面向节点能量消耗的无线传感器网络故障管理方法研究

发布时间：2019-06-22 16:49

【摘要】：由于无线传感器网络（Wireless Sensor Networks，WSNs）节点具有能量供应有限、硬件资源有限、节点数目众多、网络自组织和动态拓扑等特点，导致网络故障时有发生，即使设计成多冗余的容错系统，依然无法完全避免故障的发生。尤其是节点能量耗尽导致的能量故障非常普遍。因此，本文从节点能量耗尽导致节点失效进而引起网络故障的角度出发，研究故障预防、故障检测和故障修复等问题，基于减少能量消耗和均衡能量消耗的原则，设计相应的解决方法。本文的研究内容主要包括以下三个方面：（1）针对冗余通路故障预防方法中工作状态冗余节点过多、能量大量浪费的问题，提出一种基于冗余通路控制的WSNs故障预防方法。该方法利用汇节点收集网络内所有节点能量状态，，计算节点健康度等相关参数，使用A-Star算法选择最优工作通路，控制其余冗余通路分批轮流休眠，从而达到减少网络工作过程能量消耗，预防某些节点能量提前耗尽导致网络能量故障发生的目的。仿真实验和实际节点实验的结果表明，在保证网络适当冗余通路的前提下，与其它相关方法比较，该方法可以显著均衡网络能量消耗，有效延长网络寿命。（2）针对WSNs故障检测的邻居协作方法中，由于邻居数据与被检测节点实际数据存在差异导致的故障检测准确率问题，本文提出一种基于历史数据与邻居协作的节点自检测方法，并据此提出基于簇内数据聚类的故障检测方法来降低虚警率和普通节点的能量消耗。该方法将节点自检测方法和邻居协作方法相结合，首先由普通节点融合节点稳定的历史数据和实时邻居数据，采用分布式故障检测思想，完成自身故障检测工作，然后由能量不受限的汇节点利用聚类算法将按照空间位置分簇的WSNs节点感知数据进行聚类，并通过阈值比较进行故障检测。实验结果表明，与类似的WSNs故障检测方法相比，该方法具有良好的故障检测准确率和较低的故障检测虚警率，且普通节点能量消耗较小。（3）针对WSNs中节点故障导致网络出现“覆盖空洞”的问题，本文提出一种基于最小距离冗余节点替代的故障修复方法。该方法由能量不受限的汇节点获取全网所有工作节点和冗余节点的位置信息，通过故障节点和邻居节点的位置信息确定“覆盖空洞”面积，计算最佳修复位置，寻找与此位置距离最小的冗余节点作为替代节点加入网络工作，从而达到修复网络故障的目的。在此基础上，提出网络覆盖率的Monte Carlo积分计算方法，对网络的“覆盖空洞”边界进行数学描述，利用概率抽样的方法估算不规则区域的面积，实现对网络覆盖质量性能的评价。然后，根据实验数据，构造网络寿命与节点余活比经验公式、网络覆盖率与网络运行时间、节点余活比经验公式，为网络设计时根据监测区域面积、网络寿命期望和网络覆盖率等参数合理部署节点数目，提供理论参考。实验结果表明，通过预先随机部署的冗余节点，该故障修复方法可以获得较高的故障修复率和较好的网络覆盖，可有效地延长网络寿命。且在保证充分抽样的情况下，网络覆盖率的计算误差控制在5%以内，程序实现简单且计算结果可信。
[Abstract]:As the wireless sensor networks (WSNs) node has the characteristics of limited energy supply, limited hardware resources, numerous nodes, self-organization and dynamic topology of the network, the network failure occurs frequently, even if the multi-redundant fault-tolerant system is designed, It is still impossible to completely avoid the occurrence of the fault. In particular, the energy failure caused by the energy depletion of the node is very common. Therefore, this paper, based on the principle of reducing energy consumption and balancing energy consumption, design the corresponding solution based on the principle of reducing energy consumption and balancing energy consumption. The contents of this paper mainly include the following three aspects: (1) A WSNs fault preventing party based on redundant path control is proposed in order to solve the problem of excessive redundancy of redundant nodes and a large amount of energy in the redundant path fault prevention method. The method comprises the following steps of: collecting all node energy states in a network by a sink node, calculating the related parameters of the node health degree and the like, selecting an optimal working path by using the A-Star algorithm, and controlling the rest of the redundant paths to be alternately dormant in batches so as to reduce the energy consumption of the network working process; The purpose of preventing the failure of network energy due to the depletion of energy in some nodes. The results of the simulation experiment and the actual node experiment show that, on the premise of ensuring the proper redundant path of the network, the method can significantly balance the energy consumption of the network and effectively prolong the life of the network. In order to solve the problem of fault detection accuracy caused by the difference between the neighbor data and the actual data of the detected node, a node self-test based on historical data and neighbor collaboration is proposed in this paper. A fault detection method based on intra-cluster data clustering is proposed to reduce the false alarm rate and the energy of the common node. in that method, a node self-detection method and a neighbor cooperation method are combined, and then clustering the WSNs node-aware data according to the spatial position clustering by using a clustering algorithm by a sink node with an energy non-limitation, and comparing the data with a threshold value, The experimental results show that the method has good fault detection accuracy and low false alarm rate compared with similar WSNs fault detection method, and the energy of common node is eliminated. (3) In view of the problem of network "cover hole" caused by node failure in WSNs, this paper proposes a method based on minimum distance redundant node replacement The method comprises the following steps of: acquiring the position information of all the working nodes and the redundant nodes of the whole network by a sink node with an energy non-limitation, determining the area of the coverage hole by the position information of the fault node and the neighbor node, and calculating the maximum value of the coverage hole by the position information of the fault node and the neighbor node, the best repair location is to find a redundant node with the least distance from the location as an alternate node to join the network, thereby achieving the repair network The purpose of the fault is to put forward the Monte Carlo integral calculation method of network coverage, to describe the "cover hole" boundary of the network, to estimate the area of the irregular area by the method of probability sampling, and to realize the network coverage quality. then, according to the experimental data, the empirical formula of the network service life and the residual live ratio of the node, the network coverage rate and the network running time and the node residual live ratio experience formula are constructed, and the network design time is reasonable according to the parameters such as the monitoring area area, the network service life expectation, the network coverage rate and the like Number of nodes of UN-Women, mention The experimental results show that the fault repair method can obtain higher fault repair rate and better network coverage and can be effectively extended by the redundant nodes which are randomly deployed in advance. Long network life. In case of sufficient sampling, the error of network coverage is controlled within 5%, and the program is simple and easy to implement.
【学位授予单位】：哈尔滨工业大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP212.9;TN929.5

【参考文献】