Internet-Access-Oriented Wireless Mesh Network Performance Analysis And Protocol Optimization

By the developing of wireless communication technologies, we can enjoy the convenient wireless network services for Internet accessing. Cellular Network has a high cost in construction and maintenance. Wireless Local Area Network (WLAN) based on IEEE 802.11 is widely used for Internet accessing. However, most WLANs used today are designed in Infrastructure mode. All the clients connect to the wired network through an access point within one hop. Limited by the wireless transmission range, the wireless coverage of WLAN is small. In addition, the coverage of wireless network depends on the availability of cable lines. It increases the cost to expand the coverage by building multiple WLANs. A new concept"Wireless Mesh Network"is provided to reduce the cost. Wireless Mesh Network uses wireless multi-hop technologies to expand network coverage. And it can provide more convenient and more affordable wireless broadband access.Wireless Mesh Network uses wireless technology as the communication medium. Wireless transmission is interference because of the shared spectrum. Impacted by the environments, data transmission is unstable on wireless. In wireless mesh network, data packet is forwarded to destination by multiple hops. Conventional communication protocols cannot fully adapt to this new feature, which makes the end-to-end transmission inefficient. This dissertation aims to improve network capacity and communication efficiency of Internet-access-oriented wireless mesh network. It provides several optimizations of communication and network protocols based on the performance analysis.First of all, in order to find the factors which affect the communication efficiency, we present a performance prediction model for 802.11-based wireless mesh networks. Based on the analysis, several strategies are provided to optimize the network for throughput, such as"load-balancing","routing design","rate adaption", and so on.Since wireless mesh network backbone is a multi-hop based wireless network, load-balanced routing protocol is investigated here for data forwarding. Based on the usage model of Internet-access-oriented wireless mesh network, a layered transmission model is suggested to control the network load. The entire network load is balanced through the balance of inner-layer. Then, we design a new routing metric for multi-radio and multi-channel wireless mesh network, which considers the transmission interference and takes the load balance between relevant links into account. Subsequently, a multi-gateway multi-path routing protocol is proposed to explore path diversity, cope with link failure and increase network redundancy. It constructs multiple paths between source and destination nodes using proactive tree architecture.Rate adaption (RA) and transmission scheduling are studied to improve the single hop performance of wireless mesh network. The common rate adaption algorithms do not consider the short-term past performance. This"blind probing"can lead to significant loss and throughput degradation. We propose a history-aware rate adaption algorithm for legacy 802.11 wireless networks. It uses a simple generic adaptive probe time window mechanism to capture short-term history. It takes the tradeoff between history information and short-term probing. New challenges of rate adaption are brought in MIMO scenarios. A novel MIMO RA scheme is proposed in 802.11n wireless networks. Considering the characteristics of MIMO technologies, it zigzags between intra- and inter-mode rate options. Furthermore, in order to improve the radio efficiency, we design and implement a packet-level scheduling mechanism to explore parallel transmissions between wireless links. Radio diversity is explored here to increase network capacity.In order to deploy a wireless mesh network in campus scenario, a cost-minimization deployment mechanism is provided finally. This mechanism takes both the different user demands and heterogeneous wireless devices into consideration.On the whole, a performance analysis model is proposed in our dissertation. Based on the analysis result, four protocol optimizations are designed to improve network capacity, end-to-end performance and user experience, which include load-balancing, routing design, rate adaption and transmission scheduling. All these work can provide a strong foundation for other related topics in WMNs.