【作者】 陈宇；

【导师】 林亚平；

【作者基本信息】 湖南大学 ， 计算机软件与理论， 2005， 硕士

【摘要】 传感器网络由于其独特的优点在军事和许多民用领域具有广泛的应用潜力,正得到越来越多的关注,对传感器网络各个方面的研究也成为目前学术界的研究热点。路由算法是传感器网络在网络层上的一个关键研究课题。由于节点的能量、存储和计算等方面的资源约束,传感器网络广泛应用数据聚合技术以节约能量,因此广播和组播通信也是传感器网络中的重要通信手段。本文着重探讨传感器网络中的组播路由算法。 本文首先简要介绍了组播路由算法在传统有线网络以及移动Ad-hoc网络中应用的情况,并讨论了相关的比较指标,对相关组播路由协议进行了定性的比较,为设计适合于传感器网络的组播路由算法提供依据。针对传感器网络节点稠密分布的特点,本文提出了一种基于最小连通支配集的组播路由算法,算法给出了一种最小连通支配集的分布式计算方法,利用该方法在网络中建立一个由支配集节点组成的连通子网,并设定支配集节点为簇头对网络进行分层,使得报文的转发限制在簇头节点之间,算法能够有效减少报文的产生,进而节约节点能量。针对传感器网络中每个节点都可以充当信源的特点,本文提出了一种基于梯度的路由算法,算法使用了最短路径树结构,具有较低的时延,可以较好地适应由于节点失效或节点移动而引起的拓扑变化。实验结果显示,该算法可以节省相当的能量开销,具有良好的性能。基于上述算法,本文还引入了螺旋树概念,提出了一种新的路由算法,算法采用螺旋树以保证树的健壮性,使得在链路/节点失效时数据报文仍然可以沿着“备用”路径转发而不致丢失,在节点具有一定移动性的情况下,算法仍然具有较好的性能。更多还原

【Abstract】 Sensor network has intensive potential on military and other fields for its unique advantages. More attention is paid to it and the research to each aspect of sensor network is also a active field in current academe. Routing algorithm is a key issue on network layer in sensor network. Because of the constrain of energy, memory and computation capacity, data-integration is intensively used to save energy in sensor network, so broadcast and multicast are also important methods to communicate. This thesis emphasis on multicast routing algorithm in sensor network.Firstly, this thesis introduces simply some multicast routing protocols in wire networks and ad hoc, defines some measures and evaluates the their performance. According to these results, we can design more suitable multicast routing algorithm. Considering the high density of nodes in sensor network, thesis proposes an algorithm based on minimal connected dominating set (MCDS). Algorithm gives a distributed method to calculate MCDS and constructs a connected subnet composed by cluster in CDS. It can deduce the number of packets and limit the forwarding of packets and save the energy of nodes. In sensor network, each node can do as a source, so this thesis proposes a new algorithm based on gradient. The algorithm use minimal-path-tree, has low delay and can adapt the change because of nodes failing or moving. The simulation shows the algorithm has good performance and save much energy. According to foregoing algorithms, we use the concept of spiral tree and propose a new algorithm. New algorithm use spiral-tree to assure the robustness and packets can be forwarded along with spare path when links or nodes fail. Even when nodes move, algorithm can still have good performance.更多还原