A multiagent system for topology control on multi-radio wireless mesh networks

University of Technology, Sydney. Faculty of Engineering and Information Technology.NO FULL TEXT AVAILABLE. Access is restricted indefinitely. The hardcopy may be available for consultation at the UTS Library.NO FULL TEXT AVAILABLE. Access is restricted indefinitely. ----- The growth of wireless telecommunication networks has been accompanied by a demand for better network management techniques. As the centralised network management paradigm is no longer suitable given the diversity, complexity and the size of wireless networks, there is a need to provide a new level of flexibility and scalability of network management. The effectiveness of use of radio resources has a central role in ensuring that wireless telecommunication providers are able to successfully supply users with an appropriate level of services when and where required. This thesis investigates the applicability of lightweight, co-operative, multiagent systems to the distributed management of wireless multi-radio mesh networks. An overview of the background of the problems of topology control and channel assignment on such networks is presented, examining existing approaches, their theoretical underpinnings and deficiencies. The ramifications of identified shortcomings are discussed to highlight how well multiagent systems are placed to guarantee scalability and stability, and to provide flexibility and extensibility to wireless network management in a distributed and dynamic environment. This thesis introduces a novel topology control and channel assignment algorithms, based on interference cost and network traffic load metrics, and allowing for proactive radio resource management. Our overall goal is simply to reduce maintenance costs for such networks by removing the need for humans to tune the network settings. As the solution is distributed, it simply seeks to continually improve localised network performance as demands change. In any case, we argue that this form of optimisation of mesh network performance is the only feasible in real situations that are subjected to unanticipated perturbations and external interventions. The experiment results have been obtained for realistic scenarios of various network node densities and topologies. We have shown that the lightweight multiagent system proposed is robust, flexible, and extensible, and delivers good performance