Recently I was involved with trying to determine the cause of very low throughput on a number of Wi-Fi access points, providing coverage over a large outdoor area.
We were equipped with an Anritsu S332D Sitemaster, with spectrum analyzer, and after testing the antennas we had a look at the 2.4-2.5GHz spectrum in search of any sources of interference. The location was surrounded by 3 cellular towers, with the nearest being less than 100m away, so we suspected that there may be some interference in the 2.4GHz band due to intermodulation. Surprisingly the band was very quiet, and there were no signals that did not belong there.
Next we looked more closely at the cellular frequencies, the nearest of which was centered on 2.115GHz. To our surprise the level was much higher than expected, with our low gain, wideband antenna, the received power was -20dBm! This made us think that the high level cellular signal was swamping the access point's front end receiver leading to receiver desensitization, but surely access points would have a pre filter before their sensitive receiver circuitry wouldn't they? Well I can't answer this for certain, but I have seen some application notes showing the first stage after the antenna being a Low Noise Amplifier. If this is the case I am not surprised that high power transmitters, even if they are out of band, will impact on the performance of Wi-Fi. I should say, the cell site does appear to be operating within their license.
In order to test this theory two band pass filters (which allow only the frequencies from 2.4GHz to 2.5GHz to pass and severely attenuate all others) were ordered and installed on the two antennas of one access point on site. These filters were able to attenuate the offending signal by 60dB. Throughput tests were carried out before and after the addition of the filters, and the improvement was clear. Before installation of the filters, performance at about 20m was down to about 1Mbps. After installation this jumped to above 30Mbps, and the engineer was able to get around 30Mbps to a distance of almost 100m.
We also visited some other locations near cell sites, and measured the levels there also. At most sites we received levels between -45dBm and -60dBm, and there were no reports of poor Wi-Fi performance (although no thorough throughput tests were carried out).
So if you have ruled out all other possible causes of your poor Wi-Fi performance, it is possible that a nearby cell site could be the cause of your problems (or lack of decent receive path pre-filtering depending on your angle). You will however need to get your hands on spectrum analyzer in order to determine this conclusively.
