Research On Real-time Data Broadcast Scheduling And Index Organizing Strategy

In the environment of embedded mobile real-time database management system (EMRTDBMS), data broadcast technology is provided to support a large number of mobile clients accessing data in server concurrently. Data broadcast efficiently utilize the bandwidth asymmetry of mobile network, broadcast periodically the popular data and solve the problem of numerous clients availably. Data broadcast research mainly focuses on broadcast mode, broadcast scheduling, broadcast index and broadcast structure. Among these topics, most research concern the broadcast scheduling and broadcast index.Data broadcast scheduling includes traditional data broadcast scheduling and real-time data broadcast scheduling and time constraints should be especially considered in real-time data broadcast scheduling. Broadcast index consists of tree index, hash index, hybrid index. Tree index can efficiently reduce tuning time and bring on much research.Most real-time data broadcast scheduling are based on priority. The fair scheduling for real-time data broadcast is designed from the angle of data productivity. It is called FS-RDB that incorporates the urgency, the number of pending requests and fairness. To reduce the computation cost of data productivity, we give the approximate calculation.Broadcast index is designed to decrease the tuning time and to reduce the power consumption amount. NOSIT (Nearly Optimal Search Index Tree) is designed through the analysis of tuning time formula. As the target is almost the same with nearly optimal search tree, we can design the similar algorithm. The middle node also can index data item, so the space consumption is reduced and access time is decreased.We implement a prototype system of data broadcast and apply the designed FS-RDB and NOSIT. Results show that DRSR (Data Request Success Ratio) of FS-RDB is higher than EDF-T and PRDS and TSR (Transaction Success Ratio) of FS-RDB changes smoothly with transaction interval. The tuning time of NOSIT is lower than B+ tree, especially in declined data access probability.