Radar Events and Voice Gaps at an Airport Hotel
An airport hotel with 300 rooms and 60 access points sits 1.1 km from a primary surveillance radar. The 5 GHz plan uses channels 36–64 and 100–144 with automatic channel selection and 40 MHz widths on half the APs. Over a typical week the controller logs 40 radar events, each forcing the affected AP off its channel; the AP must then complete a channel availability check before transmitting again and cannot return to that channel for 30 minutes. Lobby and front-desk VoWiFi handsets show 6–11 second audio gaps and dropped calls; self-service check-in kiosks on 5 GHz freeze and require reboots. Guests report poor performance in the east wing. Association counts are healthy and RSSI passes everywhere. The APs are Wi-Fi 6E capable but 6 GHz is disabled. Management will not accept fewer guest-visible SSIDs and will not fund recabling.
1. Explain why the radar events produce the observed symptoms, and what the standard requires an AP to do when radar is detected.
Read the worked answer
This is expected 802.11h behaviour, not a fault. Before using a DFS channel an AP must complete a channel availability check, typically 60 seconds, and while operating it performs in-service monitoring. On detecting radar it must vacate immediately, and it cannot return to that channel for a 30-minute non-occupancy period. With automatic channel selection, the AP then picks a new channel and must run the availability check again, so that AP has no 5 GHz service for a period measured in tens of seconds while its clients hunt for another AP. Forty events a week means forty local outages and forty client stampedes; the handsets show 6–11 second audio gaps because a voice call cannot survive the roam cascade, and the kiosks freeze because their session state does not survive re-association. Healthy RSSI is irrelevant here: the failure is a channel-availability event, not a coverage hole, which is why a coverage survey shows nothing wrong.
2. With only nine non-DFS 20 MHz channels in the band, how would you allocate spectrum across 60 APs? Show the trade-off you are making.
Read the worked answer
Nine non-DFS channels for 60 APs is about 6.7 APs per channel site-wide, so blanket pinning is not an option and would trade radar outages for permanent co-channel interference. Tier the plan by consequence. Pin the roughly 12 APs serving the lobby, front desk, back office and kiosks to the non-DFS channels in UNII-1 and UNII-3, at 20 MHz, so the revenue-critical and real-time clients never see a radar vacate. Leave the guest-room APs on DFS, where a 60-second outage affects a small number of sleeping devices and room attenuation keeps co-channel interference manageable. Drop all 40 MHz widths, which doubles the reuse plan for free. Then enable 6 GHz for the guest SSID: it carries no radar obligation under low-power-indoor rules and adds a large amount of clean spectrum, though its smaller cells and the 6E-only client population mean it offloads rather than replaces 5 GHz. Add 802.11k/v so clients move deliberately instead of stampeding.
3. What would you validate before and after the change, and how would you distinguish genuine radar from false DFS triggers?
Read the worked answer
First establish whether the events are real. Pull the DFS event log per AP per channel with timestamps and correlate it against the radar's operating pattern and against spectrum-analyzer captures at the affected APs; genuine primary radar shows a characteristic chirp in the 5 GHz band, while false triggers often cluster on specific AP models or channels and do not correlate with anything external. That distinction matters, because pinning APs to non-DFS to escape phantom detections gives up real capacity for no benefit. After the change, validate with the clients that matter: run VoWiFi calls across the lobby and front desk during a shift change, measure roam time, call quality and packet loss, and confirm the pinned APs generated no vacate events over two weeks. Track channel utilization and co-channel counts on the pinned APs to prove the interference cost stayed acceptable, and re-check the east-wing complaint against a fresh survey, since that may be an unrelated problem the radar issue was masking.