The WiSE Article Series: Wi-Fi Subtleties Explained (Channel Bonding)By CWNP On 02/14/2013 - 70 Comments
In this third installment of the WiSE article series from AirTight Networks, channel bonding is considered. Some surprising results will cause you to rethink your network design plans and possibly how you will implement newer 802.11 technologies.
Guest Blogger: AirTight Networks Author: Bhaskaran Raman, PhD. Series Editor: Tom Carpenter
About the WiSE article series: Wireless is inherently complex; its study spans at least two engineering disciplines: Electrical Engineering and Computer Science. Add to this the nuances of various standards, vendor implementations, RF environments, and protocol interactions, and it is not uncommon to feel a little lost in understanding the various aspects of Wi-Fi network operation. In this series of short articles, we explain various Wi-Fi subtleties, to work toward a better understanding of Wi-Fi network deployments.
WiSE Article No.3 Wi-Fi Throughput Measurement - Subtleties in Channel Bonding
An important parameter affecting the Wi-Fi client throughput in the 802.11n protocol is the channel width of operation; the second WiSE article explained that there can be as much as 30% improvement in the throughput with the use of 40 MHz channels compared to 20 MHz channels. This article explains three subtleties regarding 40 MHz channel operation.
The option to use 40 MHz transmissions in the 802.11n protocol is called channel bonding. Two adjacent 20 MHz channels are “bonded” together to make one 40 MHz channel. Of these, one is known as the primary channel of operation. The secondary channel must be immediately above or immediately below the primary channel. Several commercial APs and clients have a configuration parameter to turn on this channel bonding feature.
Subtlety-1: 40 MHz Operation Cannot Be Forced
The first subtlety is that the 40 MHz operation cannot be forced through configuration. The 802.11n standard mandates that an AP (say A) can support 40 MHz operation (transmission or reception) only after a scanning procedure to check for nearby APs, and only when a set of stringent conditions are satisfied by the scan result. These conditions are:
- Any other AP X detected by A as supporting the 40 MHz operation should have the same primary channel as A, and also the same secondary channel as A.
- Any other AP Y detected by A as supporting only 20 MHz operation should be operating on the primary channel of A.
- The AP A should not have received a “20/40 BSS Intolerant Channel Report” citing any channel other than its primary channel in the report.
All of the above conditions must be satisfied before the AP A can use 40 MHz transmission/reception. In fact, in the 2.4 GHz band, the scan to check for the above conditions must include all channels whose center frequencies are within 25 MHz, not just 20 MHz, of the center of the chosen 40 MHz range. The conditions as well as the width of channel scan required are explained in the figure below.
[caption id="" align="aligncenter" width="776"] Channel Scan Conditions[/caption]
So effectively, 40 MHz operation cannot be forced through configuration; although the configuration parameter wording may say “40 MHz”, it really means “40 MHz if possible under the standard-specified stringent conditions”. In fact, the 802.11n standard refers to the 40 MHz BSS operation as 20/40 MHz operation, and a fall back to 20 MHz transmissions is possible at any time, in any 802.11n compliant device.
Subtlety-2: The 2x transmission range effect
The second subtlety is that an AP A may be forced to switch to 20 MHz operation by an AP X which is well outside of A’s transmission range! This is because, in doing scans to check for the above-specified conditions, the AP A may ask its associated client C to perform the scans on its behalf (the standard allows this). And in a typical deployment, it is very well possible that C detects AP X which is outside of A’s range, as depicted in the figure below.
[caption id="" align="aligncenter" width="740"] Client Scan Option[/caption]
Subtlety-3: 20 MHz operation is more efficient!
The third and final subtlety in this article is the most non-intuitive. In the clamor for higher single-client throughput, and thus 40 MHz operation, it is easy to miss that 20 MHz operation is actually more efficient from an overall network performance perspective.
An analogy helps us understand this effectively. Suppose a car manufacturer comes up with the design of a wide-car, which is twice as wide as a normal car. If your commute on a busy morning is on a congested highway, the figure below is enough to convince you that you wouldn’t want to see any of these wide cars near you. It is also clear from the figure that the use of wide cars is actually quite inefficient from a road usage perspective, during busy periods.
Taking this analogy to the 802.11 protocol, when you use 40 MHz channels, the carrier sense backoff mechanism has to deal with interferers in the entire 40 MHz band. So on a given channel access opportunity (Tx opportunity) you may transmit more, but opportunities will come less frequently due to interferers blocking your transmission more often. Also the chances of collisions are higher.
One can now see why the standard specifies such stringent conditions for the 40 MHz operation; the conditions are somewhat akin to ensuring that the wide-cars, should they exist, are used only when the road is free, and not during busy periods.
[caption id="" align="aligncenter" width="514"] Narrow Lanes vs. Wide Lanes[/caption]
- Channel bonding in 2.4 GHz: The popular 2.4 GHz band has 3 non-overlapping channels (in N. America and many other countries). As the wide-cars on a 3-lane highway analogy indicates, 40 MHz operation on the 2.4 GHz band is almost never possible, and rightly so from a network efficiency perspective, as the 2.4 GHz band resembles a busy highway with peak-time traffic.
- Channel bonding in 802.11ac: The upcoming standard 802.11ac specifies channel bonding up to 160 MHz, in the 5 GHz band of operation, with the requirement that 80 MHz chunks have to be contiguous. The line of reasoning in this article would apply to channel bonding in 802.11ac as well, with the exact implications becoming clear only after the standard has been specified and released.
Bhaskaran Raman is a scientist at AirTight Networks working on high performance Wi-Fi architecture. Bhaskar received his B.Tech in Computer Science & Engineering from IIT Madras in May 1997 and his M.S. and Ph.D. in Computer Science from University of California, Berkeley, in 1999 and 2002 respectively. He was on the faculty in the CSE department at IIT Kanpur (India) from 2003-07. Since July 2007, he has been a professor at the CSE department at IIT Bombay. His research interests and expertise are in communication networks, more specifically in wireless & mobile networks.Tagged with: Wi-Fi, WiFi, 802.11, tom carpenter, throughput, airtight networks, channel bonding, Bhaskaran Raman