In a conference hall, training room, cafeteria or event area, the complaint always comes in the same sentence: “Full signal, no internet.” The problem isn’t coverage; it’s capacity. Hundreds of devices share the same limited airtime; adding power to the AP doesn’t fix this and often makes it worse. Below we explain high-density Wi-Fi design in the order of a worked capacity calculation, channel reuse, minimum data rate settings, mounting options and pre-event checks.
Coverage design and capacity design aren’t the same thing
Coverage design asks “is the target signal present at every point in the area?” and tries to keep the number of APs as low as possible. Capacity design asks “can all the devices here run their applications at the same time with acceptable performance?” The two approaches often produce opposite results:
| Criterion | Coverage-focused | Capacity-focused |
|---|---|---|
| Cell size | Large | Small, controlled |
| Transmit power | Medium–high | Low |
| Channel width | Can be 40/80 MHz | Usually 20 MHz |
| Determining variable | Area (m²) | Number of devices and applications |
| Success criterion | RSSI map | Channel utilization, latency, user experience |
Wi-Fi is a shared medium: only one transmitter talks on a channel at a time. OFDMA and MU-MIMO make this sharing more efficient but don’t eliminate it. A client talking at a low data rate spends several times the airtime of a fast client sending the same data. In a high-density area the job is two-sided: increasing airtime (more independent channels) and using it efficiently (high data rates, little management traffic). A slow client slows everyone down.
A worked capacity calculation by user and application profile
The calculation rests on four assumptions: the number of users, devices per person, concurrent activity and bandwidth per application. The values below are example assumptions; replace them with the user profile in your own environment and the bandwidth values recommended by the application vendor.
Example: a 200-person conference hall
Connected devices: 200 people × 1.5 devices = 300 devices
Video calls : 60 users × 2 Mbps = 120 Mbps
Other traffic : 240 devices × 0.25 Mbps = 60 Mbps
Total demand : 180 Mbps
Assumption: the realistic total throughput of a 5 GHz radio
with a 20 MHz channel in a dense environment ≈ 50 Mbps (to be verified by measurement)
By throughput : 180 / 50 ≈ 3.6 → 4 radios
By device count : 300 / 50 devices/radio (design) = 6 radios
Result : the larger value → at least 6 × 5 GHz radios
In this example, it was the target number of connected devices per radio, not throughput, that determined the number of APs. This is common in high-density areas: most devices generate little traffic, but each one consumes airtime with management traffic, reconnections and background sync. The devices-per-radio target isn’t the vendor’s technical limit; it’s a value the designer sets from experience and measurement. The maximum client count on the datasheet is far above this target and isn’t used as a design value. That’s exactly a mistake you may encounter.
The 2.4 GHz band contributes very little to capacity in a high-density area; because there are no non-overlapping channels at 20 MHz other than the common 1, 6 and 11 plan, some of these radios need to be turned off after a few APs. If the device ecosystem supports it, the largest capacity increase comes from the 6 GHz band; for choosing a generation, see the Wi-Fi 6, 6E and Wi-Fi 7 comparison article.
Number of APs, channel reuse and cell size
The number of radios from the calculation turns into real capacity only if each radio can operate on a separate channel. Two APs that land on the same channel and hear each other share the airtime of a single AP; this is called co-channel interference. Hanging six APs and squeezing them all into three channels means six radios on paper, three in practice. That’s why in high-density areas:
- A 20 MHz channel width is used. A wide channel increases a single client’s peak speed but reduces the number of independent channels.
- Transmit power is lowered and cells are made smaller. The aim is for APs using the same channel to hear each other as little as possible.
- DFS channels are considered. A significant share of the 20 MHz channels in the 5 GHz band require DFS; when radar is detected, the AP has to change channels. The hall’s location and past radar events should be reviewed from controller logs. The channel list permitted in the country and power limits should be verified against the regulations of the Information and Communication Technologies Authority (BTK).
- Directional antennas make it easier to split cells into seating blocks in large halls.
Automatic RF management (RRM) alone may not be enough in a high-density area. Decisions it makes while the hall is empty become invalid once the hall fills up, because of the attenuation created by human bodies. A static channel and power plan, or a narrowed power range for RRM, gives more predictable results. The general tuning approach is in the channel width and transmit power settings article.
Minimum data rates and reducing broadcast overhead
Beacon and probe response frames are sent at the SSID’s lowest basic data rate; broadcast/multicast frames are also transmitted at a low basic rate close to this on most vendors. With default settings, this rate may be 1 Mbps (2.4 GHz) or 6 Mbps (5 GHz). The order for reducing broadcast overhead in a high-density area:
- Disable low data rates: Disable the 802.11b rates (1, 2, 5.5, 11 Mbps) at 2.4 GHz; at 5 GHz, consider raising the lowest basic rate to 12 or 24 Mbps. This setting also steers clients at the edge of a cell to a closer AP. Before the change, test that older clients continue to connect.
