5.3 Primary Channels, OBSS, Throughput and Guard Interval

PHY amendments tell you which modulations and widths exist. Objective 2.4 asks you to operate those PHYs as a channel plan. Independent OpenExamPrep notes for this objective cover primary channels, overlapping BSS (OBSS), adjacent overlapping versus non-overlapping channels, the difference between throughput and data rate, what bandwidth actually buys you, and guard interval. These ideas decide whether a 1.3 Gbps VHT radio delivers a fast user session or a busy collision domain.

Primary channels and secondary widths

Every BSS has a 20 MHz primary channel, even when the BSS is advertised as 40, 80, or 160 MHz. Beacons, probe responses, and most management frames are sent so that a 20 MHz receiver on that primary can still find the network. Clear channel assessment for the BSS is anchored on the primary: if the primary 20 MHz is busy, a station generally must not transmit a wide PPDU, even if the secondary portions look idle. That rule protects legacy 20 MHz clients and neighboring BSS that share only the primary.

Secondary channels fill out the operating width:

Example: an 80 MHz BSS using 5 GHz channels 36, 40, 44, and 48 with primary 36 uses 40 as secondary 20 and 44+48 as secondary 40. The same 80 MHz block with primary 40 is a different BSS identity for CCA and scanning, even though the RF occupancy looks similar on a spectrum analyzer. 802.11ac can negotiate dynamic bandwidth with RTS/CTS: if a secondary is busy, the pair may fall back to 40 or 20 MHz for that exchange instead of waiting for the entire 80 MHz to be idle. The primary still has to be idle.

In 2.4 GHz HT 40 MHz, the extension (secondary 20) sits above or below the primary. Because the band only has three classic non-overlapping 20 MHz slots in North America, a 40 MHz BSS almost always collides with a neighbor's 20 MHz BSS. That is why section 5.1 treated 2.4 GHz 40 MHz as a special case.

OBSS: overlapping basic service sets

An overlapping BSS (OBSS) is another 802.11 BSS whose coverage overlaps yours and whose channel use overlaps yours. Two APs on the same primary 20 MHz, or on widths that include that primary, form an OBSS for stations that hear both. They share airtime: each must defer when the other is transmitting (unless spatial-reuse rules in 802.11ax say an OBSS frame is weak enough to ignore). OBSS is co-channel contention, not a cabling error. Turning up power often creates more OBSS by stretching overlap. Better answers are channel reuse planning, width reduction (80 MHz down to 40 MHz), AP placement, and, on HE, BSS color plus OBSS PD used carefully.

Do not confuse OBSS with a hidden node. Hidden nodes may be in the same BSS and simply cannot hear each other. OBSS is a second BSS on overlapping spectrum. Adjacent-channel interference (ACI) is a third idea: energy spills into a neighbor channel that is not supposed to overlap, often from poor masks, excessive width, or 2.4 GHz plans that used channels 1 and 4 together.

Adjacent overlapping versus non-overlapping channels

2.4 GHz in North America uses channels 1 through 11 (centers 5 MHz apart). A 20 MHz (or ~22 MHz DSSS) transmission is much wider than 5 MHz, so adjacent numbers overlap heavily. The classic non-overlapping 20 MHz set is channels 1, 6, and 11. Center frequencies 2412, 2437, and 2462 MHz are 25 MHz apart, which is enough for three 20 MHz OFDM BSS to coexist without being the same channel. Channels 1 and 2 are overlapping. Channels 1 and 6 are the designed reuse pair. Some regions have channels 12–14; some European 20 MHz plans discuss 1/5/9/13. For this exam's North American story, memorize 1 / 6 / 11 as the three-cell 20 MHz plan. Using 1, 4, 8, and 11 looks like more capacity on a spreadsheet and produces overlapping energy in the building.

