Date of Award


Document Type


Degree Name

Master of Science (MS)


Computer Science

First Advisor

Dr. Hesham H. Ali


Recently, the number of applications demanding real-time performance from their multiprocessor systems has been significantly increasing. At the same time, due to the possible catastrophic consequences from missing deadlines, fault tolerance has become a critical issue in real-time systems. A scheduling problem is a classical problem that involves finding a feasible allocation and order of tasks into processors while optimizing a given objective function. In addition, in a fault tolerant multiprocessor scheduling system, all performance requirements must be satisfied even if a processor failure occurs. If a given problem requires time that is exponential in the problem size, the problem is called NP-complete. In this sense, a scheduling problem is known to be NP-complete in its general form as well as in many restricted cases. Therefore, many researchers have proposed heuristic algorithms to solve various scheduling problems. As an alternative to traditional heuristic algorithms, several optimization methods such as simulated annealing, tabu search, and genetic algorithms have been adapted to solve various NP-complete problems and proven their effectiveness. Nonetheless, almost none of these methods have been used for fault-tolerant scheduling problems. In this thesis, we first introduce a variety set of existing heuristic algorithms and optimization methods specifically designed to solve scheduling problems. Then our focus moves toward genetic algorithms. We present a generic algorithm and take a new approach to address real0tie fault tolerant scheduling. We also modify the existing heuristic algorithm to fit into our problem and compare these two algorithms to analyze the effectiveness of the generic algorithm.


A Thesis Presented to the Department of Computer Science and the Faculty of the Graduate College University of Nebraska In Partial Fulfillment of the Requirements for the Degree Masters of Science University of Nebraska at Omaha. Copyright 2004 Yoshitsugu Hashimoto

Files over 3MB may be slow to open. For best results, right-click and select "save as..."