It's a rich topic and the results depend on the AP, the client & client density, the environment and test methodology.
There is an extensive set of benchmarks in the Aruba Hign Density Design Guide in the section title "HD WLAN Testbed." It's on the web here: http://www.arubanetworks.com/pdf/technology/DG_HighDensity_VRD.pdf
While the document title suggests it applies to (only) high density designs, the peformance tests are really informative and start with low client counts on the AP. There is an extended discusison of factors that drive performance/capacity.
Cisco also has performance guidlines that start at low user density in their High Density Design Guide. It's on the web here: http://www.cisco.com/en/US/prod/collateral/wireless/ps5678/ps10981/design_guide_c07-693245.html
Cisco's design guide adds a discussion of antenna selection in high density venues but provide less background on test methods and performance claims.
Both documents are both good (but not quick) reads.
I'd recommend those design guides also.
This test is interesting too.
It can help you push 5GHz. Look at the high to N clients when a G client enters the area!
I don't get the half duplex means half the throughput comment. You aren't sending and receiving the exact same about of data. I am of the thought the slow down is from the overhead needed in the error correction code
I was under the impression that there was 50% overhead due to contention protocols CSMA/CA, because of the half-duplex medium limitations. With other factors, as the Cisco design guide (above) shows, with the shared medium, environment...
From the CWNA PWO-104 text " ....(CSMA/CA) helps to ensure that only one radio device can be transmitting on the medium at any given time. Because of the half-duplex nature of the medium and the overhead generated by CSMA/CA, the actual aggregate throughput is typically 50% of the data rate. ....throughput is affected differently based on the frequency used." (though wavelength was a wrong answer in another question about throughput)
On another thread I am asking about propagation, SNR and signal level/quality affect on throughput if there is anyone who knows.
Expect 50% of the transmitted rate, whatever that is in your situation. Simple !
The part that's not simple, is calculating what the rate will be. There are usually too many variables to accurately predict that, under less than optimal conditions.
BrianW's links above are a great place to get more information. I suggest you read them.
The Cisco document that BrianW posted is pretty good. It's more focused than some of the other Cisco wireless documents I've seen, and I think better because for it.
Page 11 has an informative table showing aggregate throughput that gets to the heart of some of the other issues here.
I did find one slight error. On page 26, the same equation is given for both Inner and Outer radius measurements (for antennas with downtilt). Unless my trig is completely off, the
tangent factor for the Inner radius should be A - (BW/2), not A + (BW/2).
All in all, two really good references BrianW. Thanks for posting them.
Franman, I haven't gotten to your links yet. But I will :-)