There are many aspects to managing unlicensed Wi-Fi spectrum to avoid interference and optimise wireless LAN application performance. Interference can be caused by traffic butting up against other traffic in overlapping channels, another operator’s 802.11 devices contending for your spectrum, non-802.11 devices operating in the spectrum, and environmental factors blocking or degrading signals, to name a few.
A less-obvious interference culprit is the “hidden node.”
Hidden nodes are basically client devices that are all within range of the WLAN access point (AP) but are not necessarily within range of each other. Picture an AP in the middle of a circle. Client A is at 9 o’clock, 50m from the AP. Client B is at 3 o’clock, 50m from the AP. The distance between the two clients – or the diameter of the circle – will be 100m (328 feet).
As you likely know, the 802.11b/a/g suite of standards uses a media access control (MAC) mechanism called carrier sense multiple access with collision avoidance, or CSMA/CA. Client nodes more than 300 feet apart are not likely to “hear” each other transmitting in order to avoid a collision. Two nodes transmitting on a common channel at once causes collisions, which results in interference and lowered throughput and response times.
Tools to detect the problem
Some of the Wi-Fi RF monitoring and management products can detect hidden nodes. AirMagnet, for one, just announced this capability in the latest version of its laptop analyser product, Laptop Analyzer 7.0 PRO. If performance is suffering, of course, one of the first things you have to do is find the source of the problem, so this type of capability can be a handy tool.
What do you do?
Having a very dense AP deployment tends to help this situation, because the overlapping “circles” of coverage are smaller, which means their diameter (or distance between client nodes) shrinks, and nodes are more likely to sense one another. If using the 2.4GHz band, though, your circles can only be so small before the three nonoverlapping channels will interfere with the nearest APs to use the same band.
Once the situation is detected, one option is to increase the power levels in the client nodes, so they are more likely to see one another. This won’t help, however, if there is something in the way – such as a cement or steel wall – obstructing the devices’ signals. This, of course, is something to account for in your upfront site survey.
You might also consider replacing directional antennas with omnidirectional antennas. This will let the antennas sense in 180 degrees, delivering a greater chance of the client’s picking up another’s signals. Doing so has to jibe with your security strategy; you may not wish to have WLAN signals spilling outside of the building’s walls in all directions.
Note that the 802.11 request to send/clear to send (RTS/CTS) protocols help. These protocols send some clients permission to transmit while asking others to wait a time period, in an attempt to coordinate transmissions. However, given the dynamic nature of wireless networks, and as the number of clients increases, this isn’t foolproof.
The 802.11e QoS standard, meanwhile, now also contains protocols for call admission control, a stronger traffic coordination function aimed primarily at giving VoIP QoS. It’s possible that this mechanism can be used for further coordinating client-to-AP transmissions for other traffic, as well.