skip to content

WPA2

WPA2 as the long-standing Wi-Fi baseline: AES-CCMP for confidentiality and integrity, Personal and Enterprise modes, the four-way handshake, and the KRACK flaw that forced a patch wave.

on this pageshow

explore

questions

22

On a WPA2 network, what does the "AES" encryption setting actually refer to, and what does it protect in each data frame?

level: juniorimportance: must knowfreq 52%

answer

  1. a mode, not just a cipher
  2. Counter Mode with CBC-MAC Protocol
  3. header plus MIC around the body
  4. addresses stay readable

basics

~20 s

In WPA2, "AES" means CCMP: AES in CCM mode with a 128-bit per-session temporal key. It encrypts each data frame's body and adds an 8-octet MIC that detects tampering, while the MAC header travels in the clear.

solid answer

~40 s

"AES" in a WPA2 setup is shorthand for **CCMP**, the Counter Mode with CBC-MAC Protocol from 802.11i-2004. It runs AES in CCM mode with a 128-bit temporal key: counter mode encrypts the frame body (the IP packet and everything above it), and CBC-MAC produces an 8-octet MIC over the body and most of the MAC header, so a modified frame is dropped. An 8-octet CCMP header carries a 48-bit packet number, and the receiver rejects any frame whose number is not higher than the last one it accepted, which stops replays. What it does not hide: MAC addresses, frame sizes and timing, and management frames unless Protected Management Frames is negotiated. The label "WPA2-TKIP" names the retired RC4-based stopgap that CCMP replaced; in hostapd the setting to write is `rsn_pairwise=CCMP`.

go deeper

for a junior

Recall that WPA2's AES is CCMP, that it uses a 128-bit key, and that it both encrypts the frame body and adds a MIC. Say plainly that MAC addresses stay visible.

for a middle

Explain the frame layout: MAC header, 8-octet CCMP header with packet number and Key ID, encrypted body, 8-octet MIC. Name the RSN element's pairwise suite as the place the cipher is announced.

for a senior

Show you verify rather than trust labels: read the RSN element, set rsn_pairwise explicitly in hostapd, and know which header fields and frame types CCMP leaves unprotected.

for a principal

Frame the answer as a trust boundary: CCMP protects data frames against outsiders on the link, while admission, management-frame protection and key distribution are separate controls a security standard must also require.

