The Lunchtime Noise Floor
A 2.4 GHz office WLAN uses channel 6 with a 20 MHz channel. At 11:30–13:00 each weekday, users near the cafeteria report stalls. The AP near the cafeteria hears a client at -72 dBm; the idle noise floor is -95 dBm, but during lunch the spectrum analyzer shows the noise floor rising to -78 dBm across 2.43–2.47 GHz. The spectrogram shows a wideband, pulsed signature with roughly 60 Hz amplitude modulation and a duty cycle near 50%. A Bluetooth headset and a 2.4 GHz cordless phone are also present, but their emissions are narrowband and frequency-hopping. Channel 1 and 11 show much lower energy. The AP's retry rate rises from 8% to 45%, and throughput falls from 90 Mbps to 12 Mbps.
1. Identify the most likely interferer and justify it from the spectrum signature. Why are the Bluetooth and cordless phone less likely?
Read the worked answer
The signature points to a microwave oven. Microwave leakage is high-power, wideband energy centered near 2.45 GHz, and the 60 Hz pulsed envelope matches AC-powered operation; a 50% duty cycle during lunch is also consistent. The energy spans 2.43–2.47 GHz, which covers channel 6 and raises the noise floor across the whole 20 MHz channel. Bluetooth and cordless phones use frequency hopping and are narrowband at any instant; they would not raise the noise floor uniformly across a 40 MHz span. They also tend to produce short, frequency-varying spikes rather than a steady wideband pedestal. Timing and location complete the case: the effect appears only during cafeteria hours and near the cafeteria AP, which is exactly where a microwave oven would be operated.
2. Calculate the approximate SNR at the client during lunch and explain how that produces the retry rate.
Read the worked answer
The client signal is -72 dBm and the noise floor during lunch is -78 dBm, so the approximate SNR is 6 dB. That is very low for 802.11 OFDM; only the lowest-order modulation and coding schemes have any chance, and even those will have high error rates. The AP's rate adaptation will drop MCS, but if the noise is wideband and pulsed, frames still fail FCS and require retransmission. The retry rate rising to 45% and throughput collapsing to 12 Mbps are consistent with a link operating near its sensitivity limit. The idle noise floor of -95 dBm would have given about 23 dB SNR, which explains why the same link works outside lunch. The calculation also shows that raising transmit power alone is a weak fix: it may improve the client-to-AP direction, but the AP-to-client direction and the noise floor remain problems.
3. Recommend a remediation and validation plan, including why simply changing to channel 1 may not fully solve it.
Read the worked answer
The primary recommendation is to move the cafeteria AP and its clients to 5 GHz, where microwave leakage is not present, then verify client band-steering capability. If 2.4 GHz must be used, move to channel 1 or 11, because the oven energy is centered near 2.45 GHz and channel 6 is most exposed. However, channel 1 or 11 is not guaranteed clean: oven emissions can spread, and the 20 MHz channel edges may still see elevated noise. Validate with a spectrum analyzer during the lunch period on the new channel, measuring noise floor and duty cycle. Then confirm the wireless metrics: SNR should rise, retry rate should fall toward baseline, and throughput should recover. Also check for Bluetooth/cordless interference after the move and consider physical relocation or shielding if the oven is the source.