Wi-Fi 802.11 - BSS, ESS, Association and Frames
- An access point (AP) is a layer-2 bridge between the radio and an Ethernet LAN. One AP and its stations form a BSS (the cell). The BSSID names the cell and is normally the AP radio's MAC address; the SSID is only the human-readable network name.
- Several APs that share one SSID and sit on one wired LAN form an ESS. To the client it looks like one network; each AP is still its own cell on its own channel.
- Joining is scan (beacons or probes) then authenticate then associate, then the key handshake (WPA), and only then the IP side: DHCP, ARP, DNS run as ordinary traffic over the new link.
- An 802.11 frame has up to four address fields because the AP sits between the real sender and receiver. The To DS and From DS bits say which fields hold the BSSID, source and destination.
- Headline Wi-Fi speeds are PHY rates under stated conditions (streams, channel width, modulation). Real throughput is a fraction of it, shared by everyone on the channel.
Mental model
A cafe with a host at the door. The cafe has a name on the sign (the SSID) and a host who knows every guest by seat (the AP and its association table). You walk in, the host checks you may come in, writes you down, and from then on all your orders go through the host, even a message to the guest at the next table. A chain with the same name over every door (an ESS) lets you move between branches, but each branch has its own host and its own room.
Where it breaks: a cafe host serves one conversation at a time by choice. A radio cell has a hard limit: everyone in the cell shares one channel and takes turns on the air, so a guest speaking slowly holds up all the others.
An AP is a layer-2 switch whose "port" to the clients is one shared half-duplex radio: closer to a hub than a switch on the wireless side, with a bridge table of associated stations. An ESS is several such switches on one VLAN that present the same network name, so it behaves like one broadcast domain. A BSSID is to a cell what a bridge ID is to a switch.
Where it breaks: a wired switched LAN never moves a host between ports on its own. In an ESS the client walks from one AP to another, and the wired switch's MAC table still points at the old port until it relearns the client (see Switches, MAC Learning and Multiple Access). Roaming is a re-association plus a layer-2 table update, not a routing event.
How it works
Cell, SSID, BSSID and ESS
In infrastructure mode, the only mode that matters for everyday Wi-Fi, every frame goes through an AP, even between two stations in the same cell. The terms are easy to mix up, so pin each to what it identifies:
- AP - the radio and bridge. It bridges the wireless medium to the wired distribution system (DS), normally an Ethernet switch.
- BSS (basic service set) - one AP plus the stations associated with it.
- BSSID - the 48-bit identifier of the BSS. In infrastructure mode it is the MAC address of the AP's radio. An AP that offers several SSIDs on one radio uses a different BSSID per SSID, so a BSSID names a cell, not a box.
- SSID - a name of up to 32 bytes, chosen by the operator. It is not unique and not an address: two neighbours can both call their network
Home. - ESS (extended service set) - several BSSs joined by one DS that advertise the same SSID. Clients treat it as one network and may move between its APs.
- IBSS - the ad hoc case with no AP, where stations talk directly. Rare today; mesh is covered below.
The cells overlap on purpose, so a client leaving AP1's cell can already hear AP2. The channels 1, 6 and 11 are the three non-overlapping 20 MHz channels in the 2.4 GHz plan used in the US (see Wireless Fundamentals for why), so neighbouring APs do not share airtime.
Finding an AP: beacons and probes
An AP announces itself with a Beacon frame, a management frame broadcast at a regular interval. The default interval is 100 TU, where one time unit (TU) is 1024 microseconds, so a beacon goes out every 102.4 ms. A beacon carries the SSID, the supported rates, the channel, the capabilities (including which security suites the BSS requires), and a timestamp.
A station finds networks in one of two ways:
- Passive scanning - tune to a channel and wait for a beacon. Costs no airtime but the station must dwell at least one beacon interval per channel.
