Dividing point value selections for Improved Quick Processor-demand Analysis

Earliest Deadline First (EDF) is an optimal scheduling algorithm for uniprocessor real-time systems. Improved Quick Processor-demand Analysis (QPA*) provides efficient and exact schedulability tests for EDF-scheduled systems with arbitrary relative deadlines. The values of the dividing points can significantly affect the performance of QPA*. In this paper, we provide an efficient approach based on experiments to find suitable values of the dividing points; we also investigate the best value of the dividing points by extensive simulations on a large number of randomly generated task sets. We show that the proposed approach can significantly reduce the required calculations to perform a dividing point´s value selection.