2

I was told that after a collision, the first device that detects the collision and sends the JAM signal has somehow priority to send first after the collision.

I do not believe that this is true and tried reading the IEEE 802.3 section 1 to find the details.

But all I understand so far, is that after sending the JAM signal, the device tries again after a random time.

My theory is that after the JAM signal ends, every device that is in the collision domain waits for a random time after which each device tries to send again. So there would not be any priority.

I hope that any of you can clarify for me, if it is random who sends first after a collision or if any of the devices in the collision domain has any priority after a collision.

3

Your read of the spec is correct. After the random backoff period, the NIC is free to transmit again. The key to this working is everyone picks a different amount of time. Naturally, someone will pick a smaller number and be free to transmit first -- collision, thus, resolved. If a collision happens again, everyone pauses for a longer random period. After several successive collisions, an excessive collision error is raised and the packet is dropped.

(This is all academic as nobody does half-duplex anymore. No one has made true hubs for decades.)

3
  • 1
    Thanks for you answer, but just to make sure: Other NIC then the one sending the JAM signal can have a lower random wait time then the one that sends the JAM signal. correct? – Flikk Jan 5 '16 at 9:27
  • 2
    Yes. Random means RANDOM. – Ricky Jan 5 '16 at 10:05
  • Binary Exponential Backoff Protocol ( en.wikipedia.org/wiki/Exponential_backoff ) – Lord Loh. Apr 6 '17 at 5:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.