## What "AES" names in a WPA2 configuration WPA2 is the Wi-Fi Alliance certification built on the IEEE **802.11i-2004** amendment, which defined the **Robust Security Network (RSN)**. Its data-protection cipher suite is **CCMP** — the *Counter Mode with CBC-MAC Protocol* — which runs the **AES** block cipher in **CCM mode** (RFC 3610, NIST SP 800-38C). When a router or a controller offers "WPA2-AES", it means CCMP. "WPA2-TKIP" names the Temporal Key Integrity Protocol, an RC4-based stopgap designed to run on WEP-era hardware; it is retired, and CCMP is what replaced it as the baseline. CCMP keys AES with a **128-bit temporal key (TK)**. Each associated client gets its own pairwise temporal key for unicast traffic, and the access point shares a group temporal key with all associated clients for broadcast and multicast traffic. Where those keys come from is the job of the four-way handshake, covered separately; for this question it is enough that the key is per session, derived by the devices, and never typed by anyone. ## What CCMP does to each data frame For every protected data frame — strictly, every MPDU, the unit the radio actually transmits — CCMP: - **inserts an 8-octet CCMP header** right after the 802.11 MAC header, carrying a 48-bit **packet number (PN)** and a **Key ID** that tells the receiver which installed key protected the frame; - **encrypts the frame body** — the LLC/SNAP header, the IP header, the transport ports and the application data — with AES in counter mode; - **appends an 8-octet MIC** (message integrity code), computed with AES-CBC-MAC over the plaintext body and most of the MAC header, so any modification is detected and the frame is discarded; - **sets the Protected Frame bit** in the Frame Control field so the receiver knows the body must be decrypted. | Part of the frame | Encrypted? | Covered by the MIC? | |---|---|---| | MAC header (addresses, most Frame Control bits) | No | Mostly, as additional authenticated data | | CCMP header (PN, Key ID) | No | The PN is bound in through the nonce | | Frame body (IP packet and above) | Yes | Yes | | MIC | Yes, with the keystream | It is the check itself | The packet number also provides **replay protection**: the receiver discards any frame whose PN is not higher than the last one it accepted for that key. CCMP therefore delivers three properties for data frames — confidentiality, integrity and freshness — with one key and one AES engine. ## What CCMP does not hide or cover "The Wi-Fi is encrypted" overstates what CCMP does. With CCMP in force: - the **MAC addresses** of sender and receiver, the BSSID, frame lengths and timing remain visible to anyone in radio range; - a few header fields are deliberately left out of the MIC — the Duration/ID field, and Frame Control bits such as Retry that can legitimately differ on a retransmission; - **management frames** (beacons, probe responses, deauthentication) are not protected by CCMP on a WPA2 network unless Protected Management Frames (802.11w) is negotiated, which is a separate subject; - **who is admitted** is not CCMP's decision at all: that is authentication, Personal or Enterprise, and the cipher only protects the link that authentication set up. ## Checking what is in force A reviewer or an engineer can confirm the cipher without trusting a product label: 1. Read the **RSN element** (element ID 48) in the access point's beacons or probe responses. Its pairwise cipher list holds suite selectors: `00-0F-AC:4` is CCMP, and `00-0F-AC:2` would mean TKIP is still offered. 2. Read the configuration. In hostapd, the open-source reference implementation, the WPA2 pairwise cipher is `rsn_pairwise=CCMP`. Write it explicitly: when it is unset, hostapd copies `wpa_pairwise`, whose default is TKIP unless hostapd was built with `CONFIG_NO_TKIP`. 3. Read association failures. A client that requests a pairwise cipher the access point does not offer is refused with status code 42, invalid pairwise cipher. ## Why interviewers ask it this way The question checks whether a candidate can move past "WPA2 uses AES" to what AES is doing: a **mode** (CCM), a **key size** (128 bits), a **per-frame header and MIC**, and a **counter against replays**. A strong answer names CCMP, says it gives data frames confidentiality, integrity and replay protection, and is honest that the MAC header, management frames and client admission are separate stories. A weak answer stops at "AES is strong encryption", which says nothing about what a single frame on the air actually carries.

  • Does CCMP protect broadcast and multicast data frames as well as unicast ones?
    Yes. Unicast frames are protected with each client's own pairwise temporal key, and broadcast and multicast frames from the access point with the group temporal key that every associated client holds. Both use CCMP when CCMP is the negotiated cipher, and each key has its own packet numbers, so replay protection applies to group traffic too. The group key is replaced periodically by the access point.
  • Is CCMP's 128-bit key weaker than the 256-bit keys some newer cipher suites use?
    Not in any practical sense: NIST SP 800-97 puts CCMP's key space at 2^128, far beyond exhaustive search. CCMP-256 and GCMP-256 exist for policies that demand a higher security level, such as WPA3-Enterprise's 192-bit mode, and they also lengthen the MIC to 16 octets. On an ordinary WPA2 network the real risks sit in how keys are shared and managed, not in AES key length.
  • What happens to a CCMP-protected frame whose MIC check fails at the receiver?
    It is discarded. CCM releases no plaintext when the MIC does not verify, so nothing reaches higher layers and the frame's packet number is never recorded. TKIP had countermeasures that deauthenticated stations after repeated MIC failures (reason code 14); CCMP has no such mechanism, and TKIP itself is retired in favour of CCMP.

  • WPA2-AES encrypts the whole frame, including the MAC addresses.
  • The WPA2 passphrase itself is the AES key that encrypts each frame.
  • CCMP only encrypts; integrity comes from the frame check sequence.
  • WPA2-AES and WPA2-TKIP are the same protocol with different key lengths.
  • WPA2's CCMP also protects beacons and deauthentication frames by default.
open as a page

In WPA2-Personal, how do the client and access point prove they share the passphrase without ever sending it over the air?

level: juniorimportance: must knowfreq 48%

basics

~20 s

Both sides turn the passphrase into the same Pairwise Master Key locally, swap random nonces in the four-way handshake, derive a session key from them, and prove possession with a MIC only the correct key can produce.

open as a page

What went wrong in WPA2's four-way handshake in the 2017 key reinstallation (KRACK) disclosure, and why did fixing it not need a new protocol?

