Cisco Aironet 2702i AIR-CAP2702I-E-K9 产品宣传页

 

 

© 2014 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public Information. 

Page 24 of 25 

● 

Channel access for 40 MHz. Perhaps the fundamental reason for the success of 802.11 is that anyone 

can install an AP or use a client, regardless of what other 802.11 devices are already nearby, and mostly it 
all “just works.” 

This comes from a MAC design goal that channel access be reasonably efficient and fair to all, regardless 

of number of devices, distance to AP, device capability, and so forth - 

put succinctly as “your packet is as 

important as my packet.

” We see the efficiency goal in the range of MAC techniques to reduce collisions, 

such as physical carrier sense (don’t transmit if you hear a lot of energy) and virtual carrier sense (don’t 
transmit while someone told you they’d be transmitting or receiving). We see the fairness goal in that each 

device is permitted to transmit only after meeting the same carrier sense and collision avoidance 

requirements. 

However, 40 MHz brings real challenges to both collision avoidance and fairness, since it is either 

cost-prohibitive or impossible to maintain accurate physical carrier sense and virtual carrier sense on two 

20-MHz subchannels in parallel. Instead, 

a “primary” 20-MHz channel is defined with the usual tight 

requirements on carrier sense and collision avoidance, augmented by a degraded physical carrier sense on 
the “secondary” 20-MHz channel. When a device wants to transmit, it performs channel access in the usual 

way

4

 - all on the primary 20-MHz subchannel. Also, immediately before the device can transmit a 40-MHz 

packet, the device inspects the physical carrier sense state of the secondary channel for a short duration to 

make sure that the secondary channel is clear too. If clear, the 20 MHz packet is sent; otherwise the device 

can either (1) transmit a 20-MHz packet on the primary 20-MHz channel or (2) back off again, then recheck 

to determine whether the full 40 MHz is clear. Remarkably, this simple scheme is reasonably fair, and 

option (1) is reasonably efficient. 

Still, in some topologies, devices on the secondary 20-MHz channel are treated unfairly with respect to the 

40-MHz devices, and so 802.11n has additional channel selection rules to try to avoid this scenario in the 

first place. These rules work pretty well, given the large number of 40-MHz channels available at 5 GHz. 

802.11n was notorious in the standards community for its slow progress. There were three causes: (1) The process 

that 802.11n chose to select the winning proposal invited contention. (2) 802.11n was loved nearly to death, in that 

many experts wanted to help and to contribute their technology. It took a long time to work through the contention 

by adopting many optional modes and then a long time to refine all the optional modes. (3) Operation of 802.11 

systems at 2.4 GHz using 40-MHz channel widths in the proximity of 802.15 systems (such as Bluetooth) raised 

concerns among parts of the 802.15 community. 

 

 

 

 

 

 

 

 

                                                 

4 

That is, the device checks carrier sense and, if the channel is busy, waits until it is clear, randomly backs off a number of slots, 

and waits while it counts off those slots.