In-vehicle networking today and tomorrow
In-vehicle infotainment
The future of infotainment networking can be divided into two categories. For the European luxury brands, who have driven the adoption of the MOST standard from the very beginning, the future will for sure look different than for the many followers who are more focused on the mass market than on the high-end premium segment.
As of today, only between 10 and 20 percent of all passenger vehicles are equipped with an infotainment network. At the same time, many car makers who have not been highly focused on infotainment to date, are facing the need to shape up in order to keep pace with the digital world, offering networked applications. For those car makers, it is very important to bet on a technology which is proven and shipping in high volume, offering state-of-the art performance and low price at a minimum risk. The re-use of the MOST50 or MOST150 technology is a rock solid and very straight forward approach, providing a well understood network and taking advantage of the existing MOST specifications, components, software stacks and development and debug tools. The focus is on low cost, low risk and a very fast time to market. The available data rates are more than sufficient for the transmission of even multiple High-Definition channels and the protocol stack which has been developed and debugged for more than 10 years offers almost everything needed.
With the MOST50 UTP electrical physical layer, which has been in production since 2007, even a point-to-point link between a head unit and a DSP audio amplifier can be cheaper than an analogue connection plus a digital control bus.
Looking at the premium car makers, the requirements are a little different. In-vehicle infotainment is about to saturate in terms of new and useful functions and features. Connectivity outside the car (e.g. through LTE or Car-to-X communication) is becoming much more important than pure in vehicle infotainment networking. More and more IP based applications will arise, making extensive use of the Most Ethernet Packet channel (MEP), which is relatively new and specific to MOST150. Here, any standard protocol stacks can be used on top of MOST, providing both the advantages of a synchronous network and a powerful data pipe for IP/Ethernet data.
Equally important is an appropriate High Speed Interface connecting the network to equally high speed processing devices, making use of the full network bandwidth without loss.
Bandwidth requirements will be driven by fast software updating and fast media access to on-board mass storage devices (HDD/SSD), portable consumer devices attached via USB, and car-to-x applications, connected via either WLAN or LTE.
Driver assist
The need for a Driver Assist network is a hot topic in the industry and there is the fundamental agreement by many OEMs and Tier 1 suppliers that there will be a new cluster in the Electric/Electronic Eco System of the vehicle for Driver Assist applications.
However, as with all new technologies, one of the key questions is how to make a start. Besides all technical discussions, the economic implications will play a major role in decision-making. Using MOST150 for Driver Assist networking has the clear advantage that from the very beginning, future applications can benefit and take share in the volume generated by the infotainment market. From 2013 onwards, MOST150 based Driver Assist networks can have a ‘jump start’ in terms of volume. In contrast, any other technology, which is not yet established in the automotive market, will have to blaze its trail through low initial volumes, immaturity in the vehicle and so on, like MOST and FlexRay experienced in their infancy.
It has been shown that by adding an appropriate safety layer, MOST can be enabled to address Functional Safety requirements up to ASIL D, while at the same time being able to re-use a major part of the technology, specifications (in particular car specific robustness and diagnostic features) which have been developed and debugged over the last 13 years. Over the longer term, uncompressed video transmission requirements may drive the bandwidth requirements of a Driver Assist network, too.
The MOST roadmap
A major strength of the MOST technology consists of carefully balancing technological possibilities, real market requirements and economic considerations. From that perspective, it is extremely important to show a clear cost-down roadmap parallel to technological advancements. Several different aspects need to be addresses, including data transport capacity (bitrate), physical media and topology options. As there will be an increasingly strong tie between infotainment/telematics and Driver Assist, requirements from both areas need to be fulfilled.
At the MOST Forum 2010, a prototype version of a Multiport-INIC for MOST150 was presented. The OS81111 implements a two port architecture which enables prototype design for alternative network topologies.
