Study Of Mobile Network Database Access Based On User Behavior Prediction

Together with the rapid developing of Mobile Computing and the widely use of Smart Device, Mobile Network Database System (MNDB), which is known as a new form of the Database System, is being used in more and more applications. However, the long Round Trip Time (RTT) and low throughput of the mobile network largely impact the performance of MNDB, which is restricting its further use.In this thesis, we present a design, which is able to improve the performance of mobile database access over mobile networks. This is accomplished by applying the theory of behavior prediction into database access. Behavior prediction provides the foreknowledge of user's following operations, and it is possible for mobile to get the required data from remote database even before receiving the request from user, which results in a smaller response time for user and gives a better user experience even in the face of weak network connectivity. Content Addressable Storage (CAS) is also being used to reduce data transmission volume from remote database, which will give a further decrease on the response time.We implemented the proposed mobile database access method on Android mobile platform. The experiment results show that our method decreases the response time while keeping the data consistency, compared to the traditional access method.