| dc.description.abstract | User demand for seamless connectivity has encouraged the development of alternatives to traditional communications infrastructure
networks. Potential solutions have to be low-cost, easily deployable and adaptive to the environment. One approach that has gained
tremendous attention over the past few years is the deployment of a backbone of access points wirelessly interconnected, allowing users to access the wired infrastructure via wireless multi-hop communication. Wireless Mesh Networks (WMN) fall into this category
and constitute a technology that could revolutionize the way wireless network access is provided. However, limited transfer
capacity and interference resulting from the shared nature of the transmission medium will prevent widespread deployment if the
network performance does not meet users' expectations. It is therefore imperative to provide efficient mechanisms for such networks.
Resource management encompasses a number of different issues, including routing. Although a profusion of routing mechanisms have been proposed for other wireless technologies, the unique characteristics of WMNs (i.e. fixed wireless backbone, with the
possibility to embed multiple interfaces) prevent their straight forward adoption in WMNs. Moreover, the severe performance degradations that can result from the interference generated by concurrent data transmissions and environmental noise call for the development of interference-aware routing mechanisms.
In this thesis, we investigated the impact of interference on the
network performance of wireless mesh networks.
We designed algorithms to associate routers to gateways that minimize the interference level in single-channel and multi-channel
networks.
We then studied the performance of existing routing metrics and their suitability for mesh networks.
As a result of this analysis, we designed a novel routing metric and showed its benefits over existing ones.
Finally, we provided an analytical evaluation of the probability of finding two non interfering paths given a network topology. | en |
