Research On Scheduling Algorithms Of Temporal Validity Data In Real-time Database System

The real-time database system is a modern database system combined with theory of database and real time technology. As the core of modern information system, it has been widely used in the field of production and process control, power system, air traffic control, mobile communications and other time critical areas.The value of the data object in the real-time database has validity interval. The current real-world state can be truly reflected by this data object value whithin its validity interval and can not be reflected if the time is beyond the data object’s validity interval. The real-time data object value should be updated in time to keep its validity, but too frequent refreshment will cause heavy load of the system. Therefore, one of the core issues of the real-time dabase sysem domain to be researched is how to keep the data object value temporally valid when the scheduling algorithm executed while the system load is reduced as far as possible.The usually used scheduling algorithms of real-time syetem such as RM,DM, EDF and LST are researched in this thesis. Based on this research, several algorithms about real-time data updating transaction such as HH, ML and DSFP are analyzed and studied. A novel approach, named FPDS-EDF, is proposed by combination of the delayed scheduling idea in DSFP algorithm and the periodically calculating method in ML algorithm. And then, the related performance of ML, DSFP and FPDS-EDF are anlyzed and compared. The experiment results shows that, FPDS-EDF algirithm is better than ML in system resource utilization and it is better than DSFP in the calculation comlexity of time consume.