Fig1: Topology Options using OS81111 Dual Port INIC for MOST150
In a very elegant way, several topology options can be realized, included ring (key to allow for the absolute minimum number of connections in a network), daisy chain and a star. Obviously, in such a ‘pseudo-ring’ architecture, the bandwidth is shared between the different branches in the event that a star topology is implemented.
Within the last year, technology has advanced and the requirements have become more defined. To meet these needs, a true multiport INIC chip is being developed by SMSC, eliminating the bandwidth sharing of the prototype:
Fig2: Block diagram Multiport INIC (for full resolution, click here)
With this chip, each branch will provide the full 150 Mbit/s data transport capacity. In addition to the well established MediaLB 6 pin interface providing 300 Mbit/s bandwidth, a standard high speed interface offering minimum 1.2 Gbit/s speed is implemented. The interface will match to state-of-the art interfaces used by powerful media processors used in multimedia or driver assist processing units. The different branches themselves will run on MOST150, with the add-on option that each branch can be hot plugged or cut off without impacting data communication within the rest of the system. Of course, each branch may again consist of a ring itself, or a daisy chain, in case the Dual-Port INIC is used. With such an architecture, those use cases which require a true star architecture, e.g. coming from the driver assist domain, are addressed, while at the same time maintaining the principal advantages of MOST, including the synchronicity and the low latency of the network.
Higher bit rate? Why? When?
It has always been a particular strength of MOST to not shoot for the highest bitrates technically possible, but rather find the appropriate ‘sweet spot’, as discussed previously. Instead, the development of new functions and features, speed grades and physical layers has always been driven by the MOST Cooperation, collecting the real market requirements of a broad community. In fig.3 the driving functions for Most25, Most150 and also the future generation are shown.
Fig3: MOST roadmap
The absolute speed grade for the future generation of MOST is not entirely determined yet. The same applied for MOST150 at the time: It was clear that a bandwidth of more than 100 Mbit/s would be required, but the exact bandwidth was determined by other parameters, e.g. the strong requirement of keeping the POF and the connector system identical to MOST25.
Technically, the feasibility of implementing a speed grade in the range of 5-10 Gbit/s for a next generation has been investigated already. As discussed previously, the even more interesting question is: What does the automotive market really need? Why would people like to transport data at such high speed and which data shall be transported at all?
And, even more importantly, what is the appropriate physical medium and what is the overall system cost, including an automotive grade Network Interface Controller, physical medium and connector system? Solutions are being developed for POF, PCS and Coax Physical Layer, with each offering certain advantages but also trade-offs. For instance, staying with POF and boosting up the bandwidth to the 1 Gbit/s range by smart modulation techniques may be a very interesting option, allowing to keep the advantage of handling a very well known optical cable and connection system, taking full advantage of the MOST150 volumes in the market. Another question is, what will be the appropriate time line to introduce a next generation to the market, given the fact that MOST150 is just about to be launched in production this year?
Some boundary conditions are crystal clear already:
One fact is that the bit rate will be beyond 1Gbit/s. Both the optical and electrical physical layer will be available, and most importantly, MOST will stay a synchronous network providing a payload efficiency of nearly 100%, with the ability to seamlessly transport IP packet data. In order to serve the Driver Assist market, relevant parameters will be considered to fulfill the functional safety requirements according to ISO26262. Other open items, such as appropriate interfaces to the application processor and of course overall system cost are being considered before the product and system concept can be finalized.
MOST in the market
As mentioned before, it is absolutely critical for any new technology to cross the chasm between being a niche technology and becoming a widely accepted standard. Several factors need to align in order to make it happen, both from the technical and commercial sides.
Looking at semiconductors for the automotive industry, the difference between the semiconductor world and the automotive industry is quite significant: e.g. the definition of ‘reasonable volumes’ required for state-of-art silicon solutions in terms of performance, quality and price, is quite different between the semiconductor and the automotive world. The same applies for product life cycles, particularly in the area of fast moving technologies like infotainment and Driver Assist.
