Date of Award

Summer 1998

Document Type


Degree Name

Doctor of Philosophy in Computing Sciences - (Ph.D.)


Computer and Information Science

First Advisor

Murat Tanik

Second Advisor

Peter A. Ng

Third Advisor

Daochuan Hung

Fourth Advisor

Franz J. Kurfess

Fifth Advisor

Ajaz A. Rana

Sixth Advisor

Ali Hikmet Dogru


Software development processes, as a means of ensuring software quality and productivity, have been widely accepted within the software development community; software process modeling, on the other hand, continues to be a subject of interest in the research community. Even with organizations that have achieved higher SEI maturity levels, processes are by and large described in documents and reinforced as guidelines or laws governing software development activities. The lack of industry-wide adaptation of software process modeling as part of development activities can be attributed to two major reasons: lack of forecast power in the (software) process modeling and lack of integration mechanism for the described process to seamlessly interact with daily development activities.

This dissertation describes a research through which a framework has been established where processes can be manipulated, measured, and dynamically modified by interacting with project management techniques and activities in an integrated process modeling environment, thus closing the gap between process modeling and software development.

In this research, processes are described using directed graphs, similar to the techniques with CPM. This way, the graphs can be manipulated visually while the properties of the graphs-can be used to check their validity. The partial ordering and the precedence relationship of the tasks in the graphs are similar to the one studied in other researches [Delcambre94] [Mills96]. Measurements of the effectiveness of the processes are added in this research. These measurements provide bases for the judgment when manipulating the graphs to produce or modify a process.

Software development can be considered as activities related to three sets: a set of tasks (τ), a set of resources (ρ), and a set of constraints (y). The process, P, is then a function of all the sets interacting with each other: P = {τ, ρ, y). The interactions of these sets can be described in terms of different machine models using scheduling theory. While trying to produce an optimal solution satisfying a set of prescribed conditions using the analytical method would lead to a practically non-feasible formulation, many heuristic algorithms in scheduling theory combined with manual manipulation of the tasks can help to produce a reasonable good process, the effectiveness of which is reflected through a set of measurement criteria, in particular, the make-span, the float, and the bottlenecks. Through an integrated process modeling environment, these measurements can be obtained in real time, thus providing a feedback loop during the process execution. This feedback loop is essential for risk management and control.