, infotainment, safety and various other purposes. Data transfer levels have, as a result, been raised considerably in recent years and this is putting a great deal of strain on the network infrastructure currently in place.
It is clear from this that existing communications protocols need to be revamped and, in addition, new protocols must be explored. Furthermore, there are major implications for the supporting semiconductor technology. The table summarises the properties of the major interface technologies.
The complement of electronic control units (ECUs) that make up a modern automobile design is almost double what it was a decade ago. It is not uncommon now for a luxury model to have more than 100 ECUs installed. For all these units there needs to be adequate provision for them to communicate with one another in a fully effective manner.
Existing technologies
There are at least 400 million in-vehicle network nodes in the car models being driven on our roads. The average automobile today will have a minimum of eight to ten nodes. The vast majority of these will be based on controller area network (Can) and local interconnect network (Lin) protocols.
Can is regarded as the foundation of any vehicle’s internal communications resource. This deeply entrenched protocol offers event-triggered, asynchronous operation at relatively high speed (transferring data at 1Mbit/s across its twisted pair cables), as well as possessing a respectable bandwidth capacity. Can is also very robust, with reliability enhanced thanks to its built-in automatic error detection feature.
Can is accompanied by Lin infrastructure, which only supports quite modest data rates (20kbit/s), so it is really just suitable for latency tolerant applications such as climate control, interior lighting control, seat positioning and vehicle entry systems. Lin may lack bandwidth, but it is very easy to implement and to scale up, with the adding of additional nodes being relatively simple to accomplish – though it must be noted that it can only support approximately half the number of nodes that are possible with Can (50 compared with 100). In addition it is extremely cost effective, presenting a very low investment per node.
Emerging networks
With Can infrastructure edging ever closer to its capacity ceiling, there is growing demand for a higher performance version of this protocol to be deployed in automotive designs. With Can Flexible Data-Rate (Can FD), the core architecture used in normal Can implementations remains, but the bandwidth overhead is raised significantly. Can FD is able to support 8Mbit/s – almost an order of magnitude greater than conventional Can – thanks to the employment of dual bit data transfer and extended data lengths. Its compatibility with legacy Can infrastructure means there is already a well-defined migration path in place.
Targeted mainly at safety related applications, Flexray can deliver 10Mbit/s data transfer capacity. It also offers time-triggered and event-triggered behaviour, plus redundancy. Flexray is able to deliver the error tolerance, as well as the speed-time-determinism performance needed for x-by-wire deployment in next generation vehicles. However, though it has been hyped up considerably, this in-vehicle networking protocol is still to make any serious impact on the automotive industry. Among the main reasons for this are the higher costs associated with it, along with the difficulties involved in scaling the network up, as there is very little scope for adding further resources after the initial implementation.
Flexray may as a result become marginalised, with Can FD able to deliver very similar bandwidth levels, but without any of the inconvenience.
There seems to be a great deal of opportunity for Ethernet, which is already commonly used in the computing and telecoms sectors, to be deployed within automobiles in the future. With 100Mbit/s data rates achievable, this could provide the bandwidth needed to deal with all manner of data-intensive automotive applications. It seems there will be quite a few car models featuring Ethernet technology destined to hit the market in the 2016-17 timeframe.
It can be employed in both infotainment and safety applications, dealing with the transfer of imaging data and so on. The longer term goal will be to install 1Gbit/s data rate Ethernet however, as this will be of value as cars start to be fitted with various advanced driver assistance system (adas) components delivering an abundance of information to processing units for enhanced safety and functionality.
Other interconnect technologies include media-oriented systems transport (Most) and single edge nibble transmission (Sent). Intended for the transfer of high-resolution multimedia data, these offer impressive bandwidths (150Mbit/s in the case of Most), though so far, just like Flexray, they have only witnessed relatively low market uptake. This is because the implementation costs for each of these are quite high.
There is a strong possibility that Ethernet, which can support data rates almost as high while still being simple and economical to put into operation, may prove to be a more attractive alternative.
With the cost per node that is less than for other protocols, combined with the obvious advantages of backwards compatibility, it seems certain that Can and Can FD will play a big part in vehicle networking for a long time into the future. The prospects for the rest of the protocol technologies discussed are a little harder to predict.
Protection
With the on-going reduction in the mechanical elements of automobile designs, to lower weight and thereby improve fuel economy, we are rapidly reaching a stage where in-vehicle networks will become the backbone of most modern cars. Gateway modules within vehicles will integrate multiple nodes supporting a plethora of different standards. These will need to talk simultaneously to the various network structures.
Every effort must be made, despite this intensified degree of intricacy, to ensure that data quality is maintained at the highest levels, otherwise operational performance could be compromised and the safety of vehicle occupants might be placed in jeopardy.
With highly sensitive semiconductor components needed to support the array of in-vehicle networking protocols outlined here, sufficient measures must be taken to safeguard them against the effects that automotive deployment can have on performance and operational longevity. The ICs specified need to be able to cope with noisy environments and have sufficient protection against voltage transients.
The automotive industry must be adequately armed to combat the threat posed by electro-magnetic interference (EMI) and electro-static discharge (ESD). With a multitude of nodes in these gateway modules, component densities are increasing and available space in short ration. It is vital, therefore, that the protection components that are addressing such implementations are extremely compact in form.
At the elevated data rates that will now start to be supported by in-vehicle networks, tighter capacitance matching between high-speed paired cables such as those for Can FD will be necessary, so that acceptable signal-to-noise levels can be maintained. Low capacitance of the protection diodes will be mandated throughout the entire signal voltage range.
Likewise, the clamping voltage will have to be constrained throughout the full spectrum of possible ESD and transient conditions. Support for flow-through routing is going to be called for to reduce sharp turns and facilitate better impedance matching.
Flexray and Ethernet protocols both have quite effective capacitance matching and lower capacitance level requirements. This translates into the need for improved protection products that provide the low on state resistance to ensure the needed protection while having low capacitance necessary for high signal integrity.
There is a heightening complexity of electronic circuitry within automobiles, as a broader spectrum of different tasks is tackled, and there is no doubt connectivity requirements will only continue to increase. This means current in-vehicle networking technologies are beginning to reach their limits. A more sophisticated approach to the protection of the semiconductor hardware involved is therefore obligatory.
Deres Eshete works for On Semiconductor
