Rony,
Several comments on 2196:
1. why can the STA not use an intra-BSS PPDU to start the backoff counter?
( i can guess why, but I believe that your rationale is flawed )
NOTE that the end of ANY BUSY activity on the medium sets up the next set of slots for all STAs on that channel
(provided, of course, that everyone on the channel saw that same BUSY condition, and there is no guarantee of that)
(but to my guess, there is no guarantee that every NPCA STA sees every intra-BSS PPDU either)
(and don't forget about EIFS!)
(and don't ignore collision recovery!)
(All of these ensure that slot boundaries are never uniform)
But again, if you want to at least try to get some of the STAs lined up, then
you do not have to receive a PPDU or even a header of a PPDU to synchronize:
any type of medium BUSY will cause a STA to line up a new set of slot boundaries
and any other STA that sees that same BUSY will be lined up to the same slot boundaries
And there is nothing that you can add to the standard to make certain that everyone sees the same BUSY
(Well, there is something that you can try to do, but it is extremely inefficient - you can have the AP be the one guy that always declares the end of a BUSY -
but that does not work well either because then STAs might be ignoring local BUSY conditions that the AP did not see,
plus, that only definitively aligns intra-BSS STAs, and OBSS STAs are competing on the same medium - slot alignment is important for all competitors, not just myBSS
the correct thing to do is allow ANY BUSY condition to allow the backoff to start
again, this will not "fix the problem" that you have identified
because there is never a fix to that problem because every STA has a different view of the medium
so every STA will see different BUSY to IDLE transition times depending on which BUSY events they observe
and every STA will have different slot boundaries
and this is the situation today, on the non-NPCA channel
and it always has been the situation
certainly, I appreciate the attempt to make it better where NPCA might have introduced a new mechanism to create disparate slots boundaries
sure, in a situation where the NPCA channel is pretty quiet, then this might help
But even in that case, your rules are WAY TOO RESTRICTIVE - change to ANY BUSY!
2. in 37.18.4 - it is possible for a STA to not meet either condition 1) or 2), in which case, the STA is stuck, frozen, unable to count down
you need to have a third way out
of course, maybe that is what you intended - i.e. for a STA that has a delay that is larger than what the AP advertised will be punished and can only start after it receives a frame (or by my proposal, a BUSY to IDLE transition) [ i.e. the STA is not really stuck, it is waiting for a BUSY to IDLE transition ]
( I re-read the proposal and I do see that you have a bold highlighted section stating this )
3. the very thing that you point out as a problem in fact leads to implementations becoming better
i.e. an implementation that has a slow switching time will, as you state, receive lower priority, hence, the creator of that implementation is incentivized to improve their delay
with this proposal, you have removed the incentive for implementers to create better implementations
Ok, not quite - you have simply given the AP the ability to decide who the winners and losers will be by letting the AP set the cutoff point and then, actually, you are increasing the punishment, by not allowing a slow STA to participate at all, until it sees an intra-BSS PPDU - pretty harsh
To unsubscribe from the STDS-802-11-TGBN list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBN&A=1