in 802.11, when must a station go to the "random fall back slot" mechanism?

Question

i've been reading Michael Gast's 2005 book on 802.11, and he talks about how each station, after it has seized a channel and finished transmitting, will then have its NAV timer expire (along with all the other stations in range). Then it, along with the other stations, will randomly choose a slot, and whoever has the lowest slot will be the next station to transmit.

he also writes how stations can fragment data and continually refresh the NAV timer so they maintian control of a channel until the final fragment is sent. NAV timers have a max of 32,000~ microseconds, but can be refreshed through "non-final" fragments, for additional 32,000 microseconds (if i'm reading correctly)

So is there a limit to how much a station can transmit before it goes into the fallback slot mechanism?

for example... if there's 5 stations, and all of them want to upload 5 gigs... what prevents a single station that wants to upload 5 gigs to the internet through the AP from just grabbing the channel and holding onto it until all 5 gigs are uploaded?

Answer 1

I believe the concept you're looking for is a Transmission Opportunity (TXOP). Introduced in the 802.11e amendment, it provides a station a preset duration for transmitting and receiving frames. For different access categories, different TXOP durations are allocated to achieve the required QoS. At the end of the TXOP, the station has to again contend for channel access, so that fairness is maintained between all the stations (this is what ensures that a single station does not grab the channel for its entire transmission).

For older standards like 802.11a/b/g (or a TXOP duration = 0), only a single frame can be sent (along with the reception of the corresponding 802.11 ack from the recepient) for each time channel access is granted. So in this scenario, if you're sending 5 gigabits of data and your packet size is ~2000 bytes, you would need a staggering number of channel access attempts.