SecurityL2-L3

Security Attacks & CVEs

Security Troubleshooting Analysis

Wi-Fi Security Attacks & CVE Reference - Sources: Vanhoef et al., NIST NVD, Wi-Fi Alliance Security Bulletins10 rows#

Attack Name	CVE(s)	Year	Standards Affected	Technical Mechanism	Attack Outcome	Wireshark / Detection Signature	Mitigation / Patch
KRACK (Key Reinstallation)	CVE-2017-13077 to 13088	2017	WPA2 (all) 802.11r FT 802.11 group key	AP retransmits EAPOL Msg 3 if it doesn't receive Msg 4. STA reinstalls same TK with nonce reset to zero. In CCMP: nonce reuse with same TK → attacker can XOR two ciphertexts to recover keystream. In TKIP: MIC forgery possible. 802.11r FT: Reassoc Req replays allow PTK reinstall.	Decrypt encrypted unicast traffic (CCMP). Inject/replay frames (TKIP). TKIP RC4 keystream recovery. 802.11r: full PTK reinstallation = total session compromise.	Capture: EAPOL Msg 3 retransmit + Msg 4 already ACKed Filter: eapol && wlan.fc.retry==1 after eapol.keydes.nonce == [previous nonce] Look for: duplicate Msg 3 with same ANonce	AP fix: never retransmit Msg 3 after receiving Msg 4. Client fix: do not accept Msg 3 if TK already installed (check for reinstall). Patch: WPA supplicant 2.7+, iOS 11.1+, Android Oct 2017. FT: disable 802.11r if unpatched.
PMKID Attack	- (Steube 2018)	2018	WPA2-Personal WPA3-SAE (partially)	RSN IE in Assoc Req/Resp contains PMKID = HMAC-SHA1-128(PMK, 'PMK Name' \|\| AP_MAC \|\| STA_MAC). Attacker captures a single Assoc frame (no full handshake needed). PMKID can be cracked offline: try all dictionary passwords → compute PMK via PBKDF2 → compute PMKID → compare.	Offline dictionary/brute-force against PMKID. No need to capture 4-way handshake. Works instantly if AP broadcasts RSN IE with PMKID (most do). hashcat -m 22000.	Capture: Assoc Req with RSN IE Extract: wlan.rsn.ie.pmkid hcxdumptool + hcxtools to extract hash Hashcat mode 22000	WPA3-SAE: PMK derived via Dragonfly → not susceptible to dictionary attack. WPA2: Use strong (20+ char random) passphrases. Enterprise (802.1X) immune - PMK derived from MSK. Mitigated by: WPA3 transition mode or pure WPA3-Personal.
DragonBlood (SAE Side-Channel)	CVE-2019-9494 CVE-2019-9496	2019	WPA3-Personal (SAE / Dragonfly)	Two attacks: (1) Timing side-channel: hunting-and-pecking loop iterations vary with password → timing differences expose hash-to-group operation count → offline dictionary with ~40 timing measurements. (2) Cache-based: branch mispredictions in H&P expose cache access patterns → offline attack. (3) Downgrade: AP accepting WPA3 + WPA2 transition can be forced to WPA2.	Partial offline password recovery (timing oracle). Cache side-channel allows distinguishing password guesses. Downgrade to WPA2 enables PMKID/dict attack.	Timing attack: measure SAE Commit response latency across multiple connections with different guessed passwords. Downgrade: wlan.rsn.ie showing WPA2 only after WPA3 AP seen	Fix: H2E (hash-to-element) - constant-time, cache-oblivious PWE derivation. Mandated in Wi-Fi Alliance WPA3 v3.0+. Transition Disable element: AP tells STA to disable WPA2 fallback permanently. Software: hostapd 2.10+ with sae_pwe=1.
Deauth Flood (Pre-PMF)	-	2003- ongoing	WPA/WPA2 without PMF (legacy networks)	IEEE 802.11 Deauth frames are unauthenticated pre-802.11w. Attacker spoofs AP BSSID as source, sends broadcast Deauth (reason 3 or 6) to all clients. Client state machine → State 1 immediately. Repeat flood = sustained DoS.	All clients disconnected from BSS. Continuous flood = no client can stay connected. Used in evil twin attacks to force clients to reassociate to attacker AP.	High rate: wlan.fc.type_subtype==12 && wlan.sa == [AP_BSSID] Broadcast: wlan.da == ff:ff:ff:ff:ff:ff Tool signature: aireplay-ng -0 output	802.11w PMF (Protected Management Frames) - Deauth is CCMP-encrypted + MIC-protected. WPA3 makes PMF mandatory. Legacy mitigation: WIDS/WIPS deauth flood detection + client-side rate limiting.
