Provenance-Based Information Trust in Wireless and Distributed Networks

With the advancement of networking and information technologies, dynamic wireless network environments are often used by individual people, business corporations as well as governmental and military organizations to obtain information from different sources and share information with other entities. In such networking environments, information transmissions and sharing are the essential activities. Information from different sources makes it possible to extract more accurate and complete knowledge and thus support more informed decision making. Hence, a trust analysis system to assess the quality and trustworthiness of information for the entire network is crucial. It will not only allow us to assess the quality of information, but also enhance the overall security level of the network.;Information can be generated from multiple source nodes, e.g., stationary sensors or mobile devices. It may then go through a series of intermediate nodes before reaching the destination. In order to assess the information trust, we need to take the provenance (i.e., derivation history) of the information into consideration. The provenance of information plays a very important role in a dynamic network environment. Not only it can be used for information trustworthiness assessment, but also it is crucial for copyright clearance, data reconciliation, and data replication. In this work, we investigate the relation between provenance and trust as well as their security and privacy implications in various mobile and distributed networking scenarios.;We begin our study by exploring the security properties of provenance meta-data in a distributed network environment. We introduce a "chain-structure" provenance scheme to provide security assurance for provenance meta-data in three dimensions - confidentiality, integrity and availability. Based on secured provenance, we first present a trust model to evaluate the trustworthiness of information as well as information publishing entities. Taking a step further, we consider an information trust metric with three dimensions: correctness, completeness, and timeliness. Based on such a trust metric, we propose a decision fusion framework that considers the provenance dependence of information items to enhance the fusion reliability. On top of trust assessment based on multiple data items that describe a single event or observation, we propose the concept of Provenance Logic, to reason about the logical relations between multiple events by jointly recognizing and linking events from successive observations in the network.;The identities of the information sources or subsequent owners often need to be recorded and validated in provenance. However, this violates the anonymity requirements of many applications. To study this problem, we focus on mobile sensing applications as they deal with a lot of personal information, e.g., users' locations at a specific time, and thus need deeper attention to user privacy. We propose ARTSense (Anonymous Reputation and Trust for Participatory Sensing), which is a provenance-based trust and reputation framework to solve the problem of "trust without identity" in mobile sensing.;Besides provenance that describes the ownership history of the data, our final step is to study the spatial-temporal provenance, i.e., the location history of mobile users. We propose the STAMP (Spacial-Temporal provenance Assurance with Mutual Proofs) protocol, which is based on co-located mobile devices mutually generating location proofs for each other. STAMP ensures the integrity and non-transferability of location proofs and maximize users' anonymity and location privacy.;We have done analytical and empirical validations for all of the above studies, which have shown that the proposed solutions in this work achieve our pre-defined security and privacy objectives with a satisfactory performance.