Fast authentication methods for handovers between IEEE 802.11 wireless LANs

Reviewer: Jorge Jose Ortiz

With the increase in wireless connectivity in public and private spaces comes new applications, dependent on a ubiquitous connection (for example, voice over Internet protocol (VOIP) and streaming video). In order to create ubiquitous connectivity, however, the wireless connection between mobile nodes (MNs) and access points (APs) must be handed over from AP to AP as the MN moves from one place to another. This paper elaborates on the security and efficiency of the handover process, and proposes changes in the IEEE 802.11 handover protocol on the link layer, and the Internet Engineering Task Force (IETF) Seamoby reconfiguration and network discovery protocol on the network layer. The paper starts by explaining the trust model, and the security relationships between the entities involved. The entities involved include APs, MNs, access routers (ARs), and authentication servers (ASs). There are implicit and explicit trust relationships, and the concept of old entities and new entities. The old entities are those connected to the MN currently, and the new entities are those that the MN moves to afterward. The implicit relationships exist between the MN and the new/old APs, and the MN and old AR. The explicit relationships exist between old/new AS servers, the MN and old AS, and the old/new AP and the old/new AR, respectively. Based on this trust model, the paper goes on to describe the associated handover process. Essentially, the process leverages the relationships set by the trust model, to relay sensitive information about MNs and APs for intra- and inter-domain handover and authentication. In providing an overview of the handover process, the paper essentially breaks down how the trust relationships are established through a distribution of keys. On the link layer (L2), the MN has an explicit trust relationship with the AS. They use a scheme to establish a master key (MK), and from this they derive a pairwise master key (PMK) that is forwarded to the AP that the MN is currently trying to establish an initial connection with. The MN and the AP then derive a pairwise transient key (PTK) to establish secure communication between them. They then establish a group transient key (GTK) that is handed out to all associated APs. This establishes trust between the MN, AS, and APs. The network layer (L3) trust is established from the trust between the MN and AP. When the MN moves to another AP (nAP) from the old AP (oAP), the MN establishes trust with the nAP from the GTK that was sent to it by the oAP, and then it re-associates with the nAP. Trust with the nAR is established through transitivity (for example, MN trusts nAP, nAP trusts nAR, and MN trusts nAR). Fast authentication is described in the section following the protocol section. One of the proposed techniques was proactive key distribution, whereby the AS calculates the nearest neighbors to the current AP the MN is connected to. In summary, the AS re-calculates PMKs for all of the oAP's neighbors. That way, when the MN goes to a nAP, they can skip the step of obtaining a PMK from the AS, as was described in the previous paragraph. The paper notes that this sped up the handover process from 1.1 seconds to 25 milliseconds. The next proposed solution is preauthentication. Basically, the MN authenticates itself to a nAP through the oAP. That way, the MN is now authenticated to the nAP while still connected to the oAP. When the MN finally moves to the nAP, they can start by deriving the GTK, and the MN can continue this procedure as it moves from AP to AP. Predictive authentication is another technique for fast authentication. Here, the AS sends the PMK response to multiple APs, in the location where the MN's home AP is detected to be. The process is similar to the nearest neighbor method in proactive key distribution. These fast authentication techniques are then expanded to inter-domain handovers, by creating a trusted relationship between APs of different domains. In intra-domain handovers, the APs inherently trust each other, as they all trust the same AR and/or AS. The problem in inter-domain handovers is that everything changes. The APs, ARs, and AS all change. If you can create a trust relationship between the ASs of different domains, then the same fast authentication techniques can be used. Some architectural changes are proposed as well, and are further expanded in the last section of the paper. Overall, the paper is generally well written, but difficult to follow. In the paper, as is probably true of this review, the use of acronyms and abbreviations make the procedures and protocols hard to understand at first glance. I read the paper twice before I started to get a clear idea of how the protocols worked. In summary, the organization of topics is well constructed, and the techniques are clearly explained. The use of abbreviations, while seemingly adding to the reading difficulty, actually does the opposite. Surely, if the terms were expanded, the paper would be tougher to understand. The future work section does not really go into what future work can be done, and what direction research is taking. Last, it would be interesting to test these techniques, and see their affect on real applications like VOIP. Online Computing Reviews Service