Cisco Cisco Aironet 3700i Access Point 白皮書

 

 

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

Page 9 of 10 

A device, typically an access point (more on this to follow), sends a Very High Throughput (VHT) Null Data Packet 

(NDP) Announcement frame whose only purpose is to contain the address of the access point (the beamformer) 

and of the target client recipients (the beamformees). The VHT NDP Announcement frame is immediately followed 

by a VHT NDP intended for those target recipients. Each intended recipient measures the RF channel from the 

access point to itself using the preamble of the VHT NDP and compresses the channel. The first intended recipient 

responds immediately with the compressed channel information in a VHT Compressed Beamforming frame. Other 

recipients respond subsequently when they are polled by the access point. The VHT NDP Announcement frame, 

the VHT NDP, and the VHT Compressed Beamforming frame are all similar to features in 802.11n. However, 

because of some subtle differences, the 802.11ac sounding is not backward compatible with 802.11n devices. 

ECBF is known to provide the most precise estimate of the channel that takes into account all the imperfections at 

transmitter and receiver. 

However, ECBF comes with a lot of overhead: the VHT NDP Announcement frame, the VHT NDP itself, and the 

frames carrying the compressed feedback from each recipient. For an access point with four antennas, the 

compressed feedback varies from 180 to 1800 bytes, depending on the number of client antennas and level of 

compression. Sounding just one single-antenna 80-MHz client takes about 250 microseconds. When devices can 

transmit at 433 Mbps, this has an impact, since that same time could have instead been used to send an extra 

13,000 bytes. 

Also, although any device with multiple antennas can beamform to any other device at any time, the practical 

limitation on the effectiveness of ECBF is that the target recipient ideally has fewer antennas than the device 

intending to beamform. Since many client devices have only a single antenna, they cannot beamform to the access 

point, and those clients that have multiple antennas are at a disadvantage, since the access point typically has 

more antennas. 

Therefore, technologies that solve the problem of sounding without being limited by client implementation and/or 

depend on client assistance (such as Cisco's ClientLink technologies) continue to add genuine value. They help (1) 

legacy 802.11a/n clients, (2) those 802.11ac clients that do not support 802.11ac sounding, and (3) clients at 2.4 

GHz. They also avoid the overhead of standards-based explicit sounding. 

Introducing Cisco ClientLink 3.0 Beamforming 

802.11a/g clients cannot support standards-based beamforming, either explicit or implicit, and many 802.11n and 

802.11ac clients do not support standards-based beamforming either. For this reason, it is vital that vendors aiming 

at a comprehensive solution provide beamforming modes that work for any client. As with Cisco ClientLink and 

ClientLink 2.0, ClientLink 3.0 beamforming does just that. 

ClientLink was designed for single-antenna clients or clients using one spatial stream. The access point can 

measure the wireless channel any time the client transmits, even if 

it’s just one packet, and the access point then 

uses that information to maximally reinforce the data send-back to the client. For multiple-antenna clients, vendors 

must use innovative technology such as ClientLink 2.0. and 3.0. Single-antenna clients also benefit from ClientLink 

2.0 and 3.0.