Evil Twin / Rogue AP	-	Ongoing	All standards	Attacker deploys AP with same SSID as legitimate BSS. May use higher TX power or deauth flood on legitimate AP. Client probes with SSID match → associates with evil twin. Attacker performs MitM on data path. HTTPS downgrade + sslstrip if client doesn't use HSTS.	Full MitM. Credential harvesting. Traffic interception. DHCP rogue server → DNS poisoning.	Two BSSIDs advertising same SSID: wlan.ssid == 'TargetSSID' RSSI delta between legitimate AP and rogue WIDS: new BSSID for known SSID at unexpected location	802.11u Interworking + Passpoint: clients only associate to authenticated networks (domain validated). VPN: end-to-end encryption renders MitM useless. Certificate pinning in apps. 802.1X: STA validates server certificate = detects impersonation.
KARMA Attack	-	2004	Open networks Legacy WPA2-PSK	Client sends directed Probe Request for a remembered SSID (e.g. 'HomeWiFi'). KARMA tool on rogue AP responds to ANY directed probe with a matching Probe Response. Client associates to rogue AP believing it found a trusted network.	Client silently connects to attacker AP. Full MitM on all traffic.	wlan.fc.type_subtype==5 (Probe Resp) responding to directed probe from unknown BSSID Tool: hostapd-wpe, Metasploit karma module	Random MAC for probe requests (iOS 14+, Android 10+ do this). Wildcard-only probes (don't leak SSID names). Client: forget unused networks. Use WPA3/Passpoint for automatic trusted network selection.
WPS Pixie Dust (Offline nonce)	CVE-2014-3652 CVE-2015-4141	2014	WPA2-Personal (WPS enabled)	WPS uses two half-PINs (8-digit split into 4+4 = 11000 attempts). Pixie Dust: some AP firmware generates weak/predictable random nonces (E-S1, E-S2) in WPS exchange using rand() seeded with time(). Offline computation: guess E-S1/E-S2 → recover PIN → recover PSK without brute force.	Full PSK recovery in seconds on vulnerable firmware. No need for online brute force.	WPS exchange: EAP/WPS M1-M8 messages wlan_eapol.type == WPS in Wireshark Tool: reaver, bully, pixiewps	DISABLE WPS entirely - no patch exists for design flaw. Wi-Fi Alliance deprecating WPS in favor of DPP/EasyConnect. Router vendors: some allow disabling WPS per-radio.
Beacon Flood	-	Ongoing	All standards	Attacker creates thousands of fake SSIDs with random BSSIDs. AP scan results fill with bogus SSIDs. Client scanning spends excessive time processing probe responses. Some devices/OSes crash or become extremely slow.	DoS on client scanning. Degraded user experience. Some client drivers have memory exhaustion vulnerabilities.	wlan.fc.type_subtype==8 at abnormally high rate Thousands of unique BSSIDs in scan results Tool: mdk4 b mode	WIDS detection: beacon rate anomaly per channel. Client-side: rate-limit scan processing. OS patches for beacon processing buffer limits.
FragAttacks	CVE-2020-24587 CVE-2020-24588 CVE-2020-26140	2020	WPA1/WPA2/WPA3 (all standards since 1997)	Three attack families: (1) Frame aggregation: attacker can cause STA to aggregate a plaintext frame into an encrypted A-MSDU. (2) Mixed-key: fragment and defragment with different keys. (3) Plaintext injection: inject plaintext in fragmented frame bypass. Design flaws in 802.11 fragmentation/aggregation since 1997.	Inject arbitrary plaintext into encrypted network. Bypass NAT/firewall. Potential remote code execution via injected DHCP response.	Fragmented frames: wlan.frag != 0 A-MSDU flag in QoS Control set unexpectedly Tools: fragattacks testing suite (Vanhoef)	OS/driver patches: Linux mac80211 patches April 2021, Windows KB5003173, iOS 14.5. Validate A-MSDU SubDA field is not multicast. Accept fragments only from same TA with same key.
SSID Confusion (Multi-SSID)	CVE-2023-52160 (wpa_supplicant)	2023	WPA2-Enterprise (wpa_supplicant <2.10)	EAP-TTLS/PEAP: wpa_supplicant did not verify that the SSID in the beacon matched what the user configured when no CA cert was set. Attacker creates rogue AP with different SSID but same EAP server cert → client connects thinking it's the correct network.	Client connects to wrong SSID. Traffic exposed to attacker. Credentials potentially exposed if inner method is weak.	wlan.ssid mismatch between targeted SSID and actual association EAP inner exchange on rogue AP	wpa_supplicant 2.10+: validates SSID matches configured network during EAP. Always configure CA certificate in 802.1X profile. Wi-Fi Alliance: SSID binding required in WPA3-Enterprise.

