As soon as one starts talking...and it's worse when they don't support 802.11h features such as TPC. :) Client interference will be worse than AP interference in most cases because the client devices always transmit on omni antennas.
Obviously when they start talking, but at which point does the clients interrupting each other significantly become an issue?
And how would one quantify that so that one could figure out how many people could be reasonably supported?
The client STA's talking are of course an issue to be aware of, but again this comes down to the type of environment you are in.
What direction is most of the traffic flowing? In a public Internet environment it is about 80/20 in favor of download. Rog, do you know what type of data will be on the network?
Oh, and one other point to make, the client STA's won't necessarily interfere with the other AP's if the AP antennas are designed correctly. However, client STA to client STA will be an issue, but again, knowing the traffic type would be helpful.
Unfortunately, I haven't been given any real data in these situations, but I'm assuming that given these are large tech conference (e.g. CES) or press sorts of settings (e.g. SuperBowl press room), surfing will take the bulk, as well as constant POP/IMAP access on some users and Youtube-ing by another set of users.
Given that, I think your 80/20 ratio is a good assumption here.
GTHill> 256 users of what per AP? Ping? Maybe...
>No real data... it has been proven many times check this out http://www.merunetworks.com/news/press_releases/index.php?articleID=081808
As suggested by Devinator read the Novarum report
GTHill> The protocol doesn't break at any user number
> yes and no the problem is the collisions, the more you have users the more you have collisions, the more you have to re xmit the packets and your bandwidth goes down, down ....
Meru, Xirrus> the best choice for density
Ruckus> it's a toy doesn't scale to enterprise level
Aerohive> just a concept that works ok in the lab but they haven't shown any decent enterprise deployment
Aruba, Cisco Trapeze etc.. when it comes to density they are limited by their MAC implementation no more than 20 users per AP as truly said by GTHill ;) and they can't manage co-chanel interference
The only thing I can add at this point is AP density with directional antennas. You can control the radius of the signal more. I wasn't aware that you could provide access to 25 users per radio, I was under the impression it was per AP (thanks for the enlightenment).
Also another thing I tried was limiting the users connection rate to 12 Mbs to avoid bandwidth leeches. And disabling the 1, 2, and 5 mb connection rate on the 11b radio.
As earlier mentioned the Xirrus product might be the solution howbeit a bit costly, but something I think could be of great use...
I am going to explore using this same type of solution in some larger conference rooms on a low power setting as a test bed ....
I too would disable those legacy ( clause 18 ) 802.11b rates except for VoWifi...
I know it has been a while since your original post to this thread - but I am curious if you (or anyone here) has had hands-on success with any of the solutions mentioned in a high-density, trade-show application. I am faced with providing Internet to a trade-show, 100,000 sq. ft. with estimated 1,500 to 2,000 clients fairly evenly distributed across the floor. Use will be basic surfing, email, and the occasional video stream. I do have the approval of the promoter to introduce limitations on per-user bandwidth or other helpful conditioning. Internet pipe is big and will not be the issue.
Genius! Genius Genius! (Mr. Deeds, Adam Sandler version)
Xirrus! Xirrus Xirrus!
I know it sounds weird, but to accomplish what you are trying to do take directional antennas, not omni's. The problem with omni antennas, even at 1mW power, is that they still propagate a long distance in free space. Actually, I'm feeling froggy so here is the math to back it up.
Free space path loss calc:
- 2.4 GHz
- 0dBm TX power (1mW)
- 200ft distance
Results: -76dBm at 200 ft range
What this means, is that if you have an AP transmitting at 1mW (0dBm) it will provide -76dBm at 200 ft. Almost every client on the planet can connect at -76dBm.
Sooo... the only want to control the problem, RF propagation and co-channel interference, is to use directional antennas. You could do this yourself with a bunch of AP's and directional antennas pointed around the room with 1mW (or less) TX power, or you could buy a system that has already done this, Xirrus.
GTHill, thanks for the reply. I was thinking either a common channel architecture (Meru) or Xirrus, based on all of the information out there. I will definitely contact Xirrus, and see what they can tell me.
I am still interested in "real-world hands-on" experience. Most everything I am reading is either ideas posted in threads (although most based on solid engineering principals) or propaganda / white papers from the manufacterers. I believe that an independent VAR or integrator who can say that they have successfully done this one or more times is the most valuable info!