B 61060 · Februar 2014 · Einzelpreis 19,00 € · www.automobil-elektronik.de ·
Das Automotive-Magazin von all-electronics
Interview with Wiren Perera,
Vice President at Micrel Page 14
Reprint for Micrel
Synonym for Automotive Ethernet
The advent of automotive Ethernet has been accompanied by one single semiconductor company so far: Micrel.
AUTOMOBIL-ELEKTRONIK talked to Wiren Perera, Vice President at Micrel, about the automotive Ethernet market
drivers and trends from ADAS, infotainment, diagnostics and other applications, Ethernet semiconductor solutions
and – of course – how it all began and why Ethernet will become the predominant in-car network in the future.
Author: Alfred Vollmer
Exclusive Interview with Wiren Perera, VP LAN Solutions Corporate Strategic Marketing at Micrel
AUTOMOBIL-ELEKTRONIK: What are the main market drivers for
Ethernet in automotive applications?
Wiren Perera: If you consider an event like CES, where you rarely
saw a car on the show floor some years ago, you‘ll find that a lot of
the big booths there will show one if not two cars. So most of the
interest, from a consumer aspect, has moved into the automobile
and, fundamentally, what you‘re seeing is the transition of what we
call the digital home, which is all of these capabilities utilizing In-
ternet-based services and content delivery and multimedia servi-
ces. Basically, everybody expects to do pretty much the same thing
if you‘re in the car today.
The second area, falls into a category described as advanced dri-
ver assistance systems, or ADAS. Eventually, leading to driverless
cars, they are things like navigation, rear view cameras, collision
detection, traffic sign recognition, warning systems, etc.
The third category is basically the same capabilities that you‘ve
got used to with all of your consumer devices. Both the suppliers of
the vehicles and the users want to run diagnostics and upgrade
What are the consequences for automotive Ethernet?
All of these three areas and usage models that are driving growth
in automotive center around electronics, and all of that centers
around networking and communications, and so this is pretty
much the opportunity that we‘re trying to address.
What will be the next step?
The first point of entry for Ethernet technology was the dia-
gnostics port used for two things. One is when you drive
into a service station to be able to hook up to an Ethernet
computer network system to run the diagnostics; the other
is to upgrade software both for the vehicle and periphe-
rals. In fact, the suppliers of those peripherals, just like
you upgrade your computers, tablets and smartphones,
want to make improvements, add services, offer diffe-
The future for proprietary technologies
like MOST, especially ones that are in
the speeds that Ethernet easily
meets or surpasses, is fairly limited.
Wiren Perera, Micrel.
rent applications. So that again is all network-centric and they
want to bring in those capabilities into the automobile.
But why should car manufacturers install Ethernet rather than other
Primarily, because all these applications need increasing band-
width. Ethernet gives you higher speed but it also gives you a road-
map. Today an Ethernet network in the car is centered on 100
Mbit/s, and the world of Ethernet is actually going to 40 Gbit/s and
100 Gbit/s, so there‘s plenty of room for growth.
What are the challenges for Ethernet in automotive applications?
Basically, make Ethernet work in the automotive environment.
That‘s where the uniqueness of Micrel‘s products comes in. That‘s
where we innovate, differentiate and bring value to our customers.
Electronic components and subsystems that you use in an au-
tomobile should have low emissions, because you‘ll interfere with
And, conversely, it has to be able to stand emissions coming
from other things, especially in safety. Therefore, the EMI issue is
of critical importance.
Ethernet didn‘t meet those requirements before Micrel intro-
duced a technology, that we call Quiet-WIRETM, which solves
the EMI problem. That was the first step on the innovation path:
making standard Ethernet meet automotive OEM EMI limits so
that Ethernet can be adopted in the vehicles.
Who initiated the use of Ethernet in automotive?
Really, the desire to adopt Ethernet came from the automotive
OEMs because of the multiple other benefits of Ethernet.
First, the low cost of ownership. The solutions themselves are
cost-effective due to the very high volumes of Ethernet component
shipments into other markets. The same applies to the surrounding
components and so you have a very low bill of materials cost.
