Use this URL to cite or link to this record in EThOS:
Title: Coordination of FIPA compliant sotfware agents using utility function assignment
Author: Lynden, Steven James
ISNI:       0000 0004 2746 2459
Awarding Body: Cardiff University
Current Institution: Cardiff University
Date of Award: 2004
Availability of Full Text:
Access from EThOS:
Access from Institution:
A Multiagent System (MAS) consisting of interacting autonomous agents motivated by individual objectives may be utilised to achieve global objectives in scenarios where centralised control is very difficult because of the distributed nature, complexity, or other restrictions present in a problem domain. When deploying MAS to achieve global objectives, a degree of coordination is usually required in order to ensure that the behaviour of the system is desirable with respect to these objectives. A sub-optimisation problem occurs when individual agent objectives are inconsis tent with global objectives. Current approaches towards this problem within the context of learning based agents include configuring the utility functions possessed by individual agents so that the maximisation of individual utility functions results in a desired global behaviour. Interoperability is of critical importance in Internet-scale MAS, and a promising standard for this is currently evolving in the form of the "Foundation for Intelligent Physical Agents" (FIPA) specifications. This thesis focuses on the integration of techniques based on the assignment of utility functions in order to coordinate MAS within the domain of FIPA compliant agent systems. The notion of utility is extended to form two separate types: performance and functional utilities. Whereas functional utility is based on abstract application specific objectives, performance utility concentrates on performance engineering related issues. The benefit of this approach is demonstrated by a software toolkit supporting the development of learning based FIPA compliant MAS, which is applied within two domains, an application based on market based buyer agents in artificial markets, and a computational Grid based application.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available