Channel 11 can NEVER interfer with Channel 4 as your article suggests.  Look at the picture you even have.  Channel 11 limits can not interfere with Channel 4.  Channel 11 does not even interfer with Channel 6.  The common triple channels to use are 1, 6, and 11 because the can not and do not interfere with each other.  Get it right dude..

Wireless Smart - take the time to read what he actually said.  He’s off in a few other places, but he’s not as wildly off on this as you suggest.  Sideband harmonics do cause a low level of intereference - all of those bumps to the side of the signal are kind of like echoes (gross over-simplifaction).

Things to correct:

1.  In an isolation chamber you won’t ever get anything close to 54Mbps.  The differences between the theoretical and the actual are the bits used for frame headers, signaling, forward error-correction and the “whitespace” between packets used to avoid collisions.  Perfect conditions won’t be more than around 27Mbits/second for 11g.

2.  11n is actually relatively stable and the market is rallying around it.  Draft 2.0 11n doesn’t have too many interoperability problems, the problems revolve more around the maturity of the chipsets.  Cisco has been shipping it’s 11n draft2 AP (the 1250) for a couple of months now.

3.  The real reason to wait for a bit to implement 11n is to switch over to the 5 GHz band.  No microwave, bluetooth, video-camera, or other common interferers.  There can still be some phone interference (5.8GHz phones), but it’s easier to work around since there are WAY more non-overlapping channels available.  Consumer embedded gear doesn’t generally do 5GHz though, so you’d end up having to use bridges today.  Just about any laptop you buy with 11n with use 5GHz though.

The sideband spread from a 802.11 spread spectrum single is large as you can see on the Metageek spectrum analyzer.  The 801.22 spec has a dB down limit on the band edges, I can’t remember but I think it’s 30 dB.  But if a laptop is close to a WAP on a channel it’s not listening to, and trying to receive a weak signal from a WAP on a non-overlapping channel, it’s bit error rate will suffer.  Very hard to tell how much.

T.H. is right on the 54 Mbps not being achievable.  I should have set the chart to something lower (maybe 45 Mbps or so) but 54 is the only reference point I could be absolutely sure 802.11g could NOT exceed!

802.11n is making progress, but my reluctance is based more on professional practices than technical issues.  Is it OK to install non-standard, still proprietary infrastructure without consulting with the home owner?  I don’t think so.  As the article says, if the customer knows the issues and want to “experiment”, fine. 

I DO NOT recommend using anything in the 5.x GHz band.  Attenuation in the home is just too high.  I’ve tried it.  It’s 2-3 times higher attenuation than 2.4GHz.  OK for open offices.

Why did everyone get rid of 900 MHz phones? They are impossible to find, and work much better than 2.4 ghz, especially in the winter when our Minnesota microwaves are heating up soup nonstop.

Julie -
Darn good question!  I found one (902 mhz) at the Good Will and snapped it up!  Now you can’t even find a 2.4 GHz phone, they have all gone to 5.2 GHz.  BUT, What you REALLY want is one that uses DECT (1920 - 1930 MHz).  Doesn’t interfere with anything and it’s a compromise on range (better than 2.4, not as good as 902).

Grayson -
I was specifically talking about 5GHz with 11n, not 11a.  MIMO makes a huge difference in extending range on 5GHz, and many of the other performance enhancements in the 11n MAC will make it maintain high performance in a home environment as well as an office.  You can also use 40MHz wide channels easily because all of the 5GHz channels are usable, unlike 2.4GHz.  The main issue is the lack of support in devices other than laptops, which is why I said that’s the reason to wait.

It’s not “non-standard”, and it’s not “proprietary” if you use a draft-2 product.  WiFi Alliance is certifying draft-2 products and performing interoperability testing, and every major enterprise AP vendor has announced a draft-2 product.  Two are shipping (Cisco and Meru) and the other are close.  Intel Centrino 4965 is draft-2 compliant and available as a cheap option on just about any laptop you buy.  Apple has been selling 11n cards for 9+ months now, and they became draft-2 compliant with a simple driver upgrade.  On top of that, the backwards compatibility in 11n is extremely robust, and you even get a range/reliability bump on legacy devices if you use it.

I am interested to know if you did any specific testing of Encryption and its effect on range or thorough put?

gtucker-
Good question.  Encryption (WPA or WPA2) adds to the packet size (about a 20-30% overhead).  Therefore the throughput decreases regardless of interference or range.  In theory it should be more susceptible to errors simply because there is more bits that have to arrive successfully.  Range for a given bandwidth should also decrease.
But to answer your question, I have done a little testing of using encryption on data rate vs. range.  I tried measuring throughput with encryption on and off.  I did see a little difference, with encryption on, troughput decreases with encryption.
Something I haven’t had a chance to try is to set up a situation where reception is “flakey” and turn encryption off to see if things get unflakey.  I hope I can remember to do this sometime soon.  Let me know if you have any results.
G.

related studies about wifi please!