Date of Award


Degree Name

Master of Science


Computer Science

First Advisor

Dr. Ajay Gupta

Second Advisor

Dr. Harrison Greenwood

Third Advisor

Dr. Robert Trenary

Access Setting

Masters Thesis-Open Access


The task scheduling problem is defined as a sequential algorithm, with individual tasks, t1, t2, t3… having associated execution times to be ported to a multiprocessor system. To determine the fastest parallel implementation is known to be an NP-complete problem, therefore heuristic techniques are employed to arrive at the best solution.

Some of the traditional methods used to solve this problem, are the list scheduling algorithm and the simulated annealing algorithm. However, evolutionary techniques are now also a consideration with the increased computation power available.

Evolutionary techniques, based on the popular theory of evolution, consist primarily of a solution represented in a chromosome data structure. Maintaining a population of chromosomes, generational transitions are accomplished through reproduction of the best solutions. The objective is to improve solutions by making small random changes and repeating the process.

The results found in this effort confirm the viability of the technique. Another observance was the evolutionary technique, as implemented here, had more overhead and therefore did not perform as exceptionally as the simulated annealing method.