Date of Graduation
Statler College of Engineering and Mineral Resources
Lane Department of Computer Science and Electrical Engineering
This work focuses on the empirical evaluation of distributed dispatching strategies on shared and distributed memory architectures for hard real-time systems. The dispatching model accommodates process parameter variability and analyzes the effect of variable execution times.;Hard real-time systems are modeled in the E-T-C scheduling framework and dispatched if a valid schedule exists. We examine the dispatchability of Partially Clairvoyant schedules of different sizes and varying deadlines under reasonable assumptions. The effect of scaling up the number of processors used by the dispatcher is also studied. The results validate the superiority of the distributed strategies over sequential dispatching and scalability of the distributed strategies. Certain system limitations which lead to Loss of Dispatchability in the experiments were pointed out.;The model finds applications in diverse areas like safety critical systems, robotics and machine control, real-time data management, and this approach is targeted at powering up the controllers.
Yellajyosula, Kiran S., "Distributed dispatchers for partially clairvoyant schedulers" (2003). Graduate Theses, Dissertations, and Problem Reports. 1408.