Furthermore, OEMs want semiconductors from multiple sour-
ces. The beauty of standard Ethernet is that inherently you have
multiple suppliers, which means our customers have a choice and,
of course, multiple suppliers result eventually in lower cost due to
competition. The key to that is interoperability. We proved that
Micrel solutions interoperate with solutions for example from
Marvel because we both provide a standard Ethernet solution.
Another big benefit is the ecosystem in terms of testing and ma-
nufacturing Ethernet systems, which is already in place. If you bu-
cket everything into total cost of ownership, there‘s a huge, huge
benefit in adopting standards based Ethernet.
Could you please quantify that?
When we compare ourselves against a proprietary, Ethernet-like
technology, our estimate is that we save multiple dollars per link.
Now, multiply that by several links that you need and you’ll obvi-
ously have a huge benefit. That‘s at the bill of materials level.
Tier-1 suppliers, favor the use of standard Ethernet versus so-
me proprietary method that would achieve the same capabilities,
because they supply the subsystems to the OEMs and need to ha-
ve all the test equipment and the ecosystem in place. This is an
intangible but significant.
Furthermore, to improve fuel efficiency and consumer costs
you have to reduce weight. That’s why the OEMs want Ethernet to
meet the automotive requirements but on unshielded cable, be-
cause shielded cable is a no-no in the automotive world for that
Wiren Perera during the interview with Alfred Vollmer, editor of AUTOMOBIL-ELEKTRONIK: „By 2018/19 all cars are supposed to have Ethernet. So that will
drive initial growth but the real massive growth of Ethernet will come through these camera networks and infotainment networks.“
We have made automotive Ethernet over unshielded
cable a reality.
Wiren Perera, Micrel.
very reason. We have made automotive Ethernet over unshielded
cable a reality.
Which other benefits does Ethernet provide in automotive environ-
We’ve actually come up with a power management scheme, which
we refer to as Auto POE where PoE means Power over Ethernet.
Every system needs power requiring another set of cables. So, very
simply put, we‘ve found a way to use that same set of cables to po-
wer the whole network and the system. Therefore you basically
halve the number of cables. In addition, with Auto PoE as done by
Micrel, we essentially use the same power management compo-
nents used to power those subsystems so you literally get power for
free. You have no additional component bill of materials costs and
you have no additional cable costs.
Also, using standard Ethernet you basically get what I call holis-
tic solutions. Aside from PoE, another Ethernet standard is AVB -
audio video bridging. This essentially provides quality of service
and synchronization of content; for example multiple speakers and
video with audio.
In which way does your solution pave the way towards Energy-Effici-
Energy-Efficient Ethernet (EEE) is yet another IEEE standard.
Our automotive products support EEE which provides a 50%–75%
power savings in idle mode. On top of this we’ve implemented
ultra-low power schemes on our products enabling further energy
and fuel efficiency.
How many Watts can you supply over Ethernet to a remote sensor,
like the camera in the back of the car?
The standard Power over Ethernet today goes all the way to 30
watts, but we don’t think that’s necessary in the car. We think that
something in the 6 watt range will be more than sufficient for any
end device. More than likely it will be a lot less than that. Idle po-
wer is as important as active power. We use the specified mecha-
nisms within Energy-Efficient Ethernet. But then we’ll go beyond
that. We have special cases where we really power things down and
take them down to a whole different level.
What is the maximum number of Ethernet nodes that you know of in
cars currently on the road?
Today there are probably three to four Ethernet nodes per car. A
diagnostics port that is also used for software upgrading. This gate-
way node may then be connected to the head unit, and then the
head unit may be connected to a rear seat entertainment unit.
What about Ethernet in ADAS applications?
A camera network typically consists of four to five cameras, which
would be on top of what I talked about. In a star network all these
cameras are individually wired via Ethernet to a central processing
unit, maybe in the head unit, which is where the image gets conso-
lidated and processed. In a ring network you use less cable because
you don’t have to span long distances for the individual links. So
you can easily have ten plus Ethernet nodes in a vehicle.
