Tech Report: Best-Harmonically-Fit Best-Harmonically-Fit Periodic Task Assignment Algorithm on Multiple Periodic Resources

AS computer hardware technology advances, both the number and the computational capabilities of processors used in real-time systems are increasing. To take advantages of the increased physical resource’s capacity, multiple groups of real-time applications are deployed on the same physical platform. However, since each group of real-time applications may have different time granularity [1], when multiple groups of applications share the same processors, traditional realtime system models and theoretic results may not be sufficient or even applicable to guarantee that each group of real-time applications will not have time interference among each other and that they will satisfy their timing requirements. In order to study the issues of scheduling multiple groups of real-time applications on the same physical resources, the concept of virtual real-time resources is proposed [1]. Virtual real-time resources are an abstraction of physical resources where the physical resources are shared by real-time application groups [1]. With the concept of virtual real-time resources, each group of real-time applications has its own isolated and independent virtual resource, hence avoiding interference among different groups of real-time applications. However, from an application perspective, virtual real-time resources are not continuously available to the application. Instead, virtual resources are periodic, i.e., they periodically provide certain amount of processing capability to the applications [1], [2], [3], [4], [5], [6], [7]. The study of periodic resources can be traced back to 1999 when the concept of periodic resource was first formally defined [4]. It has recently drawn more attention in the community [1], [7], [8], [9], [10], [11], [12], [13], [14]. However, until now, most of the studies about periodic resources focus on task set schedulability analysis on a single periodic resource. There has not been much work, if any, in the literature dealing with the task assignment problem on multiple periodic resources. In this paper, we study the task assignment problem in the context of assigning multiple periodic tasks to multiple periodic resources. The main goal is to decide on a task assignment strategy so that the resource capacity provided by every periodic resource is maximally utilized.