Mạng không dâyWireless Networking

Part of the Network Engineer Roadmap knowledge base.

Overview

Wireless networking replaces the copper or fiber cable of a traditional LAN with radio waves. Instead of confining a signal inside a shielded medium, a wireless device modulates data onto an electromagnetic carrier and radiates it into free space, where any receiver tuned to the same frequency and using the same protocol can pick it up. This one change — from a guided medium to an unguided, shared, broadcast medium — is the source of every property that makes wireless different: it is convenient and mobile, but also half-duplex, contention-based, subject to interference and attenuation, and fundamentally insecure unless you encrypt it.

The dominant technology for local-area wireless is Wi-Fi, the consumer-facing brand for the IEEE 802.11 family of standards. But “wireless networking” is broader than Wi-Fi: it also includes Bluetooth for short-range personal-area links, cellular (3G/4G/5G) for wide-area mobile data, and a long tail of technologies (Zigbee, Thread, LoRa, NFC) for IoT and specialized use. This note focuses on Wi-Fi — the part a network engineer designs, deploys, and troubleshoots every day — and then surveys the neighboring technologies you need to recognize.

We build up from the physics (spectrum, bands, channels, interference), through the 802.11 standards timeline (from 802.11b in 1999 to Wi-Fi 7 today), into WLAN architecture (autonomous vs controller-based vs cloud-managed), then wireless security (WEP → WPA → WPA2 → WPA3, PSK vs enterprise, and the WPA-vs-WPS confusion), and close with the neighboring wireless technologies and a practical troubleshooting model for the three classic failure classes: coverage, interference, and capacity.

Related notes:

• Network types & topologies (WLAN, PAN, WAN context): ./02-network-types-and-topologies.md
• Network devices (access point, WLC, controller): ./04-network-devices.md
• Network security (802.1X, RADIUS, encryption context): ./11-network-security.md

Fundamentals

Radio spectrum and the ISM/U-NII bands

Radio communication happens across the electromagnetic spectrum, measured in hertz (Hz). Wi-Fi lives in slices of spectrum that regulators (the FCC in the US, ETSI in Europe, etc.) have set aside as unlicensed — meaning you do not need a per-transmitter license to operate, but you must obey power and behavior rules and you must tolerate interference from everyone else sharing the band. This is why your microwave oven, your neighbor’s Wi-Fi, and your Bluetooth headphones can all legally interfere with your network.

Wi-Fi uses three main bands:

The trade-off is fundamental physics: lower frequency = longer range and better penetration; higher frequency = more available bandwidth but shorter range. A 2.4 GHz signal will reach farther and go through more drywall than a 6 GHz signal at the same power.

Channels and channel width

Each band is divided into channels — sub-slices of frequency an AP and its clients agree to use. A channel has a center frequency and a width (bandwidth). Wider channels carry more data but occupy more spectrum and are more prone to interference.

• 20 MHz — the base channel width; the only sane choice in 2.4 GHz.
• 40 MHz — two bonded 20 MHz channels (802.11n and later).
• 80 MHz — four bonded channels (802.11ac and later).
• 160 MHz — eight bonded channels (802.11ac/ax); needs lots of clean spectrum, realistic mainly in 5 GHz (with DFS) and 6 GHz.
• 320 MHz — Wi-Fi 7 (802.11be), only feasible in 6 GHz.

The 2.4 GHz channel problem. In 2.4 GHz there are 14 channels (1–13 in most regions, 14 Japan-only), but each 20 MHz channel is wider than the 5 MHz spacing between channel numbers, so adjacent channels overlap. Only channels 1, 6, and 11 are non-overlapping. Any deployment that ignores this — putting neighboring APs on channels 3 and 5, for example — creates co-channel and adjacent-channel interference that quietly destroys throughput. The rule “1, 6, 11 only” in 2.4 GHz is one of the most important pieces of practical Wi-Fi knowledge.

