Realistic Number of Clients from one 802.11n AP

I have my news reader set to not display the subject or the rest of the header glop. Make my life easy and instead of wasting bytes with "the subject says it all", just type or cut-n-paste the subject.

In theory, yes. In reality, no.

802.11n is all about speed (not range). The faster you get a users data transfered over the air, the more users can share the same air time. However, that assumes that all users will be running at speeds above the fastest 802.11g speed of 54Mbits/sec. That's reather problematic as the typical maximum range for a 54Mbit/sec association is about 8-10 meters (using a typical laptop). Most systems can't maintain anything even close to 54Mbit/sec.

I've been monitoring connection speeds at a local coffee shop. When there is only one user moving traffic, typical connection speeds at

24-54Mbits/sec. Add a few users, and interference (and possibly reflected interference) becomes a problem, resulting in a slow down. The same system now runs at about 9-12Mbits/sec.

Ummm... think again please. If everyone ran at 1Mbit/sec connection speeds, it would take 50 times as many packets (or 50 times as much air time) to deliver the same content as a 54Mbit/sec connection. If you want capacity (users * bandwidth), then you have to get the traffic off the air as fast as possible, which means higher speeds.

I don't know. I don't judge popularity contests.

I'm also not a huge fan of 802.11n. It does improve resistance to reflections, but I haven't seen much of a speed gain in real applications. The typical coffee shop environment that I see if full of MacBooks with the latest 802.11n technology from Apple. I installed an Airport Extreme (with 802.11n enabled) at one location for a day to see if it would make much difference in capacity. Nope. Users were still connecting at 9-12Mbits/sec (i.e. slower 802.11g speeds). I plan to repeat the test fairly soon, but I don't expect any different results.

If you don't mind forgetting about 802.11n for now, I'm partial to DD-WRT running on Buffalo or Linksys hardware. There are also other

3rd party firmware replacements that are useful for specific applications.

I'm curious... What problem are you trying to solve that would require

802.11n?

Sorry Jeff

OK

As most of our students have 802.11n nics now I wondered if we would gain any speed benefit for them by replacing our APs with ...n type for those who are too many sharing on AP. See the 'Share bandwidth evenly...' thread.

/geir

The fact that you will still have legacy devices, 11b/g, will cause the "N" AP's to operate at their speed as the 11b devices did to 11g AP's when they were introduced. It will probably be a while before all wireless devices are type "N" as there are PDA's, smart phones and AV devices such as iPod touch which at present only have 11b/g. When you say you wish to move from 11g to 11n I presume you were thinking of staying in the 2.4GHz band which is also not a good idea. If you wish to use a 40MHz channel in the 2.4GHz band you will take up 2/3 of the conventional 11 channels and this would effectively mean you could only use one AP with 40MHz. There are also coexistence issues with legacy AP's which could mean you would only be able to use 20MHz channels. If you are going to use more than one AP you should be looking at the

5GHz band for "N" networking and you might find it useful to replace some of AP's with simultaneous dual band AP's so legacy devices could use the 2.4GHz band and the 5GHz band could be "N" only, this would be dependent on whether the clients can work at 5GHz of course. This is a couple of years old but worth reading.

Airmagnet have a simple primer:-

Yep. More bandwidth is needed for more users which means 802.11a radios.

Xirrus has an assortement of really interesting articles on MIMO and

802.11n on their web pile.

Their approach to high density wireless is to switch to 802.11a *AND* to reduce the amount of air time used by going as fast as possible: Xirrus Application Note High-Density Wi-Fi "The best method to increase the Available Capacity is to provide the highest data rate and maximum Free Air Time."

However, at a given power level, rx sensitivity, and error rate, data speed varies roughly with the square of the distance. Double the distance, and you can only move data at 1/4th the speed for the same error rate. As you might deduce from this, high speed wireless works really well close to the access point, and really awful as you get farther away. The implication is that air time increases at the same rate as the speed. Thefore, as you get twice as far away from the AP, you consume 4 times the air time. The most common solution is to add more access points to reduce the maximum range, but that's self defeating due to mutual interference. 802.11a is a good answer with

12 non-overlapping channels available (for 20Mhz bandwidth).

You can get a really good feel for how this works by setting the access point for 802.11n ONLY and at whatever greater than 54Mbits/sec speed is available. Fire up a benchmark (i.e. IPerf or Jperf) and take a walk with the 802.11n laptop. I don't think you'll get very far. Last time I tried it with an Airport Extreme, I got about 8-10 meters. That's as far as you're going to go taking advantage of

802.11n modes to reduce air time.

Incidentally, once you get everyone to go with 802.11n for higher speeds and therefore higher capacity, you now have the additional problem of load balancing and air time sharing, which Xirrus addresses in the above article on dense Wi-Fi. For example, 20 users running at

50Mbits/sec thruput (not wireless connect speed) will saturate a gigabit ethernet backhaul. If they're all connected to the internet, they will certainly use more bandwidth than is economically available from your ISP.

Did you mean 40MHz bandwidth? I did notice that Intel were saying that only 11 channels were supported.

I can't find a dual band consumer router at a reasonable price that gets good reviews so I haven't got around to buying an 11n one. Is that airport extreme one of the simultaneous dual band ones? I did notice that there has been no price increase for that function.

I would like to know how these perform in the real world as it may become an option for me.

I count 12 for 20MHz channels. See the Intel Wireless Hotspot deployment guide illegally archived by me at:

Table 17 on page 48 lists 12 channels for the US and 19 channels for ETSI (Europe).

5.4.2.1 says: "All twelve 802.11a channels are considered non-overlapping for purpose of communications."

Same here. The only ones that have a chance are those that have totally seperate radio sections such as the 3com and Cisco products.

Yep:

Apple calls it "simultaneous dual band". That's been done before by Linksys in several products, all of which were rather slothish when running both bands simultaneously. I can also handle a USB hard disk drive plugged into the router for a file dumpster. $150-$175/ea.

Too soon for me to tell. There's only one laptop with a 5.7GHz radio and it wants to use 2.4GHz by default with no way to force it onto

5.7GHz. I didn't have the indicated problems with the Airport Utility because we did not use Time Capsule yet. So far, the only complaints have been over user interface issues. The wireless part has been flawless, but so far, only on 2.4Ghz.

Hmm, when you look at Intel's channel bonding page they have 24 non-overlapping channels which would give them 12 40MHz channels although according to my earlier link their client products only support

11 40 MHz channels.

A quick google came across this from 2003 "With the addition of 11 new, noninterfering channels to 13 existing channels in the 5GHz range, however, there will be more breathing room for both radar and wireless devices that use DFS, reducing the likelihood of conflicts, Redelfs said."

Although that was dated for 2003 it looks as though the channels weren't opened for use until 2006 for the US and 2007 for Japan, see table 1:5 on page 4.

Y'er right. 24 channels, 20MHz wide, on 5.7GHz. My bookmarks and Googling turned up all the pre-DFS channel juggling charts. Nice article from Atheros. Thanks.

There is a memorandum opinion and order document from June 2006

The Atheros article did mention that the DFS was going to be revised for ETSI, apparently there is a problem with the present devices not recognising Meteorological Radars operating in the 5600-5650MHz band and causing interference, so changes are to be implemented.

If you wish to waste some time the ETSI v1.5.1 draft doc is at