level: juniorimportance: must knowfreq 58%

basics

~20 s

In WPA2's four-way handshake, a client reinstalled the same key whenever message 3 was retransmitted, resetting its nonce and replay counter. Patches install a key only once and stay compatible, so no new protocol was needed.

open as a page

In WPA2, how does Personal (PSK) mode differ from Enterprise (802.1X) mode in who is authenticated and where the keys come from?

level: juniorimportance: must knowfreq 70%

basics

~20 s

WPA2-Personal gives every device one shared passphrase that becomes the same PMK on all of them. WPA2-Enterprise authenticates each user or device with 802.1X/EAP against an authentication server, so every session gets its own PMK.

open as a page

How does WPA2's CCMP use AES in counter mode and CBC-MAC to give each data frame both confidentiality and integrity with one key?

level: middleimportance: must knowfreq 38%

basics

~20 s

CCMP runs AES-CCM with one 128-bit temporal key: CBC-MAC over the protected header fields and the plaintext yields an 8-octet tag, then counter mode encrypts the body and the tag. The receiver decrypts, recomputes the tag and drops mismatches.

open as a page

In WPA2's four-way handshake, what does each of the four EAPOL-Key messages carry, and what does each one prove?

level: middleimportance: must knowfreq 40%

basics

~20 s

Message 1 brings the AP's ANonce; message 2 the client's SNonce and a MIC proving it holds the PMK; message 3 the AP's MIC, RSN element and wrapped GTK with the order to install; message 4 confirms installation.

open as a page

For an 802.1X Wi-Fi SSID, how do EAP-TLS, PEAP, EAP-TTLS and TEAP differ in what each demands of the client fleet?

level: middleimportance: must knowfreq 48%

basics

~20 s

EAP-TLS needs a certificate and private key on every client and no password. PEAP and EAP-TTLS need only a server certificate and run a password method inside a TLS tunnel. TEAP can chain machine and user authentication in one tunnel.

open as a page

What does the 8-octet CCMP header carry in a WPA2 data frame, and how does its packet number stop replayed frames?

level: middleimportance: should knowfreq 30%

basics

~20 s

The CCMP header carries a 48-bit packet number, the ExtIV bit and a 2-bit Key ID. The sender increments the number per frame; the receiver drops any authenticated frame whose number is not higher than the last accepted.

open as a page

In WPA2, where does the PMK come from, and what do the KCK, KEK and TK carved from the PTK each protect?

level: middleimportance: should knowfreq 30%

basics

~20 s

The PMK comes from the passphrase via PBKDF2 in Personal mode, or from the EAP MSK in Enterprise. The PTK derived from it splits into the KCK for handshake MICs, the KEK for wrapping group keys, and the TK for unicast data.

open as a page

Which networks and cipher suites did the WPA2 key reinstallation flaw affect, and does changing the Wi-Fi passphrase mitigate it?

level: middleimportance: should knowfreq 34%

basics

~10 s

It affected WPA and WPA2, Personal and Enterprise, under TKIP, CCMP and GCMP alike; TKIP and GCMP were hit hardest. Changing the passphrase does not prevent or mitigate it; updating devices does.

open as a page

When an office moves from WPA2-Personal to WPA3-Personal (SAE), which Personal-mode weaknesses does SAE fix and which still need Enterprise?

level: middleimportance: should knowfreq 42%

basics

~20 s

SAE fixes what a recorded exchange reveals: each session gets a fresh PMK and a listener cannot test passwords offline. It leaves the credential model alone: one shared password, no per-user identity, and revoking anyone still means changing it everywhere.

open as a page

A security reviewer asks which parts of a WPA2 data frame CCMP's MIC covers and how its nonce stays unique — what do you answer?

level: seniorimportance: should knowfreq 18%

basics

~20 s

CCMP's MIC covers the body plus AAD from the MAC header — addresses, most Frame Control bits, fragment number, QoS TID — but not Duration/ID or changeable bits. The 13-octet nonce is priority, transmitter address and a never-reused packet number.

open as a page

Why does a WPA2-Personal client with a mistyped passphrase associate successfully and then fail, with the access point logging an invalid MIC on message 2?

