Scheduling in multiple processor real-time systems.

Abstract

The objective of this thesis is to investigate the effects of different dynamic scheduling heuristic rules on the design of Real-Time Systems (RTS) which allow for the execution of complex responses. Complex responses are characterized by the concurrent nature of activity execution. RTS design will be done through the use of a simulator which allows for the selection of the external event characteristics, definition of the RTS (processes, due dates, activity-to-resource allocations with activity execution times) and selection of different activity scheduling schemes (first-come first-served, earliest due date, shortest activity first, etc. ...). Statistics, such as response execution times, waiting times of ready-to-execute activities, percentage of responses that are tardy and resource utilization, are gathered and provided to the designer. These statistics allow the designer to refine the design, by changing system parameters, until satisfactory RTS operation is achieved. In this study, the RTS will be implemented on a centrally coordinated multiple processor system topology. (Abstract shortened by UMI.)