| Wireless mesh networking is one of the most promising next generation network technologies. A wireless mesh network is a decentralized, self-organizing, self-configuring and self-healing multi-hop wireless network. In this thesis, we introduce the development, architectures, characteristics and applications of wireless mesh networks and present the existing channel assignments and routing protocols for wireless mesh networks.;To further improve the aggregated network performance, we propose Load-Aware CAEPO scheme based on the original CAEPO. In Load-Aware CAEPO, instead of using the busy time proportion of interfering nodes, we employ the traffic load as another main factor of the interference metric besides the channel overlapping degree. In addition, the concept of self-interference is introduced to estimate the interference metric. To facilitate the implementation of our channel assignment scheme, we modify the original AODV to be bandwidth-aware, where end-to-end delay and available bandwidth are both used as the routing constraints. Simulation results demonstrate that the proposed scheme can significantly improve the aggregated network performance.;For large networks, we introduce a node grouping algorithm in Load-Aware CAEPO and name the new channel assignment scheme Load-Aware CAEPO-G. Compared to Load-Aware CAEPO, Load-Aware CAEPO-G leads to a fairer channel assignment and achieves a minor improvement of the aggregated network performance.;Finally, performance of Load-aware CAEPO scheme is studied under voice applications over wireless mesh networks. To address the two challenges in voice over packet (VOP) applications, end-to-end delay and delay jitter, we propose VOP-AODV routing protocol. Along with VOP-AODV routing protocol, Load-aware CAEPO scheme can effectively decrease end-to-end delay and delay jitter.;In recent years, many efforts have been taken to better exploit multiple non-overlapping channels for wireless mesh networks, e.g. IEEE 802.11 a based wireless mesh networks, in which 12 or 24 non-overlapping channels are available. Although the IEEE 802.11 big standards, which govern the unlicensed 2.4 GHz industrial, scientific and medical (ISM) band, provide 11 channels, only three of them, namely 1, 6 and 11 are non-overlapping. In order to better utilize communication bandwidth and improve quality of service, in this thesis, we propose a channel assignment exploiting partially overlapping channels (CAEPO). In CAEPO, the interference a node suffers within its interference range is the main metric for channel assignment. It is defined to be a combination of the overlapping degree between channels and busy time proportion, i.e. channel utilization ratio of interfering nodes. In addition to that, packet loss ratio is another major consideration in the implementation of channel assignment. |
