| In recent years, the development in wireless communication technologies of portable computing devices has created a new class of applications, called mobile computing. As mobile computing becomes more commonplace, there will be an increasing demand for wireless access to data stored in a database. Data management in mobile environments has many new challenges due to the unique features of mobile computing, such as asymmetric communication bandwidth, limited battery power of mobile units and error-prone wireless networks. Though much research has been done in different areas of mobile data management, some areas, such as concurrency control, have received little attention.; My dissertation is mainly focused on the areas of concurrency control and adaptive data broadcasting in a mobile environment. Two concurrency control algorithms—the virtual lock and optimistic concurrency control with dynamic adjusting timestamp ordering, are proposed. The virtual lock algorithm is used for transactional systems where a transaction's data access pattern is known before the transaction begins. On the other hand, the optimistic concurrency control with dynamic adjusting timestamp ordering is used for transactional systems where a transaction's data access pattern is unknown beforehand. The second part of my dissertation focuses on adaptive data broadcasting. It studies overhead of data request messages in an adaptive broadcasting environment. Existing adaptive broadcasting algorithms ignore this overhead, which we believe is particularly important in mobile environments. A new data broadcasting approach called lazy data broadcasting is proposed. Through experimental study, we find that a significant number of data request messages can be saved using the lazy data broadcasting approach. |