To overcome this hurdle, repeated attempts have been made to import technologies from the consumer space into the car and thus gain the respective scaling effects in terms of volume and price, most recent examples being USB and Ethernet technology. However, at second glance, it becomes very obvious that this does not work well.
Technologies to be used in the car need to be optimized for the car, with respect to either the environment (temperature, EMC, other physical parameters such as cable length or number of interconnects) or the application. In the case of infotainment/consumer networking, there has been no major cross brand standard accepted and implemented by the major players of the consumer space, although repeated attempts have been made. And all the cellphone/smartphone industry could agree to harmonize is a common charger.
The conclusion is that the automotive requirements, particularly in the networking area as a part of the ‘DNA’ of the car require automotive specific developments. But, once a technology has been developed and/or optimized for the car, adding all the specific functions and features required for in-vehicle use, nobody in the consumer industry will want to pay for these things anymore because it is not needed in the consumer world. All in all, it leads to optimized solutions for the car, such as CAN, FlexRay and MOST.
In that context, single source supply situations also need to be seen in a different light.
MOST has been in the market since 2001, having crossed the chasm with implementation in approximately 12 percent of all passenger vehicles. In 2007, SMSC and Harman announced the opening of the MOST Link Layer for MOST25 including a clear roadmap for MOST150.
Perhaps not surprisingly, no company has licensed the technology for MOST25 so far. The reasons are pretty obvious: In contrast to CAN, where several hundred million nodes are sold each year, the overall market size is just too low.
For MOST50 and MOST150 however, the picture looks quite different, even if we consider infotainment applications only, not looking at Driver Assist for a moment.
MOST150 has been adopted by Volkswagen/Audi, Daimler and some more car manufacturers already. The volume curve will start to rise more steeply from 2013/14 onwards, when the roll-outs to high volume models will start. At the same time, MOST50, which is very well suited for the mass market (as Toyota/Lexus have proven) and has already found more adopters in the ‘fast follower’ community mentioned above, will follow a similar pattern.
This will lead to quick cost reductions, which makes it very attractive for non-infotainment application areas, i.e. Driver Assist, to participate in the high volume effect from MOST in the infotainment area. That said, it can be expected that MOST interfaces implementations by other semiconductor companies will now make technical and commercial sense, in contrast to MOST25.
For CAN and MOST, it has taken more than 10 years to reach such a point. Any new technology will basically have to follow the same pattern. In times with even greater cost pressure, it is obvious that the pre-investment, which has already been made for MOST for over a decade, will have to be made for other technologies, being new to the car. Crossing the chasm means offering superior technology at low price for low volume over several years, compared to the existing ones.
MOST has already crossed it – with a clear future roadmap, technology- and cost wise.
Against the background of recent announcements regarding Ethernet as the infotainment networking technology of choice, the author of this article added the following statement:
MOST is here to stay and grow. Automotive IP-based data transmission over MOST now supplements infotainment which is about to saturate in terms of new and useful functions and features. Prepared for rollout in early 2012, MOST150 provides an automotive-ready physical layer for IP/Ethernet protocols in addition to transporting high Quality of Service (QoS) audio and video within the car. This makes MOST the ideal network backbone for a broad variety of IP-based applications such as supporting apps on connected services and Internet access in general. MOST is from this perspective comparable to an IEEE802.x network while on top providing high speed, real time, synchronous A/V streaming.
With MOST having already crossed the chasm from a niche technology to mass market the overall system costs including research and design, components, cabling, software stacks, etc. are being reduced significantly. The volume curve will start to rise even more steeply from 2013/14 onwards, when the roll-outs to high volume models will start. This will lead to further cost reductions, which makes MOST very attractive also for non-infotainment application areas such as driver assist to participate in the high volume effect from MOST in the infotainment area.
About the author: Harald Schoepp is Vice President Marketing of the Automotive Information Systems division at SMSC. He is founding member of the MOST Cooperation and representative of SMSC in the steering committee.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