level: seniorimportance: should knowfreq 26%

basics

~20 s

Association never tests the passphrase. The first frame that depends on it is message 2: the client's wrong PMK yields a wrong KCK, so its MIC fails at the AP, which discards it, retries message 1 and finally deauthenticates.

open as a page

An IT team with forty branch sites gets a WPA2 key reinstallation advisory on Monday: which laptops, phones and access points need patches, and is patching only the infrastructure enough?

level: seniorimportance: should knowfreq 22%

basics

~20 s

Patch all clients first, since most are exposed. An access point needs a fix only if it supports fast transition or repeater functions. Patching only infrastructure is not enough: vulnerable clients stay exposed, and patched and unpatched devices still interoperate.

open as a page

A WPA2-Personal audit flags the PMKID in the RSN element: why is that exposure a property of the design, not a patchable bug, and what narrows it?

level: seniorimportance: should knowfreq 16%

basics

~20 s

The PMKID is an identifier derived from the PMK and both addresses, and it is visible by design. It lets a passphrase guess be tested offline without a connected client, so only a stronger passphrase, Enterprise or SAE narrows it.

open as a page

On a WPA2-Enterprise SSID, you disable a departed user's account, yet their laptop stays connected for hours: why, and how do you cut it off?

level: seniorimportance: should knowfreq 26%

basics

~20 s

Disabling the account only bites at the next EAP run; the laptop already holds valid keys, and a cached PMK can let it rejoin without EAP. Deauthenticate it, clear the cached PMK, and keep reauthentication timers bounded.

open as a page

A branch card-terminal gateway on a WPA2 SSID will never get a key reinstallation fix; what do you decide about the gateway, its access points and the accepted risk?

level: principalimportance: should knowfreq 7%

basics

~20 s

Set a retirement date and compensate meanwhile: serve the gateway from access points running hostapd's wpa_disable_eapol_key_retries=1 or an equivalent, accepting reduced robustness. The protection holds only while it is connected to such an access point, so record the residual risk.

open as a page

A 600-person company's Wi-Fi passphrase is on a meeting-room wall and known to former staff: would you choose EAP-TLS, PEAP with passwords, or a longer passphrase, and why?

level: principalimportance: should knowfreq 30%

basics

~20 s

A longer passphrase only treats the leak; it is still one shared secret. Move to per-user 802.1X: EAP-TLS as the standard for managed devices, PEAP as a time-limited bridge, and a separate Personal SSID for headless devices.

open as a page

A scanner lists CVE-2017-13082 against one device and CVE-2017-13087 against another; what do these identifiers say about the flaw, the handshake and the fix?

level: middleimportance: nice to knowfreq 9%

basics

~20 s

Each identifier in the CVE-2017-13077 to 13088 series is one instance of key reinstallation, naming a handshake and a key. CVE-2017-13082 is the fast BSS transition handshake; CVE-2017-13087 is the group key in a WNM sleep mode response.

open as a page

How does GCMP-256 differ from CCMP in protecting an 802.11 data frame when access points add a 192-bit WPA3-Enterprise SSID or Wi-Fi 7 links?

level: seniorimportance: nice to knowfreq 12%

basics

~20 s

GCMP-256 replaces CCM with AES-GCM: a 256-bit temporal key, a 16-octet MIC instead of 8, and GHASH authentication instead of CBC-MAC. The 8-octet header, 48-bit packet number and replay rule stay the same; clients must support it to join.

open as a page

On a WPA2 network, why is the GTK shared by every client, and what decides when the access point rotates it with the group key handshake?

level: seniorimportance: nice to knowfreq 14%

basics

~20 s

Broadcast and multicast frames must be readable by every client, so they share one GTK. The AP rotates it with a two-message group key handshake on a timer, daily by default in hostapd, and optionally whenever a holder leaves.

open as a page

What does turning on WPA3-Enterprise 192-bit mode change on the SSID and in the EAP exchange, and what does it demand of clients?

level: seniorimportance: nice to knowfreq 10%

basics

~20 s

WPA3-Enterprise 192-bit mode advertises AKM 00-0F-AC:12, protects data with GCMP-256, requires PMF, derives keys with HMAC-SHA-384 and restricts EAP's TLS to P-384 or 3072-bit RSA. Every client and certificate must support it; nothing weaker can join that SSID.

open as a page