| Wireless mesh network (WMN) has the advantages of low-cost, easy-maintained, good-compatibility of the underlying technology and network coverage, which makes it easy to achieve the large-scale deployment. However, due to the characteristic of the multiple-hop wireless network, WMN also has some critical problems such as end-to-end QoS guarantee, effective assignment of the wireless resources and so on. In this paper, some inherent problems about the network architecture, path selection protocol and channel assignment in WMN are analyzed and the solutions of algorithm or mechanism on these problems are given.The key technology in a WMN includes many aspects of network. The paper focuses on the performance optimization in transport layer and protocols and algorithm of path selection and channel assignment in MAC layer. Based on the in-depth research on the existing technology, it has done the following innovative works:1) The deployment problem of Ring-based and Cluster-based WMN is abstracted as the MINLP (Mixed Integer Non-Linear Problem). For Ring-based WMN, the boundary conditions are ring width and hop-counts and the objective function is the optimal coverage. The optimal deployment schemes are given under different traffic status. For Cluster-based WMN, the boundary conditions are the distances between APs and the saturation throughput and the objective function is the maximal of the average AP throughput. The solution is the optimal deployment scheme for different strategies. Through the analysis on the existing transport layer technology, the reliable LRTP and unreliable UDP are chosen as the transport technology in the paper. The optimal number of rings and ring width for different user’s speed in Ring-based WMN and the optimal distances for increment and uniform strategies in Cluster-based WMN are all tested.2) With distinguishing the target address and improving the maintenance of the routing, the C-HWMP (Coordinated Hybrid Wireless Mesh Protocol) is given which adopts proactive treeing for inter-mesh nodes communication and on-demand mode for intra-mesh nodes communication. Via the OPNET simulation, C-HWMP can reach the same routing efficiency both as the RM-AODV (Radio-Metric Ad hoc On-Demand Distance Vector) for intra-mesh communicating and the OTR (Optimized Tree-based Routing) for inter-mesh communicating.3) Starting from the melioration of the air-time metric in RM-AODV which is the on-demand mode of HWMP, it researches the energy model of the active and sleeping states for wireless mesh STA, calculates the optimal STA transmitting power and converts it into the routing radio metric. In addition, the normalized relationship between time-delay and throughput are found out. The QoS requirements are also taken as the routing weight factors. Hence the enhanced RM-AODV with the energy saving and QoS guarantee are proposed. The simulation results show this routing pattern can lower the intermediate STA’s energy consumption which prevents some mesh STA from rapid invalidation.4) The channel assignment process is divided into two parts:channel assessment and channel assignment. In the phase of channel assessment, the traffic-aware Mesh node traffic model and interference mode are introduced which enhance the accuracy of the channel assessment. STA is in a queue for waiting and every STA is forced to be visited only once. This avoids the updated of the channel result caused by the iteration operation in the general method. The treatment on the assigned queue also uses the greedy way. With the principle of channel multiplexing, the channel outside of the interference range is preferred. Then the minimal traffic channel inside the interference range is chosen. Hence, the double greedy mechanism of the channel assignment is shown. The simulation work shows that the optimized method can make the average link interference is reduced by about17%and the total system interference is reduced about14%.5) Based on the interference aware, the PTI algorithm is proposed, which aims at the hot spots problem, traffic load unbalanced problem and link interference problem simultaneously. PTI is a multiple channel assignment algorithm that merges link Priority-aware, Traffic loads-aware and Interference-aware together. The OPNET simulating results show that the PTI algorithm effectively improves network throughput in the cases of different numbers of interfaces and channels.The paper verifies the theoretical results via various simulation platform testing and solves some critical problems in the current research on the WMN in the aspects of network architecture, path selection and channel assignment. In the meanwhile, the future works are also proposed based on these researches. |
