Group Key Generation in Ad Hoc Wireless Networks Using a Subgroup Method

Most algorithms for Group Key distribution were not created with ad hoc networks in mind. In a wired network with a distinct infrastructure, generating a key distribution algorithm is based on a static environment and, therefore, more focus is placed on efficiency and security within those confines. With an ad hoc network, however, the focus must switch to the dynamic nature of the group and how to deal with the problems which it creates. These problems, which include a lack of infrastructure and link failures between nodes, have been identified in a few protocols, including CLIQUES, the Burmester-Desmedt suite, and the YTCC protocols. These protocols, however, still need to rekey in a lengthy process when a link goes down within the group. Although AGKE has made this rekeying scalable, this process can still cause communication to be delayed for an extended period of time. The solution to this problem proposed in this thesis is based on a subgroup method. This solution is based on a specific type of ad hoc network where the nodes are already organized into subgroups (such as a military operation, where nodes are broken into platoons). These subgroups choose a controller and are then organized as a tree based on the connections between them. Using this tree as a backbone, communication takes place between nodes in the group by using subgroup keys. This subgroup method also ensures that if a link is broken in the group, only that part of the tree needs to be re-keyed, providing a shorter down-time for the group.