- Name: Student A
- Abstract Due Date: 9/18/96
- Date: 9/18/96
- Article Title: Implications of Classical Scheduling Results for Real-Time Systems
- Author: Stankovic, et. al.
- Journal: Computer, June 1995, pp. 16-25.

The authors of this study in scheduling theory focus on the tools and theorems that are available for designers of real-time systems to create optimal scheduling algorithms. They investigate algorithms that are both preemptive and nonpreemptive, on-line and off-line, and static and dynamic. They also focus on the differences between the scheduling algorithms of single processor machines and dual or multi-processor computers. In examining these algorithms, they intended to determine the effects of overloading on optimal algorithms, and discuss the different methods of determining a n algorithm, including minimizing the maximum lateness of jobs, total tardiness, minimizing the weighted sum of the completion times, minimizing schedule length, minimizing the number of processors required, or minimizing the sum of the completion times.

The authors take a very traditional approach in their study of scheduling problems in real-time systems. Instead of creating new algorithms, they rely heavily o n theorems used in developing scheduling algorithms and describe the differences between real-time systems and non-real-time systems and how these differences affect the output of a process scheduler. They then modify or adapt the existing algorithms to suit the real-time systems and then evaluate the effectiveness of each different model. In some ways, this method of analysis of the algorithms is helpful, since they already know the faults of each algorithm in a non-real-time system and can apply what they know to real-time systems, however they implement very little originality or creativity in implementing new scheduling algorithms. They conclude that the rate-monotonic scheduling algorithm, which has been used by NASA, the FAA, IEEE, and the Department of Defense, is currently the most optimal for most real-time situations.

In writing the paper, the authors are very clear and explain each theorem they use in their analysis. This is useful since they use more than 20 different theorems. They also break the paper into several logical sections - one for each of the theorems for single processor systems, and one for each of the theorems used in multi-processor systems. They clearly describe in the introduction which questions they expect readers to be able to answer by the end of the paper and are thorough about explaining each of them. Their topic is summed clearly in the introduction when they write , Their explanations of scheduling algorithms are interesting and thorough, and provide a logical extension to the material we studied in class on process scheduling.