The scheduling of systems of periodic tasks upon multiprocessor platforms is considered.

Output jitter --- the variation in the inter-completion times of successive jobs of the same task --- is studied in the context of the preemptive uniprocessor scheduling of periodic real-time tasks. A formal quantitative model for output jitter is proposed. A list of properties that are desirable in any jitter-minimization schedule is enumerated. Algorithms are presented for generating suchschedules, and bounds proved for the maximum jitter in schedules generated by these algorithms. Keywords. Scheduling: uniprocessor, preemptive# periodic tasks# output jitter. 1 Introduction In many real-time control applications, periodic activities represent the major computational demand on the system. Such periodic activities typically arise from sensory data acquisition, low-level servoing, control loops, action planning, and system monitoring, which need to be cyclically executed at specific rates (derived from the application requirements). These periodic activities are often formally m...

The issue of temporal fairness in periodic real-time scheduling is considered. It is argued that such fairness is often a desirable characteristic in real-time schedules. A concrete criterion for temporal fairness -- pfairness -- is described. The weight-monotonic scheduling algorithm, a static priority scheduling algorithm for generating pfair schedules, is presented and proven correct. A feasibility test is presented which, if satisfied by a system of periodic tasks, ensures that the weight-monotonic scheduling algorithm will schedule the system in a pfair manner.

Current feasibility tests for the static-priority scheduling of periodic task systems run in pseudo-polynomial time. We present a fully polynomial-time approximation scheme (FPTAS) for feasibility analysis in static-priority systems where each task’s relative deadline is constrained to be at most its period. This test is an approximation with respect to the amount of processor capacity that must be “sacrificed ” for the test to become exact. We show that an arbitrary level of accuracy, ɛ, may be chosen for the approximation scheme, and present a run-time bound that is polynomial in terms of ɛ and the number of tasks, n. Keywords: Real-time scheduling; Uniprocessor systems; Static-priority systems; Feasibility analysis; Approximation algorithms.

The proliferation of new data-intensive applications in asymmetric communication environments has led to an increasing interest in the development of push-based techniques, in which the information is broadcast to a large population of clients in order to achieve the most efficient use of the limited server and communication resources. It is important to note that quite often the data that is broadcast is time-critical in nature. Most of the related current research focuses on a pure push-based approach (Broadcast Disks model), where the transmission of data is done without allowing explicit requests from the users. More recently, some bidirectional models incorporating a low-capacity uplink channel have been proposed in order to increase the functionality of the Broadcast Disks model. However, the impact of integration of the uplink channel has been investigated using only static client profiles or ignoring the existence of timing constraints associated with data. None of the existing...

Abstract—Each processor in a uniform multiprocessor machine is characterized by a speed or computing capacity, with the interpretation that a job executing on a processor with speed s for t time units completes ðs tÞ units of execution. The earliestdeadline-first (EDF) scheduling of hard-real-time systems upon uniform multiprocessor machines is considered. It is known that online algorithms tend to perform very poorly in scheduling such hard-real-time systems on multiprocessors; resource-augmentation techniques are presented here that permit online algorithms in general (EDF in particular) to perform better than may be expected given these inherent limitations. It is shown that EDF scheduling upon uniform multiprocessors is robust with respect to both job execution requirements and processor computing capacity. Index Terms—Uniform multiprocessors, hard-real-time systems, resource augmentation, earliest deadline first, robustness. 1

Emerging heterogeneous, multimedia network applications, such as video teleconferencing, require guarantees of timely and effective information delivery. In particular, it is important that the network does not commit to an additional service unless throughput and timeliness guarantees can be provided for both the new and existing connections. However, as both hardware changes and applications diversify, the forms of traffic patterns and of guarantees are impossible to predict a priori. In this paper, we provide a general model of traffic flows, and a schedulability test for a set of such flows on a packet-switched virtualcircuit fixed-packet network, together with a set of rules designed to make evaluation of the schedulability conditions feasible and efficient. An experimental study is presented which demonstrates the usefulness of the proposed rules and offers recommendations for building actual networks. We also derive testing sets for some standard traffic models and discuss issue...

There is an increased interest in using broadcast disks to support mobile access to realtime databases. However, previous work has only considered the design of real-time immutable broadcast disks, the contents of which do not change over time. This paper considers the design of programs for real-time mutable broadcast disks --- broadcast disks whose contents are occasionally updated. Recent scheduling-theoretic results relating to pinwheel scheduling and pfair scheduling are used to design algorithms for the efficient generation of real-time mutable broadcast disk programs.

Sporadic tasks in hard-real-time systems, as definedby Mok [14], are characterized by triples (e# d# p)# 1 e d# e p. Two successive requests by the same task will beseparatedbyat least p time units, and the task must be scheduled e time units within d time units of a request. A scheduling algorithm is said to be static if it does not depend on the sequenceofrequests# otherwise it is dynamic. We present here three major results. The first is that no static algorithm can be optimal. The second is that, modulo certain assumptions that imply scalability, no dynamic algorithm can take less than O(n) online time per slot scheduled. The third result is a fast scheduling algorithm based on pinwheel scheduling. 1 Introduction Hard-real-time systems are systems of tasks where missing a single deadline may have disastrous consequences. Such systems have been the focus of much study over the past twentyyears [13,11,9,12,14,1, 10, 7, 17, 8]. Sporadic tasks in a hard-real-time environment wer...

We have recently been studying the scheduling of real-time systems upon uniform and identical multiprocessor platforms. In particular, we have been exploring the use of EDF-scheduling in such systems; this paper summarizes some of our recent ndings in this eld.