Preauthentication

Preauthentication is used to speed up association transfer. Authentication can often cause a lag between the time a station decides to move to a new AP and the time that the frames start flowing through that AP. Preauthentication attempts to reduce the time by getting the time-consuming authentication relationship established before it is needed. Due to the overloading of the term "authentication" by both the low-level 802.11 authentication and the 802.1X authentication, there are two different types of preauthentication. As it is commonly used by network engineers, though, it usually refers to the 802.1X authentication.

802.11 Preauthentication

Stations must authenticate with an access point before associating with it, but nothing in 802.11 requires that low-level authentication take place immediately before association. Stations can 802.11-authenticate with several access points during the scanning process so that when association is required, the station is already authenticated. As a result of preauthentication, stations can reassociate with access points immediately upon moving into their coverage area, rather than having to wait for the authentication exchange.

In both parts of Figure 8-7, there is an extended service set composed of two access points. Only one mobile station is shown for simplicity. Assume the mobile station starts off associated with AP1 at the left side of the diagram because it was powered on in AP1's coverage area. As the mobile station moves towards the right, it must eventually associate with AP2 as it leaves AP1's coverage area.

Figure 8-7. Time savings of preauthentication

Preauthentication is not used in the most literal interpretation of 802.11, shown in Figure 8-7 (a). As the mobile station moves to the right, the signal from AP1 weakens. The station continues monitoring Beacon frames corresponding to its ESS, and will eventually note the existence of AP2. At some point, the station may choose to disassociate from AP1, and then authenticate and reassociate with AP2. These steps are identified in the figure, in which the numbers are the time values from Table 8-1.

Table 8-1. Chronology for Figure 8-7

Step	Action without preauthentication: Figure 8-7 (a)	Action with preauthentication: Figure 8-7 (b)
0	Station is associated with AP1	Station is associated with AP1
1	Station moves right into the overlap between BSS1 and BSS2	Station moves right into the overlap between BSS1 and BSS2 and detects the presence of AP2
1.5		Station preauthenticates to AP2
2	AP2's signal is stronger, so station decides to move association to AP2	AP2's signal is stronger, so station decides to move association to AP2
3	Station authenticates to AP2	Station begins using the network
4	Station reassociates with AP2
5	Station begins using the network

Figure 8-7 (b) shows what happens when the station is capable of preauthentication. With this minor software modification, the station can authenticate to AP2 as soon as it is detected. As the station is leaving AP1's coverage area, it is authenticated with both AP1 and AP2. The time savings become apparent when the station leaves the coverage area of AP1: it can immediately reassociate with AP2 because it is already authenticated. Preauthentication makes roaming a smoother operation because authentication can take place before it is needed to support an association. All the steps in Figure 8-7 (b) are identified by time values from Table 8-1.

802.11i Preauthentication and Key Caching

When a network is authenticated with 802.1X, the most time-consuming step in getting from the 802.11 join to the ability to send network protocol packets is the 802.1X authentication, especially if it uses an EAP method with several frame round-trips. Preauthentication, shown in Figure 8-8, allows a station to establish a security context with a new AP before associating to it. In essence, preauthentication decouples the association and security procedures, and allows them to be performed independently. WPA explicitly excluded preauthentication.

Figure 8-8. 802.11i preauthentication

Figure 8-8 shows the following sequence of steps.

1. The station associates to the first access point it finds on the network. It selects this AP based on the criteria in its firmware.
2. Once associated, the station can perform an 802.1X authentication. This step uses EAPOL frames as described in Chapter 6, with an Ethertype of (hexadecimal) 88-8E. EAPOL frames are converted into RADIUS packets by the AP, and the session is authenticated.
3. Dynamic keys for the radio link are derived on both sides through the four-way handshake for pairwise keys and the group key handshake for the group keys.
4. With keys configured, the station is "on the air" and can send and receive network protocol packets.

   Station software is in control of roaming behavior, and can use that to its advantage. As the station moves in such a way that AP2 appears to be a better choice, it can perform preauthentication to speed up the process of moving over to AP2. Rather than move everything all at once, though, it performs preauthentication to cut down on the interruption between sending network packets.

5. Preauthentication commences with an EAPOL-Start message sent from the station to the new AP. A station can only be associated with a single AP, so the preauthentication frames are channeled through the old AP. Preauthentication is a complete 802.1X exchange.
   1. Preauthentication frames use the Ethertype of (hexadecimal) 88-C7 because most APs apply special processing to the regular authentication Ethertype. The source address of the frames is the station, the receiver address is the BSSID of its current AP (in this case, the MAC address of AP1's wireless interface), and the destination address is the BSSID of the new AP (in this case, AP2's wireless interface).
   2. When received by AP1, the frames are sent over the distribution system to AP2. The AP only has a MAC address. If the two APs are not connected directly to the same Ethernet broadcast domain, they must have an alternative method of shuttling preauthentication frames between devices.
   3. During this entire step, the station remains associated to AP1, and can send and receive network packets through its existing encrypted connection. Because the station is still on the air, there is no apparent authentication occuring.
   4. The result of the preauthentication is that a security context with AP2 is established. The station and AP2 have derived a pairwise master key, which can be further processed to create keys between the station and AP2. Both the station and the AP store the pairwise master key in a key cache.
6. When the station pulls the trigger, the association is moved to AP2. As part of the initial association, the station includes a copy of its key cache to tell AP2 that it was already authenticated.
7. AP2 receives the authentication request and searches its key cache. Finding an entry, it starts the fourway pairwise key handshake immediately. By proceeding to key derivation, the station is unable to send and receive packets for only a short time.

802.11 preauthentication moves the time-consuming 802.1X EAP method to occur in parallel with sending and receiving network frames on an authenticated connection. The first association will be slow because the full EAP exchange is required. On subsequent associations, however, preauthentication can dramatically reduce handoff times.