I looked into this about 18 months ago for our campus. It never made it to the point of a trial: I learned enough to stop the project before it made it that far, and I think I can summarize here what I found.
I'll start by going back to basics: we all know that wireless channels overlap. A graph of signal from a wireless access point typically takes the shape of a parabola, or cone, with the open end pointed down and centered on the channel used by the access point. An example can be found at the following link found via quick Google Images search: http://www.bestandroidappsreview.com/wp-content/uploads/2010/04/Top-Android-App-WiFi-Analyzer-Signal-Graph.png That image (from an app I'd used with great success on iOS before it was pulled from the Apple App Store for using undocumented APIs) clearly shows overlapping signals. The wider portion as a signal flares out towards the bottom is often referred to as a signal's "skirt"... this term "skirt" will be important later. The common best practice to avoid overlap (and thus avoid interference and improve performance) is to only use channels 1,6, and 11. That much I think everyone here understands very well. Now let's move on to a 4-channel scenario. If you put four access points right up next to each other on channels 1,4,8,and 11, you *will* have interference. Channel 1 signals will collide with Channel 4, 4 will collide with 1 and 8, 8 will collide 4 and 11, and 11 will collide with 8, resulting in reduced performance throughout the spectrum. This is also not in question. But what if you separate these four access points... put some distance between them? Simplistic graphs such as from my earlier link imply that as the power level of the signal falls over distance you will have a shorter and therefore narrower "skirt". Could careful planning allow you to place access points so that channel 1 APs are never near channel 4 APs, 4 APs are never near 1s or 8s, 8s are never near 4s or 11s, and 11s are never near 8s, and in this way increase AP density beyond what you could do with only three channels, all while still avoiding interference? The short answer is "no". It comes down to the skirts again. Most low-end tools to measure wireless coverage do a poor job of showing this, but my understanding is that wifi RF is such that the skirts "flare out" quickly, and you have nearly all of the signal overlap even at fairly low power levels. These wide skirts makes it impractical to try for four channels... you're almost as bad off as if you tried to use all eleven. Joel Coehoorn IT Director York College, Nebraska 402.363.5603 [email protected] On Tue, May 8, 2012 at 11:01 AM, David Gillett <[email protected]>wrote: > ** > Our pilot deployment included four APs in a single fairly-small > building. If I recall correctly, I put the two in the middle of the > building on channels 1 and 11, with the two further out, one on ch8 > (nearest the AP on ch1) and one on ch4 (nearest the AP on ch11). I'm > pretty sure these were only doing 802.11b, so even where the interference > was low, the performance was modest, and nobody yet expected anything > better.... Essentially, I tried to take advantage of physical separation > where I couldn't rely on channel separation. > > (These days, we use Aruba, and generally let it try to find a selection > of channels for minimal interference.) > > David Gillett, CISSP CCNP > > > ------------------------------ > *From:* Lee H Badman [mailto:[email protected]] > *Sent:* Tuesday, May 08, 2012 07:34 > *To:* [email protected] > *Subject:* [WIRELESS-LAN] 4-channels in 2.4 GHz > > With no intent to open a conversational can 'o worms, I'm curious if > anyone is running a 4-channel plan on their production WLANs, that is > willing to share their opinions and experiences on the topic. > > Thanks- > > Lee > > Lee H. Badman > Wireless/Network Engineer, ITS > Adjunct Instructor, iSchool > Syracuse University > 315.443.3003 > ********** Participation and subscription information for this EDUCAUSE > Constituent Group discussion list can be found at > http://www.educause.edu/groups/. > > ********** Participation and subscription information for this EDUCAUSE > Constituent Group discussion list can be found at > http://www.educause.edu/groups/. > > ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/.
