分簇无线传感器网络的节点诊断方法
发布时间：2019-03-12 19:17
【摘要】：无线传感器网络的节点在硬件及软件的功能、内存、能源供应等方面的设计局限,容易造成节点故障、降低或失去监控能力。因此,对节点进行故障检测,故障隔离或数据恢复,对提高网络运行可靠性和安全性有重要意义。本文提出有效的无线传感器网络节点故障诊断方法,具体内容如下:首先,提出一种传感器节点故障诊断模型—LSI-BM。该模型依据节点数据的时间关联性,建立基于向量混合积的异常数据诊断准则和贝叶斯原理的节点信任值评估方法。此外,对于故障节点缺失的数据,利用Hausdorff距离度量变量的相似度,为避免故障节点的数据对最终结果的影响,提出一种基于节点信任值加权的数据恢复方法。仿真结果表明LSI-BM模型能有效诊断出异常数据和故障节点,并且数据恢复方法实现有效的数据填补。其次,为提高节点在大规模网络中能量使用效率,通过改进不均等分簇(EEUC)协议中节点竞选半径和簇头的选择机制,提出基于不均等分层分簇的网络优化协议—UHC。此外,针对随机数机制可能造成竞选节点分布不确定问题,对节点随机数的阈值进行调整,不再是一个固定的数值。最后通过计算机仿真实验,比较了基于UHC协议和EEUC协议的网络分簇效果,仿真结果表明UHC协议有效降低簇内节点密度,延长网络使用寿命,并且簇头分布呈现一定规律性。
[Abstract]:Wireless sensor network nodes in hardware and software functions, memory, energy supply and other aspects of the design limitations, easy to cause node failures, reduce or lose the ability to monitor. Therefore, the node fault detection, fault isolation or data recovery is of great significance to improve the reliability and security of network operation. In this paper, an effective node fault diagnosis method for wireless sensor networks is proposed. The main contents are as follows: firstly, a sensor node fault diagnosis model, LSI-BM., is proposed. Based on the temporal correlation of node data, this model establishes an anomaly data diagnosis criterion based on vector mixed product and a node trust evaluation method based on Bayesian principle. In addition, using Hausdorff distance to measure the similarity of variables, in order to avoid the influence of the fault node data on the final result, a data recovery method based on the node trust value weighted is proposed for the missing data of the fault node. The simulation results show that the LSI-BM model can effectively diagnose the abnormal data and fault nodes, and the data recovery method can effectively fill the data. Secondly, in order to improve the energy efficiency of nodes in large-scale networks, by improving the selection mechanism of race radius and cluster head of nodes in unequal clustering (EEUC) protocol, a network optimization protocol-UHC. based on unequal hierarchical clustering is proposed. In addition, because the random number mechanism may cause the uncertain distribution of the election nodes, the threshold of the random number of the nodes is adjusted, which is no longer a fixed value. Finally, the network clustering effect based on UHC protocol and EEUC protocol is compared by computer simulation experiment. The simulation results show that the UHC protocol can effectively reduce the density of nodes in the cluster, prolong the service life of the network, and the cluster head distribution has a certain regularity.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP212.9;TN929.5
本文编号：2439075
[Abstract]:Wireless sensor network nodes in hardware and software functions, memory, energy supply and other aspects of the design limitations, easy to cause node failures, reduce or lose the ability to monitor. Therefore, the node fault detection, fault isolation or data recovery is of great significance to improve the reliability and security of network operation. In this paper, an effective node fault diagnosis method for wireless sensor networks is proposed. The main contents are as follows: firstly, a sensor node fault diagnosis model, LSI-BM., is proposed. Based on the temporal correlation of node data, this model establishes an anomaly data diagnosis criterion based on vector mixed product and a node trust evaluation method based on Bayesian principle. In addition, using Hausdorff distance to measure the similarity of variables, in order to avoid the influence of the fault node data on the final result, a data recovery method based on the node trust value weighted is proposed for the missing data of the fault node. The simulation results show that the LSI-BM model can effectively diagnose the abnormal data and fault nodes, and the data recovery method can effectively fill the data. Secondly, in order to improve the energy efficiency of nodes in large-scale networks, by improving the selection mechanism of race radius and cluster head of nodes in unequal clustering (EEUC) protocol, a network optimization protocol-UHC. based on unequal hierarchical clustering is proposed. In addition, because the random number mechanism may cause the uncertain distribution of the election nodes, the threshold of the random number of the nodes is adjusted, which is no longer a fixed value. Finally, the network clustering effect based on UHC protocol and EEUC protocol is compared by computer simulation experiment. The simulation results show that the UHC protocol can effectively reduce the density of nodes in the cluster, prolong the service life of the network, and the cluster head distribution has a certain regularity.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP212.9;TN929.5
【参考文献】
相关期刊论文 前1条
1 李平;李宏;吴敏;;WSNs分布式事件区域容错算法[J];计算机工程;2009年14期
，本文编号：2439075
本文链接：https://www.wllwen.com/kejilunwen/wltx/2439075.html
