Secure Communications for Ad Hoc Networking

Cornell University, Ithaca NY

Investigators

Abstract

An ad hoc network is communication paradigm that does not rely on a pre-existing communication infrastructure. Rather, the network is established on-demand. Thus, the network needs to be deployed without prior planning and needs to operate in a variety of different and differing network propagation conditions, traffic demands, and user mobility patterns. Nodes in an ad hoc network frequently join and leave the network, without prior notice. Furthermore, the nodes are mobile and communicate by accessing a shared wireless channel. Consequently, the topology of the network is constantly changing i.e., connections between neighbor nodes (nodes that can communication directly) are established and torn down all the time. Since the diameter of the network may be significantly larger than the transmission range of a single node, communication between non-neighbor nodes is by multi-hop routing. Probably the most characteristic feature of ad hoc networks is that there is no central entity in the network. This is radically different than most other wireless networks, especially the cellular networks. Thus, nodes in an ad hoc network rely on peer-to-peer communications. Consequently, all network protocols need to be implemented in a distributed manner and need to be extremely efficient, as to ensure that the overhead caused by the frequent reconfiguration of the network is not prohibitively large. Application of ad hoc networks has been for a long time limited to the military communication environment only. However, recently, the commercial sector is increasingly interested in this technology, evidence of which is the establishment of the Intenet Engineering Task Force, MANET Working Group. ALthough it is still not clear what the main application of the commercial ad hoc network will be, applications for LAN, sensor networks, and virtual classrooms have been proposed. One of the main difficulties in promoting the concept of ad hoc networks is that, inherently, ad hoc networks are extremely vulnerable to security attacks. Not only do these networks rely nearly exclusively on unreliable and insecure wireless communication, but also the nodes in the network do not, in general, belong to the same administrative domain. Thus the standard security measure may not be readily applicable to ad hoc networks. But above all, the lack of centralized entity makes design of security protocols that much more difficult. In particular, due to possibly low availability of any network node, no one node can serve as a trustful entity in the network. Moreover, as an ad hoc network may consist of thousands to tens of thousands of nodes, it is essential that any solution be highly scaleable. Furthermore, it is imperative that the security protocols address the highly dynamic communication environment, for example, by frequent rekeying of the system. Finally, because of the nature of the nature of these networks and their intended application, high degree of reliability and survivability to node and link failures is necessary. Fortunately, many protocols that protect against intentional security attacks also guard the system again (unintentional) system failures. As part of the proposed research, the researchers plan on investigating a number of central schemes that will allow secure communication in the ad hoc networking environment. Secure communication needs to address the following attributes: availability, confidentiality, integrity, authenticity, and non-repudiation. Some of these attributes can be addressed through conventional security protocols, such as ensuring privacy of information through encryption. However, for other attributes, the standard security measures do not suffice. For example, authentication in an ad hoc network cannot rely on a signle central trustful entity. The researchers propose to use two main principles in securing the communication in an ad hoc network: use of redundancies in the network topology and distribution of trust. The main idea is that availability can be improved by relying on multiple routing paths in the network. So, if compromised paths can be eliminated, other paths can be used reliably. The second principle - the distribution of trust - is based on the assumption that as no single node is worth unlimited trust of other network nodes. However, a large enough quorum of nodes can be trusted, as the probability of compromising large number of nodes simultaneously is rather small. This research proposes to investigate the above mentioned issues of supporting secure communications in the ad hoc network by proposing and designing fundamental framework for such a communications. The researchers will use both analysis and simulation to evaluate our protocol and we intend to implement these protocols to demonstrate their behavior in a practical network.