Doctoral Dissertation Defense: On Peer-to-Peer Data Management in Pervasive Computing Environments

The deployment of wireless communication technologies has led to significant research in the area of mobile data management. The prior research is dominated by the client-proxy-server model. In this model, mobile devices are able to connect to the Internet and serve as clients, initiating actions and receiving information from servers on the network. The primary focus of prior research is on the development of techniques and protocols that deal with disconnection management, low bandwidth and device resource constraints. Their aim is to allow applications built for the wired world to run in wireless domains using proxy based approaches.

In environments based on the cellular infrastructure or wireless local area networks, the traditional client-proxy-server interaction is perhaps an appropriate model. The widespread use of short-range ad-hoc networking technologies, however, introduces different type of environments -- the pervasive computing environments -- that require novel data management mechanisms. A pervasive computing environment represents a resource-rich, data-intensive scenario where users and devices, including handhelds, wearables, computers in vehicles, computers embedded in the physical infrastructure, and (nano)sensors, are mobile and continuously exchange data. In order to exploit the available resources, mobile devices must act as semi-autonomous, self-describing, highly interactive and adaptive peers that employ cross-layer interaction between their data management and communication layers for inferring and expressing information they need, and for obtaining and storing such information by pro-actively interacting with other devices in their vicinity using available short-range ad-hoc networking technologies.

In order to address the data management challenges introduced by these data-intensive, highly dynamic, pervasive computing environments, this dissertation focuses on three questions: (i) What is the necessary model that will allow a device to infer and express what information its user needs based on the current context? (ii) How should a device determine when a user will need the information? (iii) How should a device effectively obtain and store such information from other devices in its vicinity?

The contribution of this dissertation is the design of a conceptual model for data management in pervasive computing environments. The conceptual model includes novel protocols relying on cross-layer interaction between data management and communication layers. The dissertation also defines a novel data-based query routing algorithm, a semantic-driven data caching and replication algorithms employing user's and device's profiles, a collaborative query processing protocol, a neighborhood-consistent transaction model, and a reputation-driven belief model.

The second contribution of this dissertation is the design and implementation of a prototype of the conceptual model - the MoGATU framework. Through experimental validation of the implementation prototype, the dissertation demonstrates that the conceptual model and its implementation are superior to existing data management techniques. The dissertation shows that the conceptual model improves the process of discovering and routing data across wireless ad-hoc networks. The dissertation also shows that the conceptual model increases data availability and transaction success rates. Additionally, the dissertation shows that the conceptual model enables devices to evaluate the integrity of their peers and the accuracy of peer provided information.