- Limit the number of SSIDs: Each SSID broadcasts separate beacons on each radio. Don’t exceed two or three SSIDs in a high-density area; instead of separate SSIDs for different user groups, identity-based VLAN assignment on the same SSID can be preferred.
- Control multicast and broadcast traffic: Features such as ARP proxy, mDNS/Bonjour filtering or multicast-to-unicast conversion (names vary by vendor) reduce unnecessary broadcast traffic.
- Band steering: Encourage clients that support 5 GHz and 6 GHz onto these bands; however, aggressive steering causes connection delays on some clients.
Mounting options: ceiling, wall, under-seat
| Mounting | Suitable environment | Point to watch |
|---|---|---|
| Ceiling (low, 3–5 m) | Training room, cafeteria | Classic placement; shrink cells with power |
| Ceiling (high) + directional antenna | Conference hall, sports hall | Antenna angle should be calculated for the seating blocks |
| Wall | Narrow and long halls | The middle of the hall may be left without coverage |
| Under-seat / floor | Amphitheater, hall with fixed seating | The human body provides natural isolation; cabling and physical protection are needed |
With under-seat mounting, because the crowd’s bodies attenuate the signal, cells stay small and the same channel can be reused more often. The cost is cabling, floor penetrations and protecting the device against impact and liquids. Whichever mounting is chosen, the design is done for a full hall, not an empty one. An empty hall isn’t a test environment.
Uplink, DHCP and authentication bottlenecks
A high-density area whose wireless side is built correctly can still collapse because of the wired network infrastructure behind it. A significant share of the faults hunted in a panic on the morning of an event aren’t in the RF, but here:
- AP uplink and access switch: For Wi-Fi 6/6E/7 APs with multigigabit ports, 2.5GBASE-T (or 5GBASE-T) uplink support and a PoE class that will run all of the AP’s radios at full power are checked. The uplink capacity from the access switch to the distribution layer is evaluated against total demand.
- DHCP scope: The scope is sized above the expected maximum number of devices; for frequently changing crowds, the lease time is kept short. When the scope fills up, the client connects but can’t get an IP. The user reports this as “Wi-Fi isn’t working.” For relay and snooping configuration, see the DHCP relay and DHCP snooping article.
- 802.1X/RADIUS load: When a session starts, hundreds of clients authenticate at the same time. The RADIUS server’s capacity, certificate validation time and features that speed up reauthentication such as PMK caching/OKC are checked.
- Captive portal: In guest access, the portal server and redirection mechanism can become a bottleneck under sudden load.
Pre-event verification checklist
- The assumptions of the capacity calculation (people, devices, applications) have been confirmed with the event owner.
- Each AP’s channel and power match the plan; there are no APs on the same channel hearing each other, or they have been minimized.
- Low data rates have been disabled, and the older client test has been done.
- The number of SSIDs is limited; unnecessary multicast/broadcast traffic has been filtered.
- The DHCP scope has enough free addresses and a suitable lease time.
- RADIUS and portal servers have been checked for sudden load.
- AP uplinks, the PoE budget and the distribution uplink have been verified.
- Video call and file download tests were done with real devices, if possible while the hall was partly full.
- Channel utilization, client count and retransmission rate are monitored live throughout the event.
We can address capacity calculation, placement and verification for high-density areas together under our wireless network installation and optimization service.
Frequently asked questions
Why is a 20 MHz channel recommended instead of 80 MHz in high-density areas?
As the channel gets wider, the number of independent channels decreases and the number of APs sharing the same channel increases. In a high-density area, total capacity rather than a single client’s peak speed is what matters; 20 MHz allows more APs to operate without disturbing each other.
Can the maximum client count stated by the AP vendor be used in the design?
No. That value is the number of associations the device can technically accept, not the number at which it will deliver acceptable performance. The design uses a much lower, measurement-verified target per radio.
Which clients are affected by disabling low data rates?
Very old devices that support only 802.11b and some embedded devices may not be able to connect; clients at the edge of a cell are also forced to move to a closer AP. Test with the older devices in the inventory before making the change.
Is a measurement taken while the hall is empty enough?
No. The human body attenuates 2.4 and 5 GHz signals; in a full hall, cells and interference conditions change. The design is done assuming a full hall and, if possible, verified with real devices in a partly full hall.