Fault-tolerant distributed real-time scheduling
Date
1991
Authors
Srinivasan, Anand
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Hard Real-Time Systems are employed in highly complex and time-critical applications where a high degree of fault-tolerance is a mandatory feature. A distributed algorithm for providing fault-tolerant optimal schedule in a simply periodic distributed real-time system is proposed. Each job is serviced by a primary or an alternate algorithm. The primary algorithm provides a desirable service that may not satisfy the timing constraints, whereas the alternate algorithm provides an acceptable service that always satisfies the timing constraints. After invoking an optimal scheduler in each individual node, the algorithm tries to schedule additional primaries on other nodes. Since primaries provide more accurate results, the distributed scheduling algorithm achieves better result accuracy without sacrificing the timing accuracy. The algorithm is first introduced and then applied to virtual ring network and binary n-cube interconnection network. A dynamic scheduler which enhances the run-time performance of the previously scheduled jobs, is also described. The results of performance tests for various randomly generated data are also given.