Models for optimization of network topology and performance in wireless mesh networks.

摘要

This dissertation presents models for the optimization of network topology and performance in Wireless Mesh Networks (WMNs). We examined the problems of network connectivity and network planning for optimizing the topology. Efficient routing of Internet traffic was considered for optimizing the performance.;We presented a methodology for accurately calculating the network connectivity in WMNs. The methodology takes into account dependencies between links of geometrically co-located nodes, as well as the reduced connectivity of boundary nodes. Closed form expressions for the upper bound for the network connectivity in a chain topology, as well as a triangular lattice topology were derived. A comparison study between models that assume independent links' connectivities and our model shows that the commonly used assumption of link independence almost always underestimate the network connectivity. We also show that the effect of boundary nodes on the network connectivity can not be usually eliminated by assuming an infinitely large network.;The optimal network plan in terms of node class selection and number of nodes required to achieve the given service requirements in a WMN was investigated. The network is said to be optimal if it minimizes the network implementation cost while achieving the applications' requirements. To tackle this problem, we developed an analytical model and methodology for WMN planning optimization. The model takes into account dependencies between links of co-located nodes, the effect of boundary nodes' reduced connectivity, in addition to fading and shadowing effects. The proposed methodology resulted in a qualitative and quantitative description of the optimal WMN planning policy.;Furthermore, we proposed a dynamic routing policy for routing Internet traffic from mesh nodes to multiple Internet gateways simultaneously. The system model for the WMN takes into account the effects of the wireless channel. Stochastic programming was used to find the optimal policy for the route controller. Our performance evaluation shows that utilizing multiple Internet gateways simultaneously results in a much more efficient use of resources. The performance of the proposed optimal routing policy was found superior compared to the performance of other policies found in the literature.