XML-aware data synchronization for mobile devices

Abstract

In everyday life, and when using computer systems in particular, it is sometimes the case that a logical datum is replicated into multiple copies, such as when we send a document by electronic mail, or inform interested parties of a new address of residence. If the datum for some reason changes, we would then also like the changes to be reflected in the copies. The problem of keeping the copies up-to-date with respect to each other is studied under the heading of data synchronization. In this thesis, we address data synchronization for mobile devices with limited energy resources and limited connectivity to the Internet, such as mobile phones. The importance of data synchronization is emphasized here, as it becomes infeasible to communicate continuously and in high volumes about the current state of each copy. The established conventions of the Internet and mobile computing environments on such matters as storage interfaces and data formats define an overall system architecture, into which we as seamlessly as possible want to incorporate our proposal. By focusing on interoperability we lower the threshold for utilizing our research in practice. We present a comprehensive approach to data synchronization for mobile devices that is optimistic and state-based, and which targets opaque and XML files on a standard file system. We consider how to use the available connectivity in an economical manner, and so that existing sources of data on the Internet can be utilized. We focus on XML synchronization, where we identify an opportunity to utilize the structure of the data the format exposes. Specifically, we present an algorithm for merging concurrent changes to XML documents which supports subtree moves, an efficient heuristic algorithm for computing tree-level changes between two XML documents, and an overall architecture and algorithms to support the use of lazily instantiated XML documents. Our data synchronization approach is evaluated quantitatively in several experiments, as well as qualitatively by constructing applications that build on top of the approach. One of our applications is an editor that processes 1 GB XML files on a mobile phone.