back to article IEEE delivers Ethernet-for-cars standard

The march of Ethernet into motor vehicles continues, with the IEEE launching the first automotive standard for 100 Mbps Ethernet over single twisted pair cables. IEEE 802.3bw, aka 100BASE-T1, is designed to give the auto sector a single homogeneous network architecture, the IEEE says. The main game for auto applications is to …

  1. Christian Berger

    It's not weight they are optimising for

    It's cost, that's of course heavily correlated to weight as the heavy stuff of a cable is also the expensive stuff.

    1. Mark Wilson

      Re: It's not weight they are optimising for

      Actually it is probably both, if well implemented, this will allow cars to save weight and cost by having a single power line to each part of the car, think how many wires are required to power the rear lights, with each having its own wire running from front to back. That is a lot of copper. With the ethernet, each light cluster could have a single data line running to it and tap power of the main conduit then decide which lights should be on.

      1. Anonymous Coward
        Anonymous Coward

        Re: It's not weight they are optimising for

        Automotive did away with one wire per light (switch, etc.) many, many years ago with the use of CAN and LIN.

        1. Kevin McMurtrie Silver badge

          Re: It's not weight they are optimising for

          CAN Bus is too slow for cameras, audio, and various analog sensors. The analog stuff is especially bulky because it needs to be kept separate from other noisy wires.

      2. This post has been deleted by its author

    2. Anonymous Coward
      Anonymous Coward

      Re: It's not weight they are optimising for

      It's both, but weight is a real driver. High end vehicles can easily have over 100 Kg in the wiring harnesses. The space taken by the wiring is also an issue as the bundles can get very difficult to manipulate during manufacture.

      1. BlackDuke07
        Trollface

        Re: It's not weight they are optimising for

        "but weight is a real driver"

        Let's just have driverless cars then.

        1. Anonymous Coward
          Anonymous Coward

          Re: It's not weight they are optimising for

          Maybe they should switch to plastic fiber optics in the cars where weight is more of a concern than cost?

        2. Chemical Bob
          Trollface

          Re: It's not weight they are optimising for

          >"but weight is a real driver" Let's just have driverless cars then.

          Or put the driver on a diet.

  2. Lee D Silver badge

    I would think that a standard for a car would actually include some kind of shared-bus, similar to token ring or similar. Would that not be more practical in terms of weight? Rather than 50 cables all trying to come back to one point, you could just tap into the nearest cable for it and piggyback from it.

    I can't believe that a car is going to want to have a full, uncontended 100Mbps to every single sensor or whatever it's wired to, so surely a shared bus or ring architecture is a much better idea in terms of wiring than a star kind of topology? Even if you made it a 1Gigabit shared bus, with proper QoS for those devices that need to take up more urgent data?

    1. John Sager

      I wondered about that, but 100M multidrop is probably hard. It's much more likely that there will be a few point-to-point 100M connections to strategically placed hubs & CAN from then on to the light clusters.

  3. Benno

    Another standard?

    Do we need another standard, is there an issue with using the current 'almost CAT6' 28 & 30 AWG cables that are available?

    e.g.

    Telegärtner:

    http://www.telegaertner.co.uk/telegaertner-uk-news/press-releases/30AWG-Reduced-Diameter-Cat-6-patch-leads_165.htm

    Panduit:

    http://www.panduit.com/en/landing-pages/small-diameter-patch-cords

    (I use both of these in various fixed and mobile applications, they're both LSZH, so what's not to like?)

  4. Anonymous Coward
    Anonymous Coward

    I wonder if this will work over telephone extension cable?

    If so, that would be very interesting as a method to retrofit ethernet into houses that have existing telephone extensions.

    1. John Mangan

      When we had our house re-wired we ran CAT5 to every room and use it to convey phone signals to a couple of rooms..

  5. Warm Braw

    Ethernet-for-cars

    And not a word about collision detection?

    1. Cynical Observer
      Coat

      Re: Ethernet-for-cars

      Collision Detection

      Surely....

      Collision Avoidance would be a far better approach here?

    2. Lee D Silver badge

      Re: Ethernet-for-cars

      Same joke.

      Routing.

      1. This post has been deleted by its author

  6. james.aka.damingo
    Mushroom

    Here we go

    “IEEE 802.3bw, aka 100BASE-T1, is designed to give the auto sector a single homogeneous network architecture,”

    100BASE-T1…. T1…. http://terminator.wikia.com/wiki/T-1

  7. Stuart Elliott
    Coat

    IEEE 802 for cars

    Shouldn't that be 802.3bmw ?

    1. David Lewis 2
      Joke

      Re: IEEE 802 for cars

      No. BMW (drivers) don't do signalling!

  8. annodomini2

    Limited use

    There are functions that will need this kind of bandwidth, but not that many.

    It will become more prevalent with autonomous vehicles, with potentially high bandwidth sensors.

    Question is does it offer more than say CAN-FD or Flexray if there is this need?

    One of the main advantages of CAN is that a large portion of the bus management is done in HW. When operating on lowest cost Microcontrollers this can make a big difference!

    Especially when they can be running between 80-90% CPU load when operating, adding extra communications processing is going to be greater cost add.

  9. Nigel 11

    Cat-5 and 100BaseT

    Cat-5 (and 5e, 6) cable has four pairs but 100BaseT uses only two of them (1000BaseT uses all four). So this new standard is saving just one twisted pair.

  10. Will Godfrey Silver badge
    Unhappy

    Meh!

    I'd be more interested in hearing about a standard for ensuring the 'toys' network can never, ever interfere with the systems network.

  11. Anonymous Coward
    Anonymous Coward

    Ethernet for real time?

    Never been a fan of ethernet for realtime stuff - collisions and all that - I know you can work around it but just saying.

    I'd prefer a time division protocol where talkers on the network are only allow when it is their turn - but I'm probably seen as a dinosaur.

    1. Sandgroper

      Re: Ethernet for real time?

      Ethernet for real time?

      Isochronous Ethernet from the 1980s? Token RIng?

    2. Lennart Sorensen

      Re: Ethernet for real time?

      Ethernet hasn't had collisions since we started using switches, so that is not much of an issue any more. Temporary block in the switch though because some other packet is currently being sent out a given port is still an issue though. And there are Ethernet extension standards to deal with bandwidth reservation and such for those cases where that matters.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon