A Study On Data Sharing Over Mobile Devices Based On Dis-tributed Shared Memory

With better performance and lower cost, mobile devices are playing more impor-tant roles in people’s work and entertainment, and slowly changing people’s lifestyle. People usually tend to share data between different mobile devices and need efficien-t management of their data in daily use. However, most existing tools often require human intervention to achieve data sharing and synchronization which increases user-s’burden. Different from traditional computers, mobile devices have their inherent characteristics. First, mobile devices are more portable and can be used anytime and anywhere. Second, mobile devices have lower performance than traditional comput-ers due to their little size. Third, existing battery technology can’t guarantee the long running of mobile devices. So it’s very hard to get satisfactory performance on mobile devices with traditional technology of data sharing.Distributed shared memory(DSM) is a model that provides the illusion of a shared memory on top of a message passing system. In this model, processes running on sep-arate nodes can access a shared memory address space. By avoiding the complexities of message passing, it allows users to use shared data without regard to the specific data synchronization process, thereby making it much easier to manage their data. To solve the problem of mobile devices and reduce the burden on users in data sharing and synchronization, this thesis focuses on the technology of data sharing over mobile devices based on distributed shared memory. More specifically, the main contribution of this work includes:Firstly, the design of DOMO, a data sharing system over mobile devices based on distributed shared memory, is presented. It isolates users from complex underlying message passing and enables efficient data sharing and synchronization over mobile devices.Secondly, DOMO separates file’s meta data from its content. It uses a tree struc-ture to manage and monitor shared data effectively and synchronizes those nodes of the tree. With meta data and version data, DOMO allows devices to work even if disconnection occurs and continue synchronizing data when devices are reconnected.Thirdly, logical time is adopted to cope with the asynchrony of the distributed system consisting of mobile devices. Based on logical time, methods to maintain the consistency of data and conflict management are presented. Further more, DOMO im-plements two consistency maintenance strategies:update after write(UAW) and update before read(UBR) as well as one conflict resolution strategy based on renaming files.Finally, we have implemented a prototype of DOMO on the Android platform and comprehensive experiment evaluations with UAW and UBR are conducted. The evaluation results show that DOMO has good performance under different loads and can deal with the disconnection between devices.