5 GHz non-overlap depends on width. Distinct 20 MHz channel numbers (36 versus 40 versus 44) are laid out not to overlap each other at 20 MHz. Bond them and the picture changes. Two 80 MHz BSS on 36–48 and 52–64 do not overlap. Two 80 MHz BSS that both occupy 36–48 do overlap, even if their primary 20 MHz numbers differ. A 40 MHz BSS on 36+40 overlaps an 80 MHz BSS on 36–48. Always ask: do the occupied 20 MHz pieces share spectrum, not only "are the channel numbers different?" 160 MHz and 80+80 occupy even more 20 MHz pieces, so the number of truly non-overlapping wide BSS in 5 GHz shrinks fast, especially once DFS channels are removed from the plan.

Throughput versus data rate, and what bandwidth does

PHY data rate (the MCS table number) is the speed of the modulated waveform while the radio is transmitting data symbols: for example 65, 300, 433, or 1300 Mbps. Throughput is what remains after 802.11 spends airtime on preambles, MAC headers, encryption overhead, SIFS, Block Ack or ACK, DIFS/AIFS, random backoff, retries, beacon time, and other BSS or OBSS frames. Throughput is therefore lower than data rate in every real BSS. Aggregation (A-MSDU, A-MPDU) raises efficiency by spreading preamble and contention cost across many MSDUs. A clean lab test of a single pair with aggregation might see TCP throughput around half to two-thirds of PHY rate; a busy OBSS office might see far less. Design to airtime, not to the MCS sticker.

Bandwidth in this objective is RF channel width: 20, 40, 80, or 160 MHz. More Hertz means more subcarriers and a higher PHY data rate for the same MCS and stream count — roughly a doubling from 20 to 40, and another large jump to 80, with extra data tones beyond a pure doubling. Wider channels also mean fewer non-overlapping BSS, more chance of secondary-channel busy events, and more noise captured in the receiver. Bandwidth is not free capacity; it is a trade against reuse. A 40 MHz plan with little OBSS can outperform an 80 MHz plan where every AP sits on the same primary.

Guard interval

OFDM sends many slow subcarriers in parallel. Multipath makes delayed copies arrive after the symbol boundary. The guard interval (cyclic prefix) is extra time copied onto the front of the symbol so those late copies do not smash the next FFT window. If delay spread exceeds the GI, you get inter-symbol interference, bit errors, retries, and a lower effective throughput even though the MCS table looked faster.

For 802.11a/g/n/ac, the default GI is 800 ns on a 3.2 microsecond useful symbol (4.0 microseconds total). Short GI is 400 ns (3.6 microseconds total), about an 11 percent PHY-rate gain when the environment is clean enough. Short GI is optional. Outdoor, warehouse, or long-delay rooms often need the long GI. Do not confuse GI with interframe spaces (SIFS, DIFS, AIFS): those are MAC idle times between frames, not the cyclic prefix inside an OFDM symbol.

For 802.11ax HE, the useful symbol is 12.8 microseconds. HE GI options at CWNA depth are 0.8, 1.6, and 3.2 microseconds, so total symbol times are 13.6, 14.4, or 16.0 microseconds. The longer HE symbol already tolerates more delay spread than 11a/n/ac; the 3.2 microsecond GI is the outdoor / long-echo choice. Short 0.8 microsecond HE GI is the efficiency choice when delay spread is small.

Work a mental example. Two 80 MHz 5 GHz APs both use primary 36. They are an OBSS on the primary: beacons, RTS, and data on that 20 MHz contend. PHY rates may still show 866 or 1200 Mbps when a station transmits, but throughput per BSS drops because airtime is shared. Switching one AP to a non-overlapping 80 MHz block (for example 52–64, respecting DFS) restores independent airtime. Enabling short GI on both APs does not fix the OBSS; it only changes symbol timing. Coloring on HE may allow reuse of weak OBSS frames, not of a next-door AP at -50 dBm.

Remember the objective as a chain: pick a primary 20 MHz, choose a width that does not create needless overlap, expect throughput below data rate, spend bandwidth only where reuse still works, and pick a GI that matches delay spread rather than chasing the highest MCS row.