Regularity-Based Partitioning of Uniform Resources in Real-Time Systems

Hierarchical scheduling is a hot topic in realtime systems. In a hierarchical real-time system, the resource partition is the intermediate level between physical resources and real-time tasks. A resource partition operates on the shared physical resources at a fraction of the rate, and serves as a scheduling interface between the lower-level real-time tasks and the shared physical resources. Thus a key problem is how to define this scheduling interface on resource partitions. Regularity-bounded methodology is one important type of resource partitioning algorithms. This paper extends Mok and Feng's Regularity-based Resource Partition Model from a single-resource platform to a uniform multiresource platform. We present a resource partitioning algorithm called AAF-Multi Scheduling for solving the time slice overlap problem on a multiresource platform without violating the schedulability bound given by Feng on a single-resource platform. AAF-Multi is a global scheduling algorithm with O(Ω · log Ω) time complexity (Ω = resource amount × hyper period), where hyper period is the least common multiple of the periods of the resource partitions.