Time/ Location. MW 12:30 – 1:45pm. SN 325.

Instructor. Sanjoy Baruah [baruah@cs.unc.edu; (919)962-1803; Sitterson 319]

Goals and objectives. The goal of this course is to bring all course participants (students and the instructor) up to date regarding the current state of the art in real-time scheduling theory. To meet this goal, we will read, understand, and explore the possibilities of extending recently published papers from major conferences and journals in this research area. We will also be coverning home-grown results – some exciting research in real-time scheduling theory gets done here at UNC, and it behooves us to be aware of, and appreciate, the work being done by our colleagues.

Grading. Since this is a seminar course, there will be a minimal number of assignments, tests, and other stress-inducing activities. Each course participant is expected to lead the discussion on some of the papers to be covered in this class -- one half of the overall grade will be determined by the performance on these discussions. Another quarter will be determined by class attendence and participation. For the final one fourth of the grade, each student must present a written report on a topic to be discussed beforehand with the instructor.

• Scheduling during overload

• Inherent bounds
• Task models such as (m,k)-firm tasks; skip-over tasks; imprecise computation models; etc.
• Some papers:

• Kalyanasundaram and Pruhs (2000). Speed is as powerful as clairvoyance. JACM
• Phillips, Stein, Torng, and Wein (1997). Optimal time-critical scheduling via resource augmentation. STOC
• Edmonds (1999). Scheduling in the dark. STOC.

• Multiprocessor scheduling

• Anderson & Srinivasan’s work on pfair scheduling
• Ramamurthy & Moir’s papers
• The heuristic algorithms introduced as part of the Spring project
• Some papers:

• Leung (1989). A new algorithm for scheduling periodic real-time tasks. Algorithmica. (erratum)
• Kalyanasundaram and Pruhs (1999). Eliminating migrations in multi-processor scheduling . SODA.

• Power-aware scheduling

• Theoretical results
• Heuristic approaches to power-efficient scheduling
• Some papers:

• Yao, Demers, and Shenker (1995). A scheduling model for reduced CPU energy. FOCS.

• Recent results on feasibility-analysis

• General task models
• Improvements to traditional feasibility tests (improving the constants, parallelization, etc.)

• Server models

• The constant-bandwidth server (and its variants) out of Pisa
• The processor capacity reserves model from CMU
• Proportional share scheduling algorithms