Degree Name

Doctor of Philosophy


School of Computer Science and Software Engineering


With the increasing importance of the world-wide telecommunication and computer networks, agent technologies and multi-agent systems are attracting more and more attention. Agent technologies aim at the design of agents that perform well in environments, which are not necessarily stable and well-structured. In this context, specifying the organisation of a multi-agent system at designtime become3s a difficult task. If the structure of a multi-agent system needs to adapt to changes in the environment, the difficult task turns to be a virtually impractical task. Therefore, the self-organising behaviour of multi-agent systems is one of the most interesting phenomena. This thesis mainly aims at investigating the challenging issues of self-organisation in multi-agent systems. In addition, this thesis also uses the notion of self-organisation in other research fields in order to achieve better results compared to those studies, which did not take self-organisation into consideration. The contributions of this thesis consist of the following five aspects.

1. A composite relation adaptation mechanism is proposed, which is used to adapt relations between agents in a multi-agent system, to achieve efficient task allocation. This mechanism integrates a trust model, and takes both multiple and crisp relations into account.

2. An integrative self-organisation mechanism is developed. This mechanism combines the three principles of self-organisation, i.e., cloning/spawning, resource exchange and relation adaptation. It can outperform those mechanisms, which consider only one principle of self-organisation, in various aspects, e.g., efficiency of task allocation and load-balancing.

3. A self-adaption based dynamic coalition formation mechanism is presented. This mechanism enables agents to autonomously and dynamically adjust their degrees of involvement in di fferent coalitions to join new coalitions.

4. A self-organisation based strategy for the evolution of cooperation in networks is devised. This strategy embodies current developed strategies as agents' knowledge. Then, each agent can autonomously select a strategy from its knowledge base to adapt its behaviour.

5. A self-organisation mechanism for packeting routing in wireless sensor networks is introduced. This mechanism is not a concrete packeting routing mechanism, but is an assistant tool, which is independent of current routing approaches and can be employed to enhance many current routing approaches.

In summary, this thesis includes both theory and application of self-organsing multi-agent systems. Experimental results demonstrate the efficiency, effectiveness and potential applicability of the work done in this thesis.