How many Ethernet nodes per car do you see in the mid and long
If we have three today, I would say in the mid-term it will probably
be around eight to ten, and probably somewhere between ten and
20 in the longer term: in diagnostics, infotainment, camera net-
works and ADAS. Then the question will arise whether to replace
all the other lower level networks that are used for control applica-
tions in the car. I suspect that once the automotive market becomes
comfortable with Ethernet and its adoption and the costs also go
down to a level that becomes acceptable, it will start replacing the
other networks as well. Today it is primarily replacing things like
MOST technologies and the higher speed parts of the network.
Does this mean that technologies like MOST will die?
Over the long term, yes, for all of those reasons that I outlined.
MOST doesn’t have the speed upgradeability that we bring. The
cost of ownership benefits that we bring from Ethernet are not the-
re in MOST. Multi-sourcing is not there. So I would say yes, the
future for proprietary technologies like MOST, especially ones that
are in the speeds that Ethernet easily meets or surpasses, is fairly
Having talked to OEMs and Tier-1s which timeline do you see for the
Ethernet introduction in cars?
I think you’ll see the initial waves of Ethernet between now and
maybe the next, say, three years, but I think after the 2016/2017
timeframe you’ll find that most of the designs for camera networks
and infotainment and so on will be Ethernet.
Which growth rates do you see for Micrel in the automotive Ethernet
The Ethernet business in the car is growing with a 10% to 20%
CAGR and there’s a hockey stick coming post 2016 when adoption
goes up. Initially it’s for diagnostics and by 2018/19 all cars are
supposed to have Ethernet. So that will drive initial growth but the
real massive growth of Ethernet will come through these camera
networks and infotainment networks.
What made Micrel enter the automotive Ethernet market?
This is probably the first market that I’ve dealt with where our
customers actually asked us to enter. We’ve never actually had to
sell the concept; it was sold to us.
We’ve been shipping Ethernet in cars since 2008, and we’re the
market leader. When it comes to automotive Ethernet, we probably
have close to 100% market share.
For such a system-critical device the OEMs will not accept a single-
source in the long term…
This is why we’ve collaborated on interoperability. The market will
grow if there are multiple suppliers.
Where – in terms of region and product segment – are the main
Ethernet design-in activities?
It depends on the application. In diagnostics our leading customers
When it comes to automotive
Ethernet, we probably have
close to 100% market share.
Wiren Perera, Micrel.
now use 100% Ethernet, so it’s not just the high-end models. That
will be a trend across all of the car manufacturers as we look out
over the next five to ten years as they adopt Ethernet for diag-
nostics. With the follow-on customers the transition from high-
end to low-end cars will be even more rapid.
The leaders worldwide are already migrating to Ethernet-based
camera systems and Ethernet-based infotainment systems. It’s
happening for the mid to high end cars for those manufacturers,
and then it will just percolate down.
What will be the implementation strategy of Ethernet? Just one sing-
le Ethernet bus or multiple?
You touched on a very important point, which is what we call
Ethernet network ubiquity. We mostly replace the CAN bus for di-
agnostics. In infotainment, we’re probably going to replace the
MOST bus and, in camera networks, there are various different ty-
pes of proprietary technologies that are used today, some analogue
and some digital. So Ethernet is the only technology that can be
used across all of these. There’s another massive benefit that you get
out of this: You don’t need bridging. Very often you will have a ca-
mera network on a certain technology. You then have to bridge it to
get it into the gateway where things are aggregated. With Ethernet
you don’t need that and you get another cost reduction.
Certainly for diagnostics, for camera networks and for info-
tainment you can see one single bus will serve all the different
functionalities. And then I think comes the interesting question
of whether Ethernet will then replace the networks used for con-
trol and things like that, and that will be mainly a question of
comfort with the technology and cost going down. n
The interview was conducted by Alfred Vollmer,
editor of AUTOMOBIL-ELEKTRONIK.
infoDIREKT www.all-electronics.de 300AEL0114
The leaders worldwide are
already migrating to Ethernet-
based camera systems and
Wiren Perera, Micrel.