2.4 GHz non-overlapping channels:

Ch1        Ch6        Ch11
|===|      |===|      |===|
2412 MHz   2437 MHz   2462 MHz

DFS (Dynamic Frequency Selection). Some 5 GHz channels are shared with weather and military radar. On these channels, an AP must listen for radar and vacate the channel if it detects one (DFS). DFS channels give you more spectrum but can cause brief outages when a radar event forces a channel change.

Interference and attenuation

Because the medium is shared and open, a wireless signal degrades in ways a cable never does:

• Attenuation — signal strength falls with distance and every obstacle. Measured in dBm (a logarithmic scale where each −3 dB halves power). A healthy client sees roughly −65 dBm or stronger; below ~−80 dBm connections struggle.
• Co-channel interference (CCI) — two APs (or clients) on the same channel must take turns; they share airtime and slow each other down. Counter-intuitively, CCI is often preferable to ACI.
• Adjacent-channel interference (ACI) — overlapping channels bleed energy into each other, raising the noise floor and corrupting frames. Worse than CCI.
• Non-Wi-Fi interference — microwave ovens, cordless phones (2.4 GHz), Bluetooth, baby monitors, wireless cameras. Measured as a low SNR (Signal-to-Noise Ratio). You want SNR ≥ 20 dB for good data rates, ≥ 25 dB for VoIP/video.
• Multipath — the signal reflects off surfaces and arrives at the receiver via multiple paths at slightly different times, causing fading. Modern Wi-Fi (MIMO) actually exploits multipath rather than just suffering from it.

SSID, BSS, and ESS

• SSID (Service Set Identifier) — the human-readable network name (e.g., CorpWiFi). It is not a security feature; hiding the SSID (not broadcasting it in beacons) only lightly obscures the network and can be trivially discovered by sniffing client probes.
• BSSID (Basic Service Set Identifier) — the MAC address of the AP’s radio for a given SSID. It uniquely identifies one cell.
• BSS (Basic Service Set) — a single AP and its associated clients: one cell of coverage.
• ESS (Extended Service Set) — multiple APs sharing the same SSID, connected to the same backbone, so clients can roam between them as one continuous network. This is how a whole office or campus presents one Wi-Fi network across dozens of APs.
• IBSS (Independent BSS) — ad-hoc mode: clients talk directly, no AP. Rare in practice.
• Beacon frames — the AP periodically (~every 100 ms) broadcasts a beacon announcing the SSID, supported rates, security, and capabilities so clients can discover and join.

How the airtime is shared: CSMA/CA

Wired Ethernet uses CSMA/CD (collision detection). Wireless cannot detect collisions while transmitting (a radio can’t listen and shout at the same time), so 802.11 uses CSMA/CA — Carrier Sense Multiple Access with Collision Avoidance. A station listens before sending, waits a random backoff, and (optionally) uses RTS/CTS handshakes to reserve the medium. The key consequence: Wi-Fi is half-duplex and a shared medium. All devices on a channel share the airtime; adding more clients or more APs on the same channel divides that finite airtime. This is why capacity planning (not just coverage) matters.

Key Concepts

Wi-Fi standards (IEEE 802.11) — full comparison

The 802.11 working group has released a steady stream of amendments. The Wi-Fi Alliance later introduced simple generation names (Wi-Fi 4, 5, 6, 7) to replace the confusing letter suffixes.

Key ideas embedded in that table:

• MIMO (Multiple-Input Multiple-Output) — multiple antennas send/receive multiple independent data streams simultaneously (spatial streams), multiplying throughput. Written as 4×4:4 (4 TX antennas, 4 RX antennas, 4 streams).
• MU-MIMO (Multi-User MIMO) — the AP talks to several clients at once using different spatial streams, instead of serving them one at a time.
• OFDMA (Orthogonal Frequency-Division Multiple Access) — Wi-Fi 6’s headline feature. It splits a channel into small Resource Units (RUs) so the AP can serve many clients in a single transmission, dramatically improving efficiency for lots of small packets (IoT, chat, voice). This is the biggest reason Wi-Fi 6 helps dense networks more than raw speed suggests.
• BSS Coloring — tags frames with a “color” so a device can tell its own BSS’s traffic from a neighbor’s on the same channel, allowing more spatial reuse (transmit when a distant same-channel signal is present rather than always deferring).
• TWT (Target Wake Time) — lets battery devices schedule when to wake and talk, hugely extending IoT battery life.
• MLO (Multi-Link Operation) — Wi-Fi 7’s headline feature: a single client uses multiple bands (e.g., 5 GHz + 6 GHz) simultaneously for higher throughput, lower latency, and reliability.

