A Linear Programming Approach for Deadline Assignment of Update Transactions in Co-scheduling Environment

Deriving deadline and period for update transactions has long been recognized as an important problem in real-time database research. Despite years of active study, the state of the art only focuses on scheduling of update or control transaction separately. In this paper, we study the problem of deriving deadlines for update transactions in co-scheduling environment upon uniprocessor platforms. We consider how to compute deadlines for update transactions to minimize the total workload of the transaction set, while maintaining the temporal validity of real-time data objects and guaranteeing the EDF schedulability of the hybrid transaction set. To address this problem, we propose a linear programming approach to derive deadlines for update transactions, namely DC-LP (Deadline Calculation based on Linear Programming). Finally, the success ratio, workload and time performance of DC-LP are evaluated through extensive simulation experiments.