A tool for controlling response time in real-time systems

In hard real-time systems, classical scheduling policies only cope with satisfaction of deadline constraints. In this paper, to every periodic task is associated a weight that models the importance of the task in terms of worst-case response time. These parameters are set off-line by the designers of the real-time software in order to control the quality of the on-line schedule. According to these weights, a set of feasible fixed-priorities are computed so that the mean weighted response time of the tasks is minimized. We propose a branch and bound algorithm to solve this problem. An example is completely detailed and numerical results on randomly generated problems are lastly presented to show the efficiency of the developed tool.