Cisco Aironet 2702i AIR-CAP2702I-E-K9 Folheto

 

 

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

Page 12 of 25 

Second, before the device addressed by the replicated RTS responds with a CTS, the recipient device checks to 

see if anyone is transmitting near itself, on its primary channel or on any other 20 MHz within the 80 MHz. If a 

portion of the bandwidth is in use nearby, the recipient responds with a CTS only 

on the available and “usable” 

20-

MHz subchannels and also reports the bandwidth of the replicated CTS inside the CTS’s PPDU. Here “usable” 

subchannels means the subchannels on which the initiating device is allowed to send something, such as a 20-, 

40-, or 80-MHz (but not 60-MHz) transmission. This is shown in 

. 

Third, the CTS is sent, like the RTS, in an 802.11a PPDU format, replicated in 20-MHz chunks across the available 

and useful bandwidth. Again, every nearby device receives a CTS that the device can understand on its primary 

channel. 

There are other variations on this protocol, for when the initiator is incapable of switching to a narrower bandwidth 

on the fly and so forth, but the previous description captures the essence of the enhancement: the recipient can 

say, 

“These subchannels are busy - don’t use them.” 

802.11 defines that every 802.11 PPDU transmission is an A-MPDU, yet the A-MPDU might contain only a single 
MPDU. Why? The short answer is that it’s complicated. 

Here’s the long answer: There are three reasons: (1) In 802.11a/n, the duration of the transmission is set by the 

number of octets and the data rate for the transmission. But a maximum-length 5.5-ms transmission at 6.93 Gbps 

could contain over 4 million bytes, and this takes 23 bits to represent. These bits would be sent at the lowest 

Modulation and Coding Scheme (MCS) rate at the start of every 802.11ac transmission and so practically would 

add 4 microseconds each time. Instead, the length of an 802.11ac transmission is constrained to be a multiple of 

the number of data bits per orthogonal frequency-division multiplexing (OFDM) symbol, and then only the number 

of OFDM symbols needs to be signaled. Moreover, the number of (assumed to be) 4-microsecond-long OFDM 

symbols is already implicitly available in the legacy portion of the preamble, so this signaling comes almost for 

free.

2

 Then we need a way to completely fill even the last OFDM symbol with data. A-MDPU makes this easy: send 

the data as MDPUs within MDPU subframes in an A-MDPU, then pad the A-MDPU with enough null MDPU 

subframes to fill up the last OFDM symbol. (2) This same padding mechanism will come in handy for the new 

MU-MIMO feature. (3) A-MDPU is in general a good idea to increase reliability for long payloads. 

802.11ac adopts a keep-it-simple approach to channelization. Adjacent 20-MHz subchannels are grouped into 

pairs to make 40-MHz channels, adjacent 40-MHz subchannels are grouped into pairs to make 80-MHz channels, 

and adjacent 80-MHz subchannels are grouped into pairs to make the optional 160-MHz channels, as shown in 

. A BSS (that is, AP plus clients) uses the different bandwidths for different purposes, but the usage is 

principally governed by the capabilities of the clients. 

                                                 

2

 Just a single bit is needed to disambiguate the number of actual OFDM symbols present if the transmission instead uses the 

short guard interval and the OFDM symbols are actually 3.6 microseconds long.