User talk:Zanna 37

p-persistent CSMA defer process

Dear Zanna37, I was trying to find the truth about p-persistent CSMA. There are people (and I am (was) among them) that, as you pointed out, say that when the channel is sensed IDLE then the station transmits immediately. Yet in many sources including the "Computer Networks" book from Tanenbaum it is written that when the channel is IDLE the station transmits with probability p, and delays the decision of 1 slot with probability 1-p. To find the answer, I had to go back to the original paper from Kleinrock and Tobagi. In there I actually found something that, in my opinion, says that if the channel is IDLE, then the station transmits with proabability p. Quoting from the paper:

The scheme consists of including an additional parameter, the probability that a ready packet persists (1 - p being the probability of delaying transmission by τ seconds). The parameter p will be chosen so as to reduce the level of interference while keeping the idle periods between any two consecutive nonoverlapped transmissions as small as possible. This gives rise to the p-persistent CSMA, which is a generalization of the 1-persistent CSMA. More precisely, the protocol consists of the following: the time axis is finely slotted where the (mini) slot size is τ seconds. For simplicity of analysis, we consider the system to be synchronized such that all packets begin their transmission at the beginning of a (mini) slot. Consider a ready terminal. If the channel is sensed idle, then: with probability p, the terminal transmits the packet; or with probability 1 - p, the terminal delays the transmission of the packet by τ seconds (i.e., one slot). If at this new point in time, the channel is still detected idle, the same process is repeated. Otherwise, some packet must have started transmission, and our terminal schedules the retransmission of the packet according to the retransmission delay distribution (i.e., acts as if it had conflicted and learned about the conflict). If the ready terminal senses the channel busy, it waits until it becomes idle (at the end of the current transmission) and then operates as above.

To me, the paper seems clear, but I agree that finding the channel idle and deferring the transmission doesn't make much sense to me. Do you have any other source? Or do you think there is something else in that section of the paper making the interpretation different? Indeed, in the paragraph above my quotation, they discuss the problem of all stations contending for the channel at the end of the transmission period in the case of the 1-persistent CSMA:

The above 1-persistent and nonpersistent protocols differ by the probability (one or zero) of not rescheduling a packet which upon arrival finds the channel busy. In the case of a 1-persistent CSMA, we note that whenever two or more terminals become ready during a transmission period (TP), they wait for the channel to become idle (at the end of that transmission) and then they all transmit with probability one. A conflict will also occur with probability one! The idea of randomizing the starting time of transmission of packets accumulating at the end of a TP suggests itself for interference reduction and throughput improvement. The scheme consists of including an additional parameter p, the probability that a ready packet persists (1 - p being the probability of delaying transmission by τ seconds).

The bold sentence seems to indicate that this randomized process should take place at the end of a transmission period. Yet, I find the overall description terribly ambiguous.

In addition, I've found an IEEE tutorial that makes reference to the original paper and states exactly the "defer on channel idle" mechanism.

Best,

Michele

193.205.210.74 (talk) 14:26, 16 October 2018 (UTC)