Abstract
dc:description.abstract<p>Agent-based systems have a key role to play in the effort to provide and support such applications, since agents embody several of the required characteristics for effective and robust operation in dynamic and heterogenous computing environments. However, there are a number of shortcomings relating to the use of the agent approach to application development. In particular, in this thesis we deal with the lack of clarity in existing agent models and address the need for models that can directly support practical application development. These are widely-accepted shortcomings that have been identified by a number of researchers in recent years [8, 32, 136, 189, 227, 231]. This thesis addresses these shortcomings with relation to the basic infrastructural concerns that are common to practically all significant agent-based applications in dynamic, heterogeneous environments. We develop <i>principled </i>and <i>reusable </i>models in support of agent-based systems construction, dealing both within individual agent construction and support for relationship identification and characterisation.</p> <p>In this thesis we make three main contributions. Firstly, through an abstract agent model we enable the characterisation of the wide range of agent types that can exist within a heterogenous environment. This facilitates development by ensuring that the underlying theory adequately models the actual application environment and provides indications as to where designers must focus their efforts. Secondly, we develop a model for agent construction which links the abstract agent model to practical application concerns and enables the specification of a range of agent architectures while also facilitating their run-time reconfiguration. This bridges the gap between abstract models and practical implementation, allows developers to choose the type of agent architecture that best suits the application at hand, and provide the flexibility for adapting architectures to changing application needs. Finally, we develop a model of agent interaction and use it to comprehensively identify all the possible relationships between two agents, as well as to relate agent goals to the abilities of agents to achieve those goals given their individual capabilities. This enables the effective identification and characterisation of agent relationships in dynamic environments, so as to guide the choice of appropriate relationship management mechanisms.</p>
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D.
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Southampton
- Year dc:date.issued
- 2004
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Ashri, Ronald
Rights
- Language dc:language
- English, eng