Designing a WLAN that delivers predictable throughput requires understanding how 802.11 clients choose their Modulation and Coding Scheme. The MCS index is the PHY layer’s truth. It reflects modulation order, coding rate, guard interval, and spatial streams. From this index you can infer actual throughput, airtime efficiency, and whether the channel can support the applications you are designing for. If you want a certain throughput outcome, you need a certain MCS. To get that MCS, you need a certain SNR. To get that SNR, you need the right AP density in the right RF conditions. Everything else is noise.
Higher MCS values rely on higher-order QAM. BPSK for MCS 0 sits at the bottom. 64-QAM defines the midrange. 256-QAM and 1024-QAM define the high end. Each jump adds bits per symbol, but the constellation points get closer. The PHY needs a cleaner signal to decode the symbol without error. Coding rate influences this further. A rate like 4/5 or 5/6 removes redundancy and increases efficiency. It also tightens the SNR requirements because there is less protection against bit errors. Coding at 7/8 or 5/6 requires an SNR buffer that the RF environment must maintain consistently, not just momentarily.
This produces a strict relationship between SNR and the MCS level the PHY can hold. While exact thresholds vary by chipset, the ranges are well known. 64-QAM 5/6 generally needs high-20s dB SNR. 256-QAM often requires around 30 dB. 1024-QAM needs mid-30s or better. If the link cannot supply that margin, the client steps down. Each step down lowers the PHY rate. Lower PHY rates stretch airtime usage because each frame occupies the channel longer, which increases contention. The whole BSS slows, even if only a few clients drop to lower MCS levels.
AP density is the primary tool for shaping this SNR environment. SNR is not just RSSI. It is RSSI minus the composite noise floor. AP density controls RSSI by controlling the maximum distance between the AP and the client. It also influences interference because smaller cells allow you to lower transmit power, which reduces overlap between co-channel APs. But density is not free. More APs increase beacon traffic, probe response activity, and the number of devices performing clear channel assessments. All of this interacts with MCS stability and capacity.
This is where CCA behavior enters the picture. Clear Channel Assessment (CCA) is a physical layer (PHY) mechanism that all wireless devices use to determine if the air is busy before transmitting data. This behavior is a core component of the Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) protocol, which prevents multiple devices from transmitting simultaneously on the same channel, causing collisions. It involves two thresholds: energy detect (ED), which triggers when any RF energy rises above a certain level regardless of modulation, and preamble detect (PD), which triggers when the radio sees a valid 802.11 preamble. If energy exceeds ED, or if a preamble is seen above PD, the medium is considered busy. The radio defers and backs off.
Dense deployments cause CCA to fire more often. When too many APs or clients are within ED or PD range, each device senses the medium as busy more frequently even when the interfering signal is too weak to decode. This effectively stretches contention windows and reduces the airtime available for actual data. The result is simple: even with strong RSSI and theoretically enough SNR to support a high MCS, the device may not transmit often enough to maintain the throughput tied to that MCS. CCA can be the breaking point where a design that looks fine on a heat map collapses under real load.
This is why effective AP density is not just a matter of placing more APs. You need enough APs to keep RSSI high and cell sizes controlled, but not so many that CCA turns into a gridlock engine. If the environment is noisy and the ED floor is high, APs will defer more often and lose effective airtime. If the RTS/CTS exchange becomes frequent due to collisions, airtime efficiency drops further. In that scenario, adding APs does not raise SNR enough to offset the hit you take from increased contention and CCA activity.
Management traffic amplifies the problem. Every AP sends beacons at fixed intervals. Each SSID adds another set of beacons. Clients send probe requests. APs answer with probe responses. Multicast and broadcast traffic transmits at the lowest basic rate, which further expands airtime consumption. ACKs, block ACKs, RTS/CTS frames, and OBSS-related signaling in Wi-Fi 6 add additional overhead. These frames count against the same airtime budget that data frames need to use. Real throughput ends up at roughly half the PHY rate in the best conditions. In congested or poorly designed networks, the usable throughput drops far below half.
The noise floor sets the hard boundary for what is possible. A clean environment with a noise floor around -92 dBm or -95 dBm gives a wide SNR margin. You can achieve high MCS values with modest RSSI. You can deploy more APs without triggering excessive CCA. You can shrink cells to stabilize MCS. But if the noise floor rises to -80 dBm or -78 dBm because of external interference, your entire design collapses. The same RSSI no longer provides enough SNR to hold high-order QAM. CCA triggers earlier. Contention rises. Adding more APs does not help because the SNR ceiling is now physically limited by the environment.
A correct design therefore starts with MCS requirements. If an application requires a certain throughput, calculate the airtime needed. From airtime, choose the MCS required. From MCS, determine the SNR required. From SNR, determine the AP spacing required. Then evaluate whether CCA, noise floor, and expected management traffic allow that AP density without locking the channel. Only if all these layers align does the design work.
This is the difference between placing APs by coverage and designing WLANs by physics. Once you understand how MCS, SNR, CCA, coding rate, and noise floor interact, AP density becomes a measurable engineering decision instead of guesswork.
Leave a Reply