WLAN architecture: autonomous, controller-based, cloud-managed

How you manage a fleet of APs is a defining architectural choice.

Autonomous (standalone/fat) APs. Each AP is fully self-contained: it has its own configuration, does its own encryption, RF management, and forwarding. Fine for a home or a tiny office (1–3 APs). Painful at scale — every config change and firmware update is per-AP, and there is no coordinated channel/power management or seamless roaming.

Controller-based (lightweight/thin APs + WLC). The intelligence moves to a central WLAN Controller (WLC). APs become “lightweight” — they handle radio, but the controller handles configuration, security policy, RF management (RRM: automatically tuning channels and power across all APs), and often data forwarding. The AP and WLC communicate over CAPWAP (Control And Provisioning of Wireless Access Points), a standardized tunneling protocol:

• CAPWAP control channel (UDP 5246) carries management/config.
• CAPWAP data channel (UDP 5248) can tunnel client traffic back to the controller (centralized/tunneled forwarding) or the AP can forward locally (FlexConnect/local switching for branch offices).

Benefits: one place to configure everything, coordinated RF, fast/seamless roaming, centralized monitoring. The classic enterprise on-prem model (e.g., Cisco WLC + Aironet/Catalyst APs).

Cloud-managed Wi-Fi. The controller function lives in the vendor’s cloud (e.g., Cisco Meraki, Aruba Central, UniFi’s controller). APs phone home to a cloud dashboard for config, monitoring, and analytics; data usually still forwards locally at the AP. Benefits: zero on-prem controller hardware, manage many sites from one dashboard, easy scaling. Trade-off: dependence on internet/cloud reachability and (usually) recurring licensing.

Access points, controllers, and roaming

An access point (AP) bridges the wireless (802.11) and wired (802.3 Ethernet) worlds: it converts frames between the two and enforces association and encryption. Most enterprise APs are powered by PoE (Power over Ethernet) so a single cable carries data and power. See ./04-network-devices.md for the AP’s place in the device taxonomy.

Roaming is the process by which a moving client leaves one AP’s cell and joins another without dropping the connection. The client (not the AP) decides when to roam, based on signal strength. Plain roaming re-does the full authentication with each AP, which can cause a noticeable gap — fatal for a VoIP call. Fast roaming standards fix this:

• 802.11r (Fast BSS Transition / FT) — pre-establishes the security keys so re-association is near-instant.
• 802.11k (Neighbor Reports) — the AP tells the client which neighboring APs/channels to consider, so it doesn’t have to scan blindly.
• 802.11v (BSS Transition Management) — lets the network suggest a better AP to the client (load balancing, steering).

Together, k/v/r give the seamless roaming enterprises need for voice and video.

Coverage vs capacity planning (the basics).

• Coverage planning asks “is there a strong enough signal everywhere it’s needed?” You design for a target signal level (e.g., ≥ −67 dBm) at the cell edge. A site survey (predictive with software, or physical with a tool like Ekahau) maps coverage and finds dead spots. Old wisdom of “put APs far apart to cover more area” fails in high-density spaces.
• Capacity planning asks “can the network carry the load of all the clients here?” A lecture hall with 300 laptops needs many APs at lower power with smaller cells so airtime is divided among more radios — even though coverage alone would need only two APs. Modern high-density design is capacity-driven: more, smaller cells; disable slow legacy data rates; use 5/6 GHz; balance clients across bands (band steering).