WIDS/WIPS Detection Signatures - Key Thresholds for Security Monitoring6 rows#

Detection Event	Threshold / Signature	Frame Filter	Severity	Response Action	False Positive Risk	Source	Notes
Deauth flood	≥ 20 Deauth frames/second from same BSSID	wlan.fc.type_subtype==12 rate>20/s	CRITICAL	Alert + block spoofed BSSID + notify clients	Low (legitimate deauth rate is <1/min)	NIST SP 800-97	Broadcast deauth especially suspicious. With PMF, legitimate deauth is encrypted.
Rogue AP (SSID match)	BSSID not in authorized AP list broadcasting known SSID	wlan.ssid in [known_list] && wlan.bssid not in [auth_list]	HIGH	Locate physically (RSSI triangulation), isolate on wired side if found	Medium (neighbors may use same SSID)	Wi-Fi Alliance WIPS Guidance	Cross-reference with wired 802.1X port authentication logs.
WPS active on AP	WPS IE present in Beacon or Probe Response	wlan_eap.type == 0xFE && vendor == 00:50:F2:04	HIGH	Alert security team; request WPS disable	Low	CVE-2014-3652 advisory	WPS should be disabled on all enterprise and secure consumer APs.
4-Way Handshake anomaly	EAPOL Msg 3 retransmitted after Msg 4 ACKed	eapol.type==3 && wlan.fc.retry==1 after msg4 seen	CRITICAL	Investigate client/AP firmware for KRACK patch status	Low	CVE-2017-13077	Unpatched AP is the most common finding. Check firmware dates.
Client probing for sensitive SSIDs	Directed Probe Req for corporate SSID from unknown device	wlan.fc.type_subtype==4 && wlan.ssid==[corp_ssid] && wlan.sa not in [known_clients]	MEDIUM	Log + investigate device identity	Medium	NIST SP 800-97	May indicate KARMA preparation or device misconfiguration.
Beacon flood / SSID pollution	≥ 50 unique BSSIDs/minute on same channel	count(unique wlan.bssid) > 50 per 60s window per channel	MEDIUM	Alert; ignore synthetic BSSIDs in scan	Low	802.11 WIDS best practice	Mdk4 tool signature. Can degrade AP scanning performance.