- Active scanning - transmit a Probe Request (with a wildcard SSID, or naming one) and get a Probe Response from APs that hear it. Faster, but the station is transmitting.
Worked example: a passive scan of the 11 US 2.4 GHz channels needs at least 11 x 102.4 ms = 1126.4 ms, over a second, before the station has heard every AP that beacons at the default rate. Active scanning waits only for probe responses, so it can finish each channel in a small fraction of that. Clients therefore mix both: probe where transmitting is allowed, and only listen where it is not (the DFS channels below).
Authenticate, associate
Once the station picks a BSS (usually the strongest signal that offers the SSID and a security suite it supports), two exchanges follow:
- Authentication: two frames. For WPA2 networks this is the legacy open system algorithm and carries no secret; it exists for historical reasons. WPA3-Personal does its real key exchange (SAE) inside these Authentication frames.
- Association: an Association Request (the station's supported rates and capabilities) and an Association Response that returns a status and an association ID (AID). After it the AP forwards frames to and from the station and keeps it in its association table.
If the network is encrypted, the key handshake (WPA's 4-way handshake, four EAPOL-Key messages) follows association; data frames are not allowed through until it completes. The security details are on Wi-Fi Security and Roaming.
The 802.11 frame
Every 802.11 frame starts with a 2-byte Frame Control field containing the protocol version, a 2-bit type, a 4-bit subtype, and flag bits including To DS and From DS, Retry and Protected. The three types:
| Type | Role | Examples (subtype) |
|---|---|---|
| 0 - Management | Find, join and leave a BSS | Association Request 0, Association Response 1, Reassociation Request 2, Probe Request 4, Probe Response 5, Beacon 8, Disassociation 10, Authentication 11, Deauthentication 12 |
| 1 - Control | Help deliver data frames on the shared medium | RTS 11, CTS 12, ACK 13, Block Ack 9 |
| 2 - Data | Carry the payload (an Ethernet-style payload inside) | Data 0, Null 4, QoS Data 8, QoS Null 12 |
The subtype numbers come from IEEE 802.11 (a Beacon is type 0, subtype 8, so Wireshark matches it with wlan.fc.type_subtype == 0x08). A basic data frame header is 24 bytes before the payload: Frame Control (2), Duration/ID (2), three addresses (3 x 6), Sequence Control (2). A fourth address adds 6 bytes (30 in all), and QoS adds 2 more. A 4-byte FCS follows the payload.
Why up to four addresses
An Ethernet frame has two MAC addresses because the sender and receiver are the endpoints of one hop. A Wi-Fi frame crosses a wireless hop and a wired side, so it must name more parties: the receiver on this hop (RA), the transmitter on this hop (TA), plus the original source (SA) and final destination (DA) when they differ from the radio hop. The To DS and From DS bits tell the receiver which role each of the address slots plays:
Read the 1/0 row for a laptop sending to a server: the receiver on the radio hop is the AP (the BSSID), the transmitter is the laptop (SA), and the final destination (the gateway's MAC, for off-link traffic) is in the third slot. The AP rebuilds this as an Ethernet frame for the wired side. The reverse, 0/1, puts the final destination first and the BSSID second. The 1/1 case is a frame travelling between two APs over a wireless link (WDS, or a mesh bridge), where both the original and radio-hop endpoints must be named.
The wired side of this is the ordinary Ethernet frame in Ethernet Frames and MAC Addresses; 802.11 adds the radio-hop fields around the same payload. The AP's uplink port also behaves like any switch port: the switch learns every associated station's MAC address on it (see Switches).
What happens when you join Wi-Fi
An ordered list, with the layer each step belongs to:
- Scan (802.11 management): beacons or probe request/response give a list of BSSs.
- Authenticate and associate (802.11 management): the station gets an AID and is in the AP's table.
- Key setup (WPA, 802.11 and 802.1X/EAPOL): the 4-way handshake derives the keys that encrypt the radio hop. Security is on Wi-Fi Security and Roaming, not here.
