| Wireless mesh network (WMN) is a hybrid wireless network that has layered struc-ture. As a key technology of the next generation, WMN has several advantages such as high?exibility, wide coverage, high reliability, broad bandwidth, high utilization, facility, andetc..Therefore, WMNs can fulfil the proliferative requirements of wireless users. However,traffic loads in WMNs are aggregated from multiple communication systems, such as cel-luar, WiFi, WiMAX, Ad Hoc, and etc. Then, high capacity is required by WMNs. However,aggregated loads aggravate the interference problem of WMNs. This is because availablewireless channels in wireless communication systems are very limited. Furthermore, traf-fic loads sent from source nodes are usually relayed by several nodes before they arrive atdestination nodes. Since adjacent wireless nodes are inclined to take the same channel, longrouting paths further aggravate the interference level of WMNs. High interference decreasesthe capacity and performance of WMNs. As a consequence, enhancing the network capac-ity is one critical issue of WMNs. The utilization rate of wireless channel can be greatlyimproved by proper channel assignment. Therefore, channel assignment is one of the mosteffective strategies to improve the network capacity of WMNs.WMNs adopt several new techniques and strategies to take full advantage of preciouswireless channels to improve the network capacity. These new techniques include equippingeach node with multiple wireless interfaces, adopting interface switching strategies, multi-channel MAC, directional/adaptive antenna, and so on. These strategies improve the perfor-mance of WMNs to some extent, whereas, they induce new problems and challenges. Forinstance, most protocols assume that every node is equipped with single interface and adoptsstatic channel assignment protocols. Thereby, these protocols cannot be directly applied tosuch environments that nodes adopt dynamic interface assignment strategies. Aiming at im-proving the performance of WMNs by interface swtiching, this dissertation mainly studiesthe issues of interference estimation strategies, broadcasting techniques, and channel assign-ment in multi-interface WMNs that adopt frequent interface switching strategies. It also researches traffic modeling and construction of simulation testbeds for WMNs. The researchresults of this thesis are as follows:(1) Putting forward a new strategy to estimate the interference level of multi-interfaceWMNs, where nodes adopt interface switching strategies. The interference problem in suchwireless networks is complicated due to frequent interface switching during communica-tions. However, few current interference estimation strategies consider the effect of frequentinterface switching on interference between adjacent links. This dissertation proposes a newmulti-interface interference estimation strategy based on interference graph. The presentedstrategy considers the impacts of frequent interface switching, the characteristics of trafficloads in WMNs, and propagation characteristics of wireless signals on interference. Simu-lations show that the proposed strategy not only is simple, but also adopts well to frequentinterface switching.(2) Presenting an efficient broadcasting scheme for multi-interface WMNs that adoptdynamic channel assignment. There is no permanent communication links in a WMN thatadopts dynamic channel assignment. Thereby, the broadcasting character of wireless chan-nels is unavailable. In order to send broadcast packets in such networks, current strategieseither need specific additional broadcasting interfaces or induce huge interface switchingoverheads. This dissertation brings forward a new broadcasting strategy that does not needspecific broadcasting interface. The proposed strategy provides efficient broadcasting sup-porting, whereas it keeps interface switching overheads under tolerable range. In addition,the proposed strategy can also decrease broadcast redundant in WMNs by preventing allnodes that have accepted broadcast packets on selecting relayed nodes.(3) Developing a queuing theory based traffic model for Infrastructure WMNs (I-WMN). In this model, every gateway node and the outermost mesh node is modeled asan infinite queuing system, whereas all nodes in each hop are modeled as a finite capacityqueuing system. Based on this model, the thesis analyzes the network throughput, packetsloss rate and packets delay on each hop nodes in I-WMN. The thesis also amends the originalmodel on account of the impacts of wireless interference, bandwidth, and etc.. Simulationresults show that the modified model is more accurate than the original one to model thetraffic character of I-WMNs. Based on result achieved by this traffic model, we proposedand implemented a hybrid channel assignment protocol for I-WMNs. According to analysis,loads on gateway nodes are the heaviest. Starting from gateways, traffic loads on a nodedecrease along with the hop count increasing. The proposed protocol then adopts differ-ent interface assignment strategies for different nodes: gateway nodes adopt static interface assignment strategy, whereas all the other nodes adopt hybrid strategy. The presented pro-tocol considers the characteristics of traffic loads in I-WMN, whereas it does not depend onprior information of traffic loads on nodes/links. The protocol can be implemented easilywithout changing current hardware. Extensive simulations show that the proposed protocolimproves the network capacity and ?exibility. It also improves per-?ow fairness of the net-work, whereas it keeps the interface switching overheads and coordination complexity undertolerable range.Besides, we extended the simulation tool NS2 (Network Simulator) to support multi-interface and frequent interface switching. The extended platform does not bind to any spe-cific protocols. Consequently, it can be easily applied to every dynamic channel assignmentstrategy/protocol. All protocols proposed in this dissertation are implemented on the ex-tended NS2 platform. Results show that the extended platform is efficient to support multi-interface and dynamic channel switching in wireless networks.KEY WORDS: Wireless Mesh Network, Multiple Interfaces, Dynamic Channel Assign-ment, Interference Estimation, Broadcast Strategy, Traffic Model, NS2 Simulation...
|
PDF Full Text Request