Is that because MOST is too well established a bus in the in-vehicle infotainment network, and automakers are less inclined to replace it?
Lau: Our sense is that automotive OEMs will start embracing automotive Ethernet as more stuff – like heads-up displays, digital amplifiers, rear-seat entertainment, information clusters, etc. – begin hanging off the infotainment network. MOST will eventually run out of network bandwidth, because it's based on a ring network. Its network architecture forces newly added clusters to share bandwidth. In contrast, Ethernet is based on a switched network -- not a shared network like MOST – so that it can adapt to adding more complex, higher performance systems to the network.
EE Times: I hear often ADAS is what’s fueling carmakers to embrace Ethernet. But take an example of the number of cameras installed at the front end of a vehicle. They’re there to detect objects running in front of a vehicle, and to determine what they are. My understanding is that automotive OEMs wouldn’t like you to compress that video. If that’s the case, the 100Mbits offered by Ethernet is not going to help ADAS.
Lau: That’s a good question. We see there are two types of ADAS –passive and active. Ethernet is perfect for passive ADAS, which includes applications like parking assist. But for active ADAS, you’re correct, car OEMs want video to be lossless so that images captured can be sent uncompressed to an image cognition block, in which its algorithm determines what it is.
IEEE standard's coming
EE Times: So, as I understand it, IEEE is putting together an automotive Ethernet standard, 802.3bw (also known as 100BASE-T1), slated to come out at the end of this year. Assume I’m a chip vendor. Shouldn’t I wait for the IEEE standard to get completed, rather than jumping the gun to become a member of OpenAlliance (which promotes BroadR-Reach)?
Lau: Nobody wants different flavors of automotive Ethernet.