- DHCP (UDP, DHCP): the station has no IP yet, so it broadcasts a Discover, gets an Offer, sends a Request and receives an Ack with an address, gateway and DNS server.
- ARP (ARP): the station asks "who has the gateway's IP" to learn the gateway's MAC, because the first hop of any off-link packet is a frame to that MAC.
- DNS (DNS): the first name lookup goes to the resolver DHCP supplied. After it, ordinary traffic flows.
Steps 4 to 6 are the same as on a cable: to the IP layer, 802.11 is just a link that now works.
Deep dive
Channels, widths and bands
Each band is divided into 20 MHz channels, and an AP can bond neighbours into 40, 80 or 160 MHz (802.11ac), and 320 MHz in 802.11be, which exists only in 6 GHz. Wider channels raise the PHY rate roughly in proportion but leave fewer non-overlapping channels, so more neighbours share each one. A quick derivation for 6 GHz: the band is 1200 MHz wide in the US, so at most 1200 / 80 = 15 channels of 80 MHz, 1200 / 160 = 7 of 160 MHz and 1200 / 320 = 3 of 320 MHz (an upper bound; the regulatory channel plan decides the exact set). The band-by-band trade-off (range against bandwidth) is on Wireless Fundamentals.
Channel plans are national. The set of allowed 2.4 GHz channels is 1 to 11 under FCC rules and 1 to 13 under ETSI. Do not assume the US numbers abroad.
DFS: sharing 5 GHz with radar
Part of the 5 GHz band is shared with weather and other radar. In the US, the affected channels are 52 to 64 and 100 to 144 (the rest, 36 to 48 and 149 to 165, are not DFS). Dynamic Frequency Selection (DFS) requires an AP on such a channel to:
- listen for radar for 60 seconds before first using the channel (the channel availability check, CAC; 10 minutes on the weather-radar channels in some regulatory domains),
- vacate within about 10 seconds of detecting radar, and
- stay off that channel for 30 minutes (the non-occupancy period).
A client's constraint follows from this: it may not transmit on a DFS channel until it hears an AP there, so it cannot send a Probe Request. That is why 5 GHz DFS channels are passive-scan only: the station listens for beacons, because a beacon proves a radar-checked AP is operating. The two timers also explain a classic complaint, "Wi-Fi drops for a minute at random": a radar event forces the AP off its channel, and a move to another DFS channel then costs the 60-second CAC. The macOS output in Hands-on shows a neighbour AP on channel 124, a DFS channel.
Regulatory sources are the FCC rules for U-NII devices (47 CFR Part 15 Subpart E) and ETSI EN 301 893 in Europe; the timers above are the commonly cited values, and a device's certified behaviour is what binds it.
Hiding the SSID is not security
An AP can leave the SSID out of its beacons ("hidden network"). The SSID still travels in the clear in the Probe Request a client sends to find it, in the AP's Probe Response, and in every Association Request. Anyone with a monitor-mode radio sees it the first time a client joins. It also makes clients probe for the hidden name everywhere they go. The protection comes from the security suite (WPA), not from the name.
Mesh, repeater and wired backhaul
All three extend coverage; they differ in how a far AP reaches the wired network:
| Design | How the far AP reaches the LAN | Throughput cost | When it fits |
|---|---|---|---|
| Wired backhaul | Ethernet cable to a switch | None: radio airtime is used only for clients | The default for any serious deployment |
| Wireless mesh | APs relay frames to each other over the air (vendor mesh, or IEEE 802.11s) | Each wireless hop reuses airtime; best systems add a separate radio for the backhaul link | Home and small sites where no cable can be run |
| Repeater / extender | Joins the main AP as a client, rebroadcasts the SSID | A single-radio repeater must receive then retransmit on the same channel, so traffic through it gets roughly half the airtime | Dead-spot fix; not for speed |
The halving is a direct consequence of the shared half-duplex medium: one frame occupies the air twice (client to repeater, repeater to AP) on one channel. A dedicated backhaul radio (for example a second 5 GHz radio used only between mesh nodes) avoids it. IEEE 802.11s is the standardized mesh; commercial home mesh systems mostly use their own mechanisms. Mesh links use the four-address format.
