A Component-Based Multi-Agent System Modeling Methodology

This dissertation defines a new feature called a
potential arc that is used to extend traditional
Colored Petri Net(CPN) models to support agent
behavior modeling. Potential arcs serve to extend the
traditional CPN model with an explicit support for
distinguishing the representation of potential
conflicts and real conflicts, and thereby avoid the
need for eliminating all potential conflicts in the
design stage.

The research reported in this dissertation involves a
broad array of tasks, including model
generation algorithms, analysis/simulation
techniques, and experimentation. The main
contributions of the research are : 1) defines a new
feature called a potential arc that is used to
extend traditional Colored Petri Net (CPN) models to
support agent behavior modeling; 2)
presents a semi-automated MAS modeling methodology,
which allow software engineers to
create MAS systems based on the extended CPN model;
3) presents an intelligent coordinator
design methodology, resulting in a set of
coordinators at different levels of intelligence; and
4)evaluates the proposed modeling methodology by
developing models and performing simulationbased
analysis.

Autorentext

Jiexin Lian got his Computer Science Ph.D degree from University
of Illinois at Chicago at 2007. His research interest mainly
focuses on Petri net based system modeling and analysis. He also
has interest in other area of software engineering. Jiexin
currently work for Amazon.com in Seattle, WA, U.S

Klappentext

This dissertation defines a new feature called a
potential arc that is used to extend traditional
Colored Petri Net(CPN) models to support agent
behavior modeling. Potential arcs serve to extend the
traditional CPN model with an explicit support for
distinguishing the representation of potential
conflicts and real conflicts, and thereby avoid the
need for eliminating all potential conflicts in the
design stage.

The research reported in this dissertation involves a
broad array of tasks, including model
generation algorithms, analysis/simulation
techniques, and experimentation. The main
contributions of the research are : 1) defines a new
feature called a potential arc that is used to
extend traditional Colored Petri Net (CPN) models to
support agent behavior modeling; 2)
presents a semi-automated MAS modeling methodology,
which allow software engineers to
create MAS systems based on the extended CPN model;
3) presents an intelligent coordinator
design methodology, resulting in a set of
coordinators at different levels of intelligence; and
4)evaluates the proposed modeling methodology by
developing models and performing simulationbased
analysis.