| An ad hoc network is a multi-hop wireless network temporarily and dynamically formed by a collection of mobile nodes without the use of any preexisting network infrastructure or centralized administration. Applications such as disaster recovery, distributed collaborative computing and automated battlefields are typical examples of where ad hoc networks are deployed. Owing to its self-organization, rapid deployment and absence of any fixed infrastructure, ad hoc networks are gaining popularity as a significant and promising research domain. The effects of node movement, signal interference, and power outages, however, make the available link state and network topology information inherently imprecise. On the other hand, heavy traffic, frequent link failure and network partition will incur transmission disruptions, causing data packets to be delayed and dropped. Dynamically changing topology and lack of network resource make the design of an adaptively distributed routing protocol challenging. In the development of routing protocol, the selection of routing metrics and the adoption of routing strategies must adapt to the dynamic changes of network topology and network resource, which help to increase the packet delivery ratio, to decrease end-to-end delay and to minimize the control overhead. The inherently centreless, multi-hop, and distributed characteristics of ad hoc networks make the design of routig protocols mainly facing several difficulties: 1) How to adapt to the rapid, frequent and unpredictable network topology change; 2) How to efficiently deal with the link breakage and improve the packet delivery ratio; 3) How to limit the control overhead, reduce the end-to-end delay, and improve the throughput; 4) How to utilize the imprecise information to implement the global network resource optimization. To address these problems, the research of this dissertation includes four issues: 1) An overview of routing protocols for ad hoc networks. Based on the analysis of architecture of ad hoc networks, this dissertation reviews and summarizes the routing protocols for ad hoc networks on the different view, disscusses unicast routing, multicast routing, and QoS routing respectively, prevents the advantages and disadvantages of different routing strategies, and gives a solution in the aspects of selection of routing metric and adoption of routing strategies. 2) Performance analysis of dynamic topology in ad hoc networks. A snapshot-based statistical analysis method is proposed, which is used to study the rules of changes of network topology determined by the two different routing metrics, i.e. the shortest path and the most stable path based on the shortest-path. The average path stability time and the average path length are selected as performance metrics to explore the effects of node density, mobility rate, and pause time on the network topology performance. The average path stability time and the average number of new intermediate nodes of paths with different length was analysed statistically, the distribution rules of different path length was obtained and the distribution probability of the path stability time of the path length with the maximum probability was conducted. Results show that the selected path according to different routing metrics has different rules of changes, but the composition of intermediate nodes between the new optimal path and the previous one is very similar. The average path stability time decreases as the path length and the node velocity increases and the average path length becomes shorter with the increased node density. 3) A periodic route update-based active source routing protocol. An on-demand active source routing protocol (ASR) is proposed. It utilizes the periodic route update mechanism and the actively gratuitous route acknowledgment for obtaining the up-to-date network topology information. In the route discovery phase and in the route maintenance phase, every node adds or updates multiple paths to the designated destination via processing the route reply messages and the route acknowledge messages, respectively. Each path associated with an expiry time will be removed to avoid the occurrence of the stale route if it has not been used or updated during such a period of time. Under light load, simulation results of ASR show that the throughput is close to the ideal and the average latency is relatively stable. The overall performance of ASR almost not changes as the pause time changes, decreases slightly as the mobility rate increases, and increases slightly as the node density increases while degrades only under heavy load. ASR is the first to utilize periodic route update mechanism for reactive routing protocols and eliminate an inherent problem that the event-triggered route maintenance approach could not acquire the optimal path as long as the current route is available. 4) A soft bandwidth constrained QoS routing protocol. An active multipath QoS routing protocol (AMQR) is presented, which utilizes multiple disjoint paths to supportsoft bandwidth constrained QoS requirement. A distributed route discovery algorithm is proposed to find multiple disjoint paths with associated path stability and network resource information. In addition, an actively dynamic route maintenance algorithm based on periodic state update combined with gratuitous QoS acknowledments is introduced to refresh network topology and resource information, which helps the source dynamically adjust the traffic load on the desired route for data dispersion. At the same time, it takes advantage of the acquired path stability and allocated resource information to find the congested node and unstable link, and utilizes backup route or pre-initiates route discovery to avoid the occurrence of QoS disruptions. Simulation results compared with ASR show that AMQR provides excellent end-to-end QoS. In brief, this dissertation argues that the ideal routing metric for path selection should enhance the longevity of the selected route, and avoid bottleneck and congestion at intermediate nodes. Routing metric should be used not only in the route discovery phase but also in the route maintenance phase. On the other hand, routing protocols should actively adapt to the dynamic changes of network topology and network resource by making the best of the rules of network topology change, and dynamically discover and maintain the optimal path according to the routing metric in the entire route maintenance phase.
|
PDF Full Text Request