Band steering
An AP that offers the same SSID on 2.4 and 5 GHz can nudge dual-band clients toward 5 GHz by delaying its 2.4 GHz probe responses. The choice stays with the client. Band steering is a vendor feature, not an 802.11 rule.
MIMO, MU-MIMO, OFDMA and beamforming
One line each, with the amendment that introduced it:
- MIMO (802.11n): several antennas send independent spatial streams at the same time over the same channel, multiplying rate by the stream count.
- Beamforming (802.11n, standardized sounding in 802.11ac): the AP shapes its transmission toward a client using feedback from that client, raising SNR at the client.
- MU-MIMO (802.11ac Wave 2, downlink; 802.11ax adds uplink): an AP sends different streams to different clients in the same transmission, instead of serving one client at a time.
- OFDMA (802.11ax): the channel is split into small resource units so one transmission carries data for several clients, which helps many small frames (chat, IoT) share airtime.
- BSS colouring and target wake time (TWT) (802.11ax): a colour tag lets a station tell its own BSS from an overlapping one and ignore the other's frames more safely; TWT schedules when battery clients wake.
Generations and peak rates
The Wi-Fi Alliance named three generations retroactively in 2018 (Wi-Fi 4, 5 and 6) and has kept the scheme (6E, 7). The "Max PHY" column is a theoretical per-link PHY rate under the conditions in the last column, not what a speed test shows; the derivation of each follows the table.
| Name | Standard | Year | Band | Max width | Max PHY, under conditions | Key feature |
|---|---|---|---|---|---|---|
| - | 802.11b | 1999 | 2.4 GHz | 22 MHz | 11 Mbit/s (CCK) | First mass-market Wi-Fi |
| - | 802.11a | 1999 | 5 GHz | 20 MHz | 54 Mbit/s (64-QAM, rate 3/4) | OFDM; 5 GHz |
| - | 802.11g | 2003 | 2.4 GHz | 20 MHz | 54 Mbit/s (same OFDM as 11a) | OFDM on 2.4 GHz, compatible with 11b |
| Wi-Fi 4 | 802.11n | 2009 | 2.4 and 5 GHz | 40 MHz | 600 Mbit/s (4 streams, 64-QAM 5/6, short guard interval) | MIMO, 40 MHz channels, frame aggregation |
| Wi-Fi 5 | 802.11ac | 2013 | 5 GHz only | 160 MHz | 6.93 Gbit/s (8 streams, 256-QAM 5/6, short GI); 3.47 Gbit/s with 4 streams | Wider channels, 256-QAM, downlink MU-MIMO |
| Wi-Fi 6 | 802.11ax | 2019 (certified); IEEE 2021 | 2.4 and 5 GHz | 160 MHz | 9.6 Gbit/s (8 streams, 1024-QAM 5/6, 0.8 us GI) | OFDMA, uplink MU-MIMO, BSS colouring, TWT |
| Wi-Fi 6E | 802.11ax | 2020 to 2021 | adds 6 GHz | 160 MHz | Same as Wi-Fi 6 | Wi-Fi 6 radios in the 6 GHz band |
| Wi-Fi 7 | 802.11be | 2024 | 2.4, 5 and 6 GHz | 320 MHz (6 GHz only) | 46.1 Gbit/s (16 streams, 4096-QAM 5/6, 0.8 us GI) | 320 MHz, 4096-QAM, multi-link operation (MLO) |
Where the numbers come from, so you can reproduce them: PHY rate is data subcarriers x bits per symbol x code rate x streams / symbol time.
