Trust management in mobile wireless networks: Security and survivability.

摘要

In this dissertation we study the security and survivability of wireless mobile network systems in two distinct threat models: the Byzantine threat model and the selfish node threat model.; Wireless mobile networks are collections of self-organizing mobile nodes with dynamic topologies and have no fixed infrastructure. Because of their dynamic ad hoc nature, these networks are particularly vulnerable to security threats. The security of such systems is, to a large extent, based on trust associations. There are several ways in which trust can be supported in a network system. The way we adopt is to establish a secure public key management infrastructure (PKI). This enables basic cryptographic functionalities, such as integrity, privacy, etc. However, due to the dynamic character of a wireless mobile network and its ad hoc topology changes, the trust associations cannot depend on any pre-established trust relations and must support a flexible, uncertain and incomplete trust model.; One of our main goals in this dissertation is to analyze the distributed nature of trust in wireless mobile networks and to consider approaches that manage trust based only on locally available information. In our analysis for this problem we use the traditional Byzantine attack model. After reviewing the trust models proposed in the literature we propose an extension that supports a distributed trust management infrastructure. In this model the trust is distributed horizontally via multiple disjoint trust flows. Compared to the traditional hierarchical trust distribution, our approach is appropriate for dynamic wireless systems for which there are no central trust authorities.; A second goal is to manage trust based on the good behavior of nodes. Mobile wireless networks rely heavily on node collaboration. However, since the nodes are often battery powered, they may behave selfishly to preserve power. The threat model for this application is restricted to selfish node attacks.; We present a simple and efficient reputation system, Locally Aware Reputation System (LARS) that mitigates selfish node behavior. We explore methods that stimulate node cooperation in mobile wireless networks, and analyze the reputation systems proposed in the literature. The performance of LARS is evaluated in terms of its packet delivery ratio, its end-to-end delay and its overhead, and compared to the other reputation systems proposed in the literature.; Finally, to enhance the security and survivability of wireless mobile networks against selfish threats, we propose a mechanism that will trace selfish node behavior.; Chapter 4, 5 and 6 are the main contribution of the dissertation.