【作者】 江自兵；

【导师】 梁祥君；

【作者基本信息】 安徽工程大学 ， 计算机应用技术， 2011， 硕士

【摘要】 无线传感器网络(Wireless Sensor Network, WSN)是由部署在监控区域大量微型、低成本的智能传感器节点,利用无线通信信道,通过多跳自组织方式构成的网络系统。传感器节点硬件结构简单且能量有限,通常部署环境恶劣且无人值守,加上无线信道本身的不可靠,导致网络经常出现通信链路断开、节点损坏或者能量耗尽等故障,造成传感器采集的数据不能正确反映事件发生区域真实的情况。同时,网络可能会受到恶意节点的攻击,或者传感器节点软件问题,也会造成采集的数据错误。对于以数据为中心的无线传感器网络,网络的容错性和可靠性的研究显得尤为重要。本文针对无线传感器网络缺乏信任管理机制,提出一种基于节点信誉度的信任管理机制,该机制的核心思想是：考虑到传感器节点采集的数据不仅与节点本身的历史数据有关,还与领域内其他传感器节点采集的数据有关,根据相似度矩阵得出数据信任度,计算可用的通信带宽和剩余能量分别得出带宽和能量的信誉度。综合无线传感器网络传感器节点的数据、带宽和能量三方面的信誉度建立信任模型,并将该信任模型引入到无线传感器网络管理体系中,此时网络组建和节点各种行为都以可信度为依据。接着将该信任管理机制引入到传感器节点的事件检测中,并提出一种节点信誉度的容错事件检测算法NCFTED(Nodes Credit Fault-Tolerant Event Detection),该算法的主要思想是：当传感器检测到异常数据时,根据故障节点排除算法FNE(Fault Nodes Exclude)对故障节点进行排除；利用事件区域节点的信任等级进行簇头的选择,分析得出多跳的方式比单跳的方式更节省能量,特别对于簇头节点和离簇头较远的节点,簇内和簇间的通信尽可采用多跳的方式,最终将数据发送给基站。文章首先叙述了本课题研究的前景和意义,简要的描述无线传感器网络应用前景,并对目前无线传感器网络存在的问题进行了分析。详细地介绍了当前无线传感器网络的信任管理机制和容错技术的国内外研究现状以及无线传感器网络的体系结构和基本特征。通过对节点信誉度的计算得出节点信任度评估模型,并将此模型引入到事件检测的容错技术中；其次,文章利用NS2网络仿真平台,对NCFTED算法进行仿真,该算法只需收集一次邻居节点的传感的数据,就可以检测到非正常的数据,通过简单的计算便可将故障节点进行排除。仿真分析出该算法识别率高,误检率和误报率低,计算复杂度低,拟合度高,由于簇的建立只在检测到有异常数据时进行,所以网络的通信量小,对大规模网络具有很好的可扩展性。更多还原

【Abstract】 Wireless Sensor Network (WSN) is a network system which is composed by a large number of miniature and low-cost intelligent sensor nodes deployed in monitoring area by using wireless communication channel through multi-hop self-organizing mode.The hardware structure of the sensor nodes is simple and has finite energy. Because the sensor nodes are usually deployed in rugged and unattended environments and the wireless channel itself is not reliable, all these lead to the frequent breakdowns of the network such as communication link disconnection, nodes damage or energy exhaustion, which brings that the data collected by the sensor network can not reflect the real situation of the area in which the events have occurred. Meanwhile, the network may be attacked by malicious nodes or the problems of the software of the sensor nodes may lead to collected data errors. To the data-centric wireless sensor networks, it is very necessary to carry on research on fault tolerance and reliability of the network.Due to the lack of trust management mechanism of WSN, this thesis proposes a trust management mechanism based on a kind of the credibility of the nodes. The core idea of this mechanism is to take into account the data collected by sensor nodes not only correlated on the node itself but also with other sensor nodes data within the field, by obtaining the data confidence level according to the similarity matrix, to calculate the available correspondence bandwidth and the excess energy and obtain the bandwidth and energy credit. Then the trust model is built by the credibility of the data, bandwidth and energy of the sensor nodes in wireless sensor network and the model is introduced to the management system, all the behaviors of the network building and the nodes are based on the credibility.Subsequently, the trust management system is introduced to the event detection of the sensor nodes and an algorithm of NCFTED (Nodes Credit Fault-Tolerant Event Detection) is proposed. The main idea of this algorithm is that when the sensors detect abnormal data, the failed nodes are excluded according to FNE (Fault Nodes Exclude) and the cluster heads established under a trust environment, are selected by taking advantages of the trust rank of the nodes in this field. By analyzing, A multi-hop mode can save more energy than that of single-hop. Specially for the nodes of the cluster head or the ones far from them, The multi-hop mode should be adopted as far as possible to communicate between inside and outside of the clusters and transmit the data to the base stations.In this thesis, the prospect and significance of this research was firstly narrated and the promising applications together with the problems of current wireless sensor network were analyzed. It also simply described the application prospect of wireless sensor network in detail and the current research status of trust management mechanism with fault-tolerant technology both at home and abroad, and the structure and basic features of current wireless sensor networks. The trust evaluation model of the nodes was obtained by calculating the credibility of them and it was introduced into the fault-tolerant detection technology.Secondly, this thesis emulated the algorithm NCFTED, which can detect abnormal data by collecting the sensor data of the neighboring nodes only for one time and exclude the failed nodes by simple calculation though NS2 network emulation platform. The emulation had figured out that the algorithm had a high rate of identification, low rate of false detection and alarm, low computational complexity and high fitting degree. Because the establishment of the clusters can only be conducted when the abnormal data is detected, the communication traffic of the network is very little and the clusters have the best expansibility in the large-scale networks.更多还原