Wireless security: WEP → WPA → WPA2 → WPA3

Because anyone in radio range can capture your frames, encryption is not optional. The security protocols evolved as each generation was broken.

WPA3 improvements worth naming:

• SAE (Dragonfly handshake) replaces WPA2’s PSK 4-way handshake in Personal mode. Even if an attacker captures the handshake, they cannot run an offline dictionary attack against it, and every session gets forward secrecy (compromising the password later doesn’t decrypt past traffic).
• 192-bit mode for WPA3-Enterprise (government/high-security).
• OWE (Opportunistic Wireless Encryption) / “Enhanced Open” — encrypts open (passwordless) networks like coffee-shop Wi-Fi, so traffic isn’t sent in the clear even with no password.

Personal (PSK) vs Enterprise (802.1X). This is orthogonal to WPA version — you can run WPA2-Personal or WPA2-Enterprise, WPA3-Personal or WPA3-Enterprise.

• Personal / PSK (Pre-Shared Key) — one shared password for the whole SSID. Simple, fine for home/small office. Weakness: everyone shares one secret; if it leaks, you must re-key everyone, and it’s the target of offline cracking (in WPA2).
• Enterprise / 802.1X — each user authenticates individually against a RADIUS server (often tied to Active Directory), using EAP (e.g., PEAP, EAP-TLS with certificates). No shared password; per-user credentials, per-user keys, central revocation, and audit trails. This is the standard for any organization. The chain is: supplicant (client) → authenticator (AP/WLC) → authentication server (RADIUS). See ./11-network-security.md for 802.1X/RADIUS/AAA details.

WPA vs WPS — a common point of confusion

These sound similar but are completely different things:

• WPA (Wi-Fi Protected Access) is the family of security/encryption protocols described above (WPA/WPA2/WPA3). It protects your traffic.
• WPS (Wi-Fi Protected Setup) is a convenience feature for joining a network without typing the passphrase — via a button press or an 8-digit PIN.

WPS is a security risk. The 8-digit PIN method has a serious design flaw: the router validates the PIN in two halves, so a brute-force attack only needs to guess ~11,000 combinations (and the last digit is a checksum), making it crackable in hours (the “Reaver” attack). Recommendation: disable WPS, especially the PIN method. WPS being enabled can undermine an otherwise strong WPA2/WPA3 network.

Best Practices

• Never use WEP or WPA (TKIP). Use WPA2-AES at minimum; prefer WPA3 (or WPA3/WPA2 transition mode for mixed clients). Enable “Enhanced Open”/OWE for guest/open networks.
• Use WPA-Enterprise (802.1X/RADIUS) for organizations; reserve PSK for home/small/IoT SSIDs, and use a long, random passphrase (20+ chars) for PSK to resist dictionary attacks.
• Disable WPS entirely (especially PIN). Convenience is not worth the exposure.
• In 2.4 GHz, use only channels 1, 6, 11. Never use overlapping channels. Prefer 20 MHz width in 2.4 GHz.
• Steer clients to 5/6 GHz with band steering; keep 2.4 GHz for legacy/IoT and long-range low-bandwidth devices.
• Design for capacity, not just coverage in dense spaces: more APs at lower power, smaller cells, and disable low legacy data rates (e.g., 1/2/5.5/11 Mbps) so slow clients don’t hog airtime.
• Do a site survey (predictive before install, validation after). Target ≥ −67 dBm and SNR ≥ 25 dB in areas that carry voice/video. Re-survey after major layout or occupancy changes.
• Enable fast roaming (802.11k/v/r) for voice/video and mobile users; test roaming with a real handset/laptop while walking the space.
• Separate SSIDs by purpose and put them on VLANs: corporate, guest (internet-only, isolated), and IoT (locked-down). Keep the number of SSIDs small — every extra SSID consumes airtime with its own beacons.
• Use PoE and plan uplinks: a Wi-Fi 6/7 AP can exceed 1 Gbps, so feed high-end APs with multi-gig (2.5G/5G) switch ports where possible.
• Keep AP/WLC firmware patched (KRACK, FragAttacks, and similar Wi-Fi vulnerabilities are fixed in firmware).
• Monitor the RF, not just the wire: watch channel utilization, retransmission rates, SNR, and client counts per AP — most “the Wi-Fi is slow” tickets are airtime/interference problems, not bandwidth problems.

