الفهرس Only 14 pages are availabe for public view
 |  | from | 118 |  |  |
| | | from | 118 | | |
Abstract
Processor task scheduling in real-time systems is a complicated problem that has been the subject of many research articles. In this thesis, we present a centralized algorithm for scheduling real-time tasks in a heterogeneous multiprocessor system. We first use an offline algorithm for scheduling periodic static hard real-time tasks where reliability and fault tolerance measures are considered. Then, in order to schedule dynamic aperiodic firm real-time tasks we suggest two approaches -other than the traditional linear search methods- to be used along with the offline algorithm. In the first approach, we use an online probabilistic algorithm, which starts the search for an adequate gap in a processor’s schedule, from the processors having the highest probability of success. And in the second approach, we propose a new online cut-off technique to be used along with the same offline algorithm. This is done by cutting off unsuitable processors, and then choosing from the remaining ones, using heuristics that we will call best-fit and worst-fit following the naming conventions of memory management.