Because the attacker can inject himself between communicating parties (the STA and AP) in a man-in-the-middle attack, the attacker has the ability to completely control the content of the communications (if encryption and message authenticity are not used); and even if encryption and/or message authenticity is used, the attacker can still deny or delay communications.
This section examines two different problems that occur because of MiM attacks. In the first case, the attacker can hijack, or take over, a session; even when robust authentication and access control are used without encryption. In the second case, an MiM attack eliminates the protection afforded by the use of an encrypted tunnel.
Shortly after the IEEE 802.1x protocol was defined, a large number of users were considering using it for authenticating users at hotspots without using WEP. While in theory this is a good idea, using 802.1x this way didn't solve any problems (Mishra and Arbaugh, 2002). Essentially, the attacker simply waits until the STA is completely authenticated and then sends a forged Disassociate or Deauthentication management frame to the STA. At this point, the STA believes it no longer has a session and attempts to reconnect (the attacker can continue to send forged management frames to the STA, keeping it from establishing a session). The AP, on the other hand, believes there is still a session, and the attacker can now use that session, masquerading as the STA up until a reauthentication event takes place?usually in five minutes.
Finally, we earlier discussed in Chapter 9 the problems with EAP and so we won't duplicate that discussion here.
PEAP was designed to protect the EAP exchange from eavesdroppers (see Chapter 9). There were two reasons for this. The first was to provide privacy and allow users to remain anonymous to eavesdroppers because traffic analysis can be a significant threat in some cases, and the second was to provide protection for the EAP control messages EAP-Success and EAP-Failure. Unfortunately, an easy MiM attack eliminates all of the protection provided by PEAP when anonymous connections are supported.
In the first phase of PEAP, an anonymous tunnel is established between the STA and the AP, with the STA sending an anonymous identity if it likes. If the STA sends an anonymous identity, then it cannot be authenticated. A TLS tunnel is created nonetheless with the anonymous credentials, and phase 2 is started, which is a normal EAP session.
The attack against PEAP works by establishing the MiM prior to phase 1 of PEAP. The attacker establishes two different anonymous tunnels. The first (PEAP phase 1) is with the AP, and second (PEAP phase 1) is with the STA. In the first tunnel with the AP, the attacker masquerades as the STA, and in the second tunnel the attacker masquerades as the AP. The STA now begins phase 2 and the attacker sees the true identity information of the STA as well as having the ability to forge EAP control messages?just as if PEAP were not being used (Asokan et al., 2003).