Abstract

Workflow has become an important paradigm for distributed data and computing systems in a wide range of application areas. In a workflow, tasks executing on autonomous, heterogeneous systems are coordinated through data and control flow constraints. An important challenge in workflow management is the scheduling of actions and operations performed by the concurrently executing tasks. The legal interleavings among the tasks must be specified, and scheduling control mechanisms must be generated so that the executions progress correctly and efficiently. The scheduling issue is particularly difficult because the execution constraints may be application-specific, and certain autonomous task executions may be outside the scope of workflow control. We use techniques adapted from supervisory control theory of discrete event systems for specifying and generating scheduling controllers in workflow environments. We use our modeling approach to specify the tasks and the dependencies as finite stat...

Citations