- 802.11n: 40 MHz has 108 data subcarriers; 64-QAM is 6 bits; rate 5/6; symbol 3.6 us with the short guard interval.
108 x 6 x 5/6 / 3.6 us = 150 Mbit/sper stream, 600 Mbit/s with 4 streams. - 802.11ac: 160 MHz has 468 data subcarriers; 256-QAM is 8 bits;
468 x 8 x 5/6 / 3.6 us = 866.7 Mbit/sper stream; 8 streams give 6.93 Gbit/s and 4 streams 3.47 Gbit/s. First-wave products (80 MHz, up to 3 streams) topped out at3 x 433.3 = 1.3 Gbit/s. - 802.11ax: 160 MHz has 1960 data subcarriers; 1024-QAM is 10 bits; the symbol is 13.6 us (12.8 us plus a 0.8 us guard).
1960 x 10 x 5/6 / 13.6 us = 1201 Mbit/sper stream; 8 streams give 9.6 Gbit/s. - 802.11be: 320 MHz has 3920 data subcarriers; 4096-QAM is 12 bits.
3920 x 12 x 5/6 / 13.6 us = 2882 Mbit/sper stream; 16 streams give 46.1 Gbit/s, and the 2-stream devices that actually ship give 5.76 Gbit/s.
Client devices rarely have more than 2 spatial streams, so a typical Wi-Fi 6 phone on an 80 MHz channel tops out near 1.2 Gbit/s PHY (980 x 10 x 5/6 / 13.6 us = 600 Mbit/s per stream, times 2), and real transfers are lower still because airtime is shared. Wi-Fi 6 does not make one client faster as much as it lets an AP serve many clients efficiently.
CSMA/CA and RTS/CTS
A radio cannot listen while it transmits, so 802.11 avoids collisions: sense the channel, wait a DIFS, count a random backoff, send, and expect an ACK. Optional RTS/CTS plus the NAV timer silences hidden stations.
The timing, contention window doubling, NAV and RTS/CTS are on Wi-Fi CSMA/CA.
Textbook vs real world
| The textbook says | Production does | Why |
|---|---|---|
| An AP is a hub or bridge for its cell | Enterprises run controller-managed or cloud-managed APs: a central controller sets channels and power and tunnels or bridges client traffic (CAPWAP is the standard tunnel) | A hundred APs configured one at a time drift apart; central control keeps channels, power and policy consistent |
| One AP covers an area; more power is better | Dense design: many low-power APs on 20 MHz channels so cells are small and each carries few clients | Airtime is the scarce resource; small cells reuse the same channel across the building |
| The label speed is what you get | Client count, airtime and the weakest client decide throughput | The advertised figure is a PHY rate under best-case conditions, shared by every station in the cell |
| An ESS is one big LAN | It is one broadcast domain, and a roam still means a re-association and a MAC-table update on the wired side | Radio cells are separate; the wired network only learns where a client moved when it sees the next frame |
In production
- Controller and cloud-managed APs. Enterprise vendors such as Cisco Meraki manage APs from a central controller or cloud dashboard. The IETF standardized the controller-to-AP tunnel as CAPWAP (RFC 5415, with the 802.11 binding in RFC 5416).
- Federated Wi-Fi. eduroam lets a user from one institution sign in at another with their home credentials, using the same SSID
eduroameverywhere and a chain of RADIUS servers to reach the home institution. - Home mesh systems extend one network over several nodes; the quality difference is mostly whether the nodes have a dedicated backhaul radio or a cable (the table above).
- Generations in the field. The Wi-Fi Alliance publishes what each generation adds (Wi-Fi 6, Wi-Fi 7); adoption figures move yearly, so use their current pages and vendor shipment reports rather than a number frozen here.
History and evolution
- 1997 IEEE publishes the first 802.11 standard, at 1 and 2 Mbit/s in 2.4 GHz (IEEE 802.11 working group).
