High-Density Auditorium: Coverage Passes, Capacity Fails
A 300-seat auditorium hosts monthly all-hands meetings. Attendees bring two to three devices each, producing a peak of about 1,200 clients. The WLAN has 12 ceiling APs. Each AP is configured with 80 MHz channels in 5 GHz and 20 MHz in 2.4 GHz, and both bands broadcast the same SSID. Users report video stalls, poll-app timeouts, and roaming failures near the stage. A predictive survey performed for coverage at -67 dBm shows adequate signal, but no capacity survey was done. The company wants live 1080p streaming to every device and plans to keep the existing AP model, which supports 802.11ax, OFDMA, and MU-MIMO. No 6 GHz spectrum is available. The team can change channels, transmit power, SSIDs, and AP count, but cannot replace client devices.
1. Using the client count, AP count, and channel plan, explain why coverage is adequate but the user experience is failing. Show the capacity arithmetic and identify the dominant bottlenecks.
Read the worked answer
Coverage at -67 dBm only proves that the signal reaches the seats; it says nothing about airtime. The first arithmetic check is clients per radio: 1,200 clients divided by 12 APs is 100 clients per AP. Even if every client used only the 5 GHz radio, 100 associated clients per radio is well above the 30-50 active clients per radio that a high-density design should target. The second check is channel reuse. An 80 MHz channel plan in 5 GHz yields roughly six non-overlapping channels in common regulatory domains. With 12 APs, that is at least 2:1 co-channel reuse, and in practice more because some channels are DFS or avoided. Co-channel interference raises retries and airtime consumption. The third check is aggregate load. 1,200 simultaneous 1080p streams at about 5 Mbps each is about 6 Gbps of application demand. Twelve 2x2 Wi-Fi 6 APs at 80 MHz might advertise 600 Mbps PHY each, but real aggregate throughput after overhead, retries, and management traffic is far lower, and it is shared airtime, not per-client bandwidth. The 2.4 GHz band has only three non-overlapping 20 MHz channels, so it cannot absorb the overflow. The dominant bottlenecks are therefore AP density, 80 MHz channel width causing high co-channel interference, and an unmanaged dual-band SSID that lets low-rate 2.4 GHz clients consume airtime. Adding OFDMA or MU-MIMO helps efficiency, but it does not fix a 100-client-per-radio design.
2. Recommend a concrete RF and WLAN configuration plan for the next all-hands, including channel width, band strategy, SSID/rate settings, and AP count. Justify each change against the bottlenecks you identified.
Read the worked answer
The design should shift from coverage-based to capacity-based. First, reduce 5 GHz channel width to 20 MHz, or 40 MHz only where the spectrum plan proves no co-channel overlap. In a 12-AP room, 20 MHz gives enough non-overlapping channels to plan a reuse pattern and cuts co-channel interference. Second, add APs so the target is roughly 40-50 active clients per radio, not 100. With 1,200 clients, that implies about 24-30 5 GHz radios if all clients are on 5 GHz, so the team should plan for roughly 24-30 APs, subject to a proper capacity survey. Use directional antennas or downtilt near the stage rather than high transmit power. Third, disable 2.4 GHz for the high-density SSID, or leave it only for legacy devices with 20 MHz, disabled 1/2/5.5/11 Mbps rates, and a minimum basic rate of 12 or 24 Mbps. Low rates consume disproportionate airtime. Fourth, use band steering to move capable clients to 5 GHz, limit the SSID count to two or fewer, and enable airtime fairness. For the video application, use multicast-to-unicast conversion if the stream is multicast, enable WMM/QoS, and mark video appropriately. Enable Wi-Fi 6 features such as OFDMA, MU-MIMO, and BSS coloring, but treat them as efficiency gains, not substitutes for AP density. Lower transmit power so cells are smaller and clients roam more predictably. If the poll app requires local discovery, verify whether client isolation or mDNS filtering breaks it before enabling isolation globally.
3. How would you validate the design before the event, and what measurements would prove it is ready?
Read the worked answer
Validation should be a capacity test, not another coverage walk. Before the event, perform a predictive capacity survey using the final AP count and channel plan, then confirm with an on-site validation survey at seated height and at the stage. Measure signal at the cell edge, but more importantly measure co-channel interference and channel utilization per AP. Run a load test with representative clients: at least 40-50 active clients per radio, concurrent video streams, and the poll app. Use iperf3 or a similar tool to measure aggregate throughput and per-client throughput, and record retries, airtime utilization, and latency. Acceptance criteria should include channel utilization below roughly 70% under peak test load, retry rates below about 10%, stable roaming between APs, and successful DHCP and authentication for the full client count. Check the DHCP scope size and lease time, RADIUS or PSK load, and controller CPU. During the event, monitor per-AP client counts, channel utilization, retries, and video quality; have a rollback plan for channel width and AP additions. If the load test shows one AP serving far more clients than its neighbors, adjust transmit power or antenna orientation rather than simply raising power, which would increase co-channel interference.