skip to content

Rogue APs & Deauth Floods

The over-the-air threats and the controls: rogue and evil-twin APs, deauth and disassociation floods and the PMF that stops them, the WPS PIN flaw, jamming, and WIDS/WIPS detection.

on this pageshow

questions

6

How does a rogue access point differ from an evil twin, and why can't a client tell a twin by SSID and security type alone?

level: juniorimportance: must knowfreq 55%

answer

  1. one is about permission, one about names
  2. a name is only a label
  3. an open network has no server identity
  4. the client profile carries the weight
  5. monitoring compares what it hears with an inventory

basics

~20 s

A rogue AP is any unauthorised access point, possibly wired into your network; an evil twin impersonates an authorised SSID. A client cannot tell a twin by name and security type, because both are advertised labels, not proof of identity.

solid answer

~50 s

A **rogue AP** is defined by authorisation: a device the organisation did not approve, often a consumer AP in a wall port, which can open a path around the perimeter into the wired network. An **evil twin** is defined by impersonation: it advertises the name of an authorised network such as `corp-example`, so clients that trust the name may join it. One device can be both, or only one. A client sees the SSID and the security type in the beacon, and any transmitter can repeat both. On an open SSID there is no server identity at all, so the weight moves to the client profile: what the device joins on its own and whether it insists on the security it was configured with. Detection is a monitoring job: compare every transmitter heard against an inventory of authorised APs.

go deeper

for a junior

Recall the pair: rogue means not authorised, evil twin means imitating an authorised name, and one device can be both.

for a middle

Explain why SSID and security type prove nothing, how an open network has no server identity, and what the client profile and certificate checks add.

for a senior

Describe classifying heard APs as authorised, neighbouring or unknown, tuning lookalike rules, and correlating with the wired side to decide whether a device is a real backdoor.

for a principal

Decide which client-profile baseline and which inventory-based classification the whole estate enforces, and who may act on a device whose owner is unknown.

## Two words, two different questions A **rogue access point** is defined by *authorisation*: it is an AP the organisation did not approve. NIST SP 800-153 describes the risk as a possible backdoor into the wired network that bypasses perimeter controls such as firewalls, and as a point where clients that connect can have their communications viewed and manipulated. In practice many rogues are harmless in intent: a consumer AP plugged into a wall port in a meeting room. An **evil twin** is defined by *impersonation*: it advertises the name of an authorised network, for example `corp-example`, so that clients which trust the name may join it. It need not touch your wired network at all; a phone sharing its mobile connection can be one. | | Rogue AP | Evil twin | |---|---|---| | Defined by | Not authorised by the organisation | Imitates an authorised SSID | | Question it answers | Is this device ours to run? | Is this the network the client meant? | | Main exposure | Wired-network backdoor, unmanaged settings | Client traffic reaches the wrong party | | Overlap | A rogue can also be a twin | A twin can also be a rogue | ## What a client can and cannot see A client sees what an AP advertises: the SSID, the BSSID (an address carried in the frame) and the security the network offers. All three are labels that any transmitter can repeat, so none of them proves the AP belongs to the organisation. - **Open SSID:** nothing in the join sequence proves who is on the other end. There is no server identity to check. - **WPA2-Personal:** the four-way handshake proves that the AP knows the key derived from the passphrase, so a device without the passphrase cannot finish the join. But every user holds the same passphrase, so it identifies a community, not one AP. The handshake itself is covered with the WPA2 material. - **802.1X Enterprise:** the client's validation of the server certificate is what proves identity. That is covered with admission over the air, not here. - **Wi-Fi Enhanced Open (OWE):** the Wi-Fi Alliance describes it as *unauthenticated* data encryption. It hides traffic from passive listeners and does not tell the client who the AP is. ## Where the weight moves: the client profile Because the beacon proves nothing on an open or shared-secret network, the decision sits with the client: which remembered networks it joins automatically, whether it requires the security type it was configured with, and whether it will join an open network without asking. For managed devices a pushed profile that requires the intended security and forgets guest networks after use is the control. Personal devices depend on the user. ## How monitoring notices them NIST SP 800-153 section 3.3 says continuous monitoring should detect unauthorised WLAN devices, including rogue APs, and impersonation attempts, and that an organisation should be able to locate a detected threat by estimating its position from signal strength at several sensors. A wireless intrusion detection and prevention system typically: 1. Compares every transmitter it hears against an inventory of authorised APs. 2. Sorts the rest into neighbouring networks and unknown devices. 3. Raises an impersonation alert when an unknown transmitter advertises a protected SSID. 4. Estimates where the transmitter is, so staff can go and look. ## The library hotspot Students joined a hotspot named like `guest-example`. Three cases look alike in the helpdesk queue: - **Exact SSID from an unknown transmitter:** an impersonation alert, if an inventory exists. - **A lookalike spelling:** caught only by a similarity rule, and such rules raise false positives. - **A student's phone hotspot:** a neighbour-style device, not on your wired network, with no rogue-AP risk to your LAN. What monitoring "should have caught" therefore depends on the inventory and the rules, and the client-side answer is a profile baseline that does not auto-join names it cannot verify. ## What to do 1. Keep an inventory of authorised APs and SSIDs and feed it to the monitoring system. 2. Push client profiles that require the intended security and forget open networks. 3. Use credentials that identify the server where you can, such as certificate validation on Enterprise networks. 4. Locate before you act: NIST cautions that devices in range may belong to other organisations.