Troubleshooting the three classic problem classes

Almost every wireless complaint reduces to one of three root causes. Diagnose by asking which one you’re facing:

1. Coverage problem — the signal is too weak where the user is. Symptoms: drops in specific spots, low signal (−80 dBm+), works fine near the AP. Fix: add/relocate APs, increase power (carefully), better antennas, remove obstacles. Confirm with a survey/heatmap.
2. Interference problem — the signal is present but noisy. Symptoms: high retransmissions, low SNR, works intermittently even with good signal, worse at certain times (microwave at lunch). Fix: change channels (respect 1/6/11), use a spectrum analyzer to find non-Wi-Fi sources, reduce channel width, move off DFS if radar events recur.
3. Capacity problem — too many clients or too much load per cell. Symptoms: fine with few users, degrades as the room fills, high channel utilization (>50–70%), high airtime on slow clients. Fix: add APs/smaller cells, enable OFDMA/MU-MIMO capable gear, band-steer to 5/6 GHz, disable legacy rates, upgrade to Wi-Fi 6/6E/7.

A quick mental checklist for any Wi-Fi issue: Is the signal strong enough (coverage)? Is the signal clean enough (interference)? Is there enough airtime for everyone (capacity)?

Other wireless technologies

Bluetooth

Bluetooth is a short-range (typically ~10 m for Class 2) PAN (Personal Area Network) technology in the 2.4 GHz band, using frequency-hopping spread spectrum to dodge interference. It connects peripherals — headsets, keyboards, mice, wearables, car audio. BLE (Bluetooth Low Energy) is a separate, low-power variant for battery-sensitive IoT and beacons (fitness bands, sensors, indoor-positioning beacons). Bluetooth and Wi-Fi both use 2.4 GHz and can interfere; modern chips coordinate to coexist. Bluetooth is not a general LAN technology — it’s about pairing devices, not carrying network traffic.

Cellular / mobile networks (3G/4G/5G)

Cellular networks provide wide-area wireless data through a grid of licensed-spectrum base stations (“cells”). A network engineer should recognize the generations:

Cellular is relevant to network engineers for WWAN failover (a 4G/5G modem as backup internet for a branch), fixed wireless access (FWA) as a broadband alternative, and IoT connectivity.

Hotspot and tethering

• Tethering shares a phone’s cellular internet with another device over USB, Bluetooth, or Wi-Fi.
• Mobile hotspot is Wi-Fi tethering specifically: the phone (or a dedicated MiFi device) becomes a small AP, broadcasting an SSID that other devices join, while it uplinks over cellular. It’s the same Wi-Fi technology described above, just with a cellular WAN backhaul instead of a wired one. Useful for travel and as emergency failover, but subject to carrier data caps and battery drain.

References

• IEEE 802.11 Working Group — the standards themselves (a/b/g/n/ac/ax/be)
• Wi-Fi Alliance — Discover Wi-Fi — Wi-Fi generations, WPA3, WPS, certification programs
• Wi-Fi Alliance — Security (WPA3) — WPA2/WPA3, SAE, Enhanced Open/OWE
• Cisco — Wireless LAN (WLAN) overview — controllers, CAPWAP, RRM, deployment models
• Cisco — 802.11 Wireless Fundamentals / CWNP body of knowledge — spectrum, channels, RF design, site surveys
• FCC — U-NII / 6 GHz unlicensed operation — spectrum and regulatory rules
• IETF RFC 5415 — CAPWAP Protocol Specification
• MetaGeek — Wi-Fi channels, interference, and SNR guides — practical RF troubleshooting