hi there, im new to this forum, so just like to day hello to everyone!!!

i know im not aloud to post homework questions, but is it ok to ask for a formula to use to answer a question?

i looking for a formula for collison backoff algorithm to find the average time to successfully transmit a complete frame.
i have the the sslot time and successive collision.

also is there any type of formual for snmp polling intervals?
i have the number of host and sustained output

i dont no which forumal to use can someone guide me in the right direction; i tried searching the web but cant find anything??


thanks so much and im sorry if im breaking any forum rules!

pure!

Well, first, the backoff algorithms are usually at the MAC (Medium Access Control) layer, so it depends what medium (thin Ethernet, UTP Ethernet, Fast, GigE, 10Gig, Wireless, etc) you're talking about. When I was taught this subject years ago, the backoff algorithm used by Ethernet was exponential, starting maybe at 2 clock ticks or something. (Then 4, then 8, then 16). But maybe there was a random fuzz factor involved. (If there is, then computing the average would be to compute the mean of the random range, and add that to whatever).

The network is full of papers on this topic.

Google: collision+backoff+algorithm

Exactly this was the case and it is one of the great ideas of IIRC Bob Metcalfe to have done so. The "fuzz factor" was a random number by which the increasing number was multiplicated. Thus two colliding nodes probably weren't colliding at their next try, because the randon numbers would be most probably different.

"Pure" ALOHA (the first CSMA network) without timeslots operated at a ~18% throughput / raw bandwith ratio, whereas "slotted" ALOHA (with timeticks) roughly doubled that efficiency. Ethernet (the successor of ALOHA) had an even higher efficieny.

bakunin