El jue, 04-11-2010 a las 20:04 +0100, Jonas Smedegaard escribió: > On Thu, Nov 04, 2010 at 02:22:18PM +0100, José L. Redrejo Rodríguez wrote: > >We're currently testing this solution that covers the items you mention > >too, and it's 1/10 of the value of a good AP. > > I guess you mean the opposite: that the solution you are testing > provides _same_ value while 1/10 _cost_ of a good AP :-) > > of course, my spanglish mixed value/costs words ;) > Really appreciate your detailed real-world experiences, José! > > > >Anyway, you've forgotten the main difference between a school and a > >hotel/restaurant setup: the number of concurrent users. Think of a > >school with only 10 classrooms, 25 pupils per classroom: 260 laptops. > >As soon as most laptops are in the school, you will soon run out of > >available wireless channels, and available bandwith. That's why only > >expensive AP working in 802.11n work (we haven't tested enough the > >hostap solution), but I can confirm you that ~100 euros AP's will not > >work as soon as the number of users increase. > > > > > >And, for sure, forget about using 802.11b/g and, if possible, use > >802.11n with dual band support and work in the 5GHz band, where you can > >get enough real available channels. > > > Interesting point about amount of users essentially flooding the radios. > > > This issue is of concern for the One Laptop Per Child project too, and > its sister-project, Sugar, too. I recall it being discussed recently > (2-4 months ago, I believe - tell me if anyone wants me to locate it > more exactly) and a proposed solution was to turn *down* the power of > the radio chips, so as to cover smaller areas per AP, instead of wasting > radio bandwidth by overlapping too much. If I recall correctly, the > proposal originates from the german Freifunk project which apparently > have had success with setting up AP software to dynamically lower the > radio power to the least needed depending on some usage patterns. That's very interesting, we have also thought of reduce the power of the AP and the laptops radios. A dynamic solution would be fantastic but I don't know how it can be done. What parameters could be taking to calculate the needed power? Do you have a link with more information about the possible algorithm? Cheers.
Attachment:
signature.asc
Description: Esta parte del mensaje =?ISO-8859-1?Q?est=E1?= firmada digitalmente