  • A student's phone hotspot with a name close to the guest network appears in the library; is it an evil twin, and would monitoring flag it?
    It imitates a name, so it behaves like a twin to confused users, but it is not on your wired network and has no authorisation question to answer. Monitoring flags it only if an inventory plus a similarity rule exist, and such rules produce false positives. The durable fix is client-side.
  • Why does WPA2-Personal still not tell a client that an AP is the organisation's?
    The handshake proves the AP holds the key derived from the passphrase, so a device without it cannot complete the join. A passphrase shared by a whole campus is known to everyone who has it, so it identifies the community rather than one AP. Per-user credentials with certificate checks identify the server.
  • How do you tell a neighbour's access point from a rogue one on your own wired network?
    From the air you can only classify a transmitter as authorised, neighbouring or unknown. Whether it is wired into your network is what makes it a backdoor, and that needs correlation with the wired side, such as switch port and address data, which radio monitoring alone cannot show.

An SSID is like a name on a shop sign: anyone can hang the same sign on another door, so the sign tells a visitor what the place calls itself, not who actually runs it.

  • Says a rogue AP and an evil twin are the same thing
  • Treats WPA2 in the beacon as proof the network is the organisation's
  • Assumes an impostor off the wired LAN is harmless to users
  • Expects a hidden or unusual SSID name to prevent impersonation
  • Reaches for disrupting any unknown AP without finding out who owns it
open as a page

Why can forged deauthentication or disassociation frames drop clients from a WPA2 network, and why does 802.11w PMF stop it?

level: middleimportance: must knowfreq 42%

basics

~20 s

Legacy 802.11 management frames such as deauthentication and disassociation carry no authentication, so a frame claiming to come from the access point can end an association. PMF protects them, so a receiver that negotiated PMF ignores unprotected ones.

open as a page

What is the WPS PIN design flaw behind the 11,000-attempt figure, and why is switching PIN registration off the answer?

level: middleimportance: should knowfreq 30%

basics

~20 s

A WPS PIN has eight digits, but the last is a checksum and the protocol confirms the PIN in two halves, so at most 11,000 attempts cover it. Lock-down only slows that; the answer is to switch WPS PIN registration off.

open as a page

Users on one floor report a dead WLAN while the access points stay up; how do you tell RF jamming or interference from a deauthentication flood, and what can you do?

level: seniorimportance: should knowfreq 17%

basics

~20 s

Jamming and non-Wi-Fi interference deny service at the radio level, so no cipher or PMF helps; they show as a raised noise floor and heavy retries, whereas a deauthentication flood shows as many session terminations. Locate the source, then work around or remove it.

open as a page

As the lead for a campus WLAN, what do you standardise for monitoring rogue and impersonating APs, floods and jamming, and how far does the response go?

level: principalimportance: should knowfreq 8%

basics

~20 s

Standardise sensor coverage, the event classes NIST lists, an inventory-based classification of authorised, neighbouring and unknown devices, and a response rule: identify and locate before acting, because a device in range may belong to someone else.

open as a page

What is KARMA-style probe answering, and which client-profile settings and monitoring signals limit the exposure?

level: middleimportance: nice to knowfreq 14%

basics

~20 s

A KARMA-style access point answers a client's directed probe request for any network name, so a remembered name seems to be nearby. Forgetting open networks and disabling their auto-join narrows it; monitoring flags one transmitter answering for many names.

open as a page