- 1999 802.11b (11 Mbit/s in 2.4 GHz) and 802.11a (54 Mbit/s in 5 GHz) are published, and the Wi-Fi Alliance brand is created; the name was coined by a branding firm and is not an acronym (Wi-Fi history, Wi-Fi). 2.4 GHz crowds up quickly.
- 2003 802.11g brings OFDM and 54 Mbit/s to 2.4 GHz while staying compatible with 802.11b devices.
- 2009 802.11n adds MIMO and 40 MHz channels, because a single stream on 20 MHz had topped out at 54 Mbit/s.
- 2013 802.11ac moves to 5 GHz only with 80 and 160 MHz channels and 256-QAM, and 802.11ac Wave 2 adds downlink MU-MIMO.
- 2018 The Wi-Fi Alliance renames the generations Wi-Fi 4, 5 and 6 so users can tell them apart.
- 2019 Wi-Fi 6 (802.11ax) certification starts. The goal is not just speed but efficiency in dense deployments: OFDMA, BSS colouring and TWT.
- 2020 The FCC opens 1200 MHz at 6 GHz, which Wi-Fi 6E devices use (FCC announcement).
- 2024 Wi-Fi 7 (802.11be) certification begins, adding 320 MHz channels, 4096-QAM and multi-link operation (IEEE 802.11be task group).
- Now Wi-Fi 6 and 6E are widespread, Wi-Fi 7 is shipping.
- Next The 802.11bn task group works on "ultra high reliability" (informally Wi-Fi 8), aimed at latency and reliability rather than peak rate (IEEE 802.11bn task group).
Hands-on
On macOS, system_profiler SPAirPortDataType needs no sudo and lists your link and the neighbours it can hear (older guides use the removed airport tool). Real output from this machine, abbreviated, with the network names redacted:
Look at three things. Network Type: Infrastructure is the BSS mode from above (the alternative is ad hoc). Channel 149 (5GHz, 80MHz) is a bonded 80 MHz channel; the centre frequency is 5000 + 5 x 149 = 5745 MHz. The neighbour on channel 124 is on a DFS channel (100 to 144 in the US), so that AP must have passed a 60-second radar check before using it, and a client can only learn about it by passive scanning. For the live details of your own link (BSSID, rates, MCS) use sudo wdutil info; it needs sudo, so its output is not shown here.
On Linux (not run on this machine; abbreviated, and the fields vary with driver and iw version):
The BSS line is the BSSID, so two lines with the same SSID and different BSSIDs are two APs of one ESS. The beacon interval: 100 TU is the default from above. The output above is illustrative of that format, with made-up addresses.
To see the frames themselves, put the radio in monitor mode and capture. On macOS sudo tcpdump -I -i en0 -e -w beacons.pcap writes a capture, and in Wireshark the display filter wlan.fc.type_subtype == 0x08 shows only beacons; wlan.fc.type == 0 shows all management frames. Monitor mode captures every frame on one channel, so lock to the AP's channel first.
Gotchas
Interview questions
Show answer
The SSID is the network name an operator chooses; it is not unique. The BSSID identifies one cell and is normally the MAC address of the AP's radio, with a separate BSSID per SSID on a multi-SSID radio. An ESS is several BSSs, usually several APs on one wired LAN, that advertise the same SSID so clients treat them as one network.
Follow-ups:
- What happens when a client roams between APs? It re-associates to the new BSSID on the same SSID. The wired switch updates its MAC table for the client once it sees the client's next frame on the new AP's port.
Show answer
It scans (beacons or probe request and response), picks a BSS, authenticates (open system, or SAE for WPA3), and associates to receive an association ID. For a secured network the WPA 4-way handshake then derives the keys. After that the link is up and normal IP bring-up runs: DHCP for an address, gateway and DNS server, ARP for the gateway's MAC, and DNS lookups.
Follow-ups:
- Where do DHCP, ARP and DNS happen? All three run above 802.11 as ordinary frames. The AP bridges them onto the wired LAN, where the DHCP server and gateway sit; the DNS server is wherever DHCP pointed.
Show answer
Because a frame crosses a wireless hop and a wired side, and the radio-hop endpoints are not always the original source and final destination. The To DS and From DS bits select the meaning: a station-to-AP frame carries the BSSID, source and destination; AP-to-station carries destination, BSSID and source; and an AP-to-AP frame (To DS and From DS both 1) adds a fourth field so that receiver, transmitter, destination and source are all named.
Follow-ups:
- Who are the four? In the 1/1 case: receiver address, transmitter address, destination address, source address.
- When do you see the fourth? Wireless distribution (WDS) or mesh links between APs.
Show answer
Passive scanning means tuning to a channel and waiting for a beacon, which costs no airtime but takes at least a beacon interval (102.4 ms by default) per channel. Active scanning sends a Probe Request and gets Probe Responses, which is faster. On a DFS channel a station may not transmit until it hears an AP there, because the AP's presence shows the channel passed its radar check, so it scans passively.
Follow-ups:
- What is the DFS timeline? About 60 seconds of radar listening before first use, vacate within about 10 seconds of detecting radar, and stay off for 30 minutes.
Show answer
The cell shares one half-duplex channel and the basic access rule gives stations roughly equal chances to send frames, not equal airtime. A client at a low rate takes much longer to send the same frame, so it holds the medium longer per byte and every other client's throughput falls toward its own. This is the performance anomaly.
Follow-ups:
- What fixes it? Airtime fairness (the AP schedules by time rather than frames), pushing weak clients to a better band, and in Wi-Fi 6, OFDMA, which lets one transmission serve several clients at their own rates.
Show answer
Wi-Fi 6 (802.11ax) targets efficiency with many clients, not just peak rate. OFDMA splits a channel into resource units so one transmission serves several clients; uplink MU-MIMO lets clients send to the AP in parallel; BSS colouring helps a station tell its own cell from an overlapping one so it can reuse the channel more safely; and TWT schedules when battery devices wake. The gain shows up when many clients send small frames, such as a stadium or an office.
Follow-ups:
- What is Wi-Fi 6E? The same 802.11ax in the 6 GHz band, with wide clean channels and no pre-802.11ax devices sharing them.
Show answer
A wired backhaul connects each AP to the LAN by cable, so the radio carries only client traffic. A repeater joins the main AP as a client and rebroadcasts, so each frame uses the air twice on one channel and throughput through it roughly halves. A mesh relays between nodes wirelessly with coordinated routing; its quality depends on whether it has a dedicated backhaul radio, which avoids sharing airtime with clients.
Follow-ups:
- What is a dedicated backhaul radio? A separate radio used only for node-to-node links, so the backhaul and the client side do not contend for the same channel.
Fun facts
- The beacon interval of 102.4 ms is 100 time units, and one time unit is 1024 microseconds.
- "Wi-Fi" is not an acronym. A branding firm coined the name in 1999, and an early Alliance slogan, "The Standard for Wireless Fidelity", made people read it as one.
- 802.11n's 600 Mbit/s needs 4 streams on 40 MHz. A single-stream device on 20 MHz gets
52 x 6 x 5/6 / 3.6 us = 72.2 Mbit/s.
Cheat sheet
- 100 TU = 102.4 ms - default beacon interval (1 TU = 1024 us)
- 24 / 30 bytes - basic 802.11 data header without / with the fourth address
- 0x08 - Beacon type and subtype in Wireshark (
wlan.fc.type_subtype == 0x08) - 1, 6, 11 - non-overlapping 2.4 GHz channels in the US plan
- 60 s, 10 s, 30 min - DFS radar check, vacate time, non-occupancy period
- 600 Mbit/s, 6.93, 9.6, 46.1 Gbit/s - 802.11n, ac, ax, be theoretical PHY maxima, under the stream and width conditions in the table