【作者】 向敏；

【导师】 石为人；

【作者基本信息】 重庆大学 ， 控制理论与控制工程， 2009， 博士

【摘要】 无线传感器网络是涉及多学科知识的新型网络,综合了传感器技术、网络技术与无线通信技术等,应用前景十分广泛,是21世纪前沿新技术之一。无线传感器网络节点能量有限,能量高效的数据管理和延长网络寿命是该领域很多学者十分关注的问题。传感器节点是一个独立的计算和控制单元,能够实现自身的数据管理,即数据感知、分析、转发以及自身状态的控制,或者是与其它节点协同完成数据管理。节点数据管理是无线传感器网络数据管理的组成部分,它与网络拓扑结构、节点自身特性以及节点感知数据密切相关。如何有效地管理节点数据对改善网络能量效率和延长网络寿命具有重要意义,目前国内外尚缺乏无线传感器网络节点数据管理技术的具体研究。通过对节点数据的有效管理,为查询者提供可靠数据,减少通信中不必要的广播能耗和数据转发能耗,从而提高整个网络能量利用率和延长网络寿命。本文围绕无线传感器网络节点数据管理和能耗处理技术,着重从节点数据管理特点、网络拓扑控制、节点分类管理和节点感知数据预测四个方面进行了比较系统的研究,具体研究内容如下:①结合无线传感器网络及其节点的特性,讨论了无线传感器网络数据管理与节点数据管理的关系。分析得到良好的拓扑管理,合理的节点感知数据处理和节点调度控制对节点数据管理与降低网络能耗有着重要作用。最后讨论了节点数据管理与能耗处理技术的主要研究内容。②分析了现有分簇算法的优缺点,提出了基于参数优化的分簇算法。以降低簇间通信能耗为目标,给出了簇间最优单跳距离、分簇角与节点物理参数和节点数目的关系;以减少簇头更换频率和降低簇内广播能耗为目标,提出了簇头连续担任本地控制中心直至其工作次数到达最优值才被候选簇头替换的簇内数据收集机制。仿真结果表明所提分簇算法能够有效降低簇内、簇间通信能耗,并能显著延长网络寿命。③利用节点的计算和分析功能,提出了基于感知数据综合支持度的节点分类算法。簇头利用误差函数和模糊函数分析成员感知数据的关联性,获取节点感知数据综合支持度,由此将成员节点划分为冲突节点、补充节点和可靠节点。针对休眠的节点,通过分析节点感知数据综合支持度及其增量,给出了相应的休眠控制规则。针对具有高综合支持度的冗余节点,给出了相应的调度规则以降低簇头能耗和尽可能实现簇间节点能耗均衡。仿真结果表明算法能够实现簇内节点分类,降低簇内数据收发量并能有效延长网络寿命。④以减少簇内数据收发量为目标,提出了面向数据收集的节点数据预测算法。簇头采用GM(1, 1)预测模型和动态更新参数阵列的机制实现对部分成员感知数据的预测。针对簇头不同的预测模式,给出了被预测节点的两种调度机制即顺序调度和选择性调度以及簇头数据融合的处理方法。实验和仿真结果表明采用预测算法能够准确预测节点感知数据,并能有效改善网络能耗和延长寿命。论文最后对节点数据管理及能耗研究所提出的算法及相关的工作进行了总结,并对有待于进一步研究的课题和方向提出自己的思考。更多还原

【Abstract】 Wireless Sensor Networks (WSN) is a new type network made up of sensor, network and wireless communication technologies and has a wide application future, and it is one of the new and high technologies in the 21st century.The nodes of WSN are extremely power constrained, so energy-efficient data management and prolonging the networks lifetime are the major concerns to many scholars in this researching area. Each node of WSN is an independent computing and controlling unit which can achieve their own data management such as sensing data, analyzing data, transmitting data and controlling their own state, or collaborate with other nodes for the data management. Node data management is a part of WSN data management, and it is closely related to network topology, node characteristics and node sensing data. How to effectively manage nodes’data is very important to improve the networks energy efficiency and extend the networks lifetime, and there is little research result for WSN node data management.Through energy-efficient node data management, the reliable data can be provided to the users, and the unnecessary energy consumption for broadcasting message and transmitting data can be reduced, and then the entire network energy efficiency can be improved significantly and the network lifetime can be extended. In this paper, the technologies of node data management and energy consumption are studied mainly from the characteristics of the node data management, network topology control and node classification management and sensing data prediction of nodes. The specific studies are as follows.①According to the characters of WSN and its nodes, the relationship between the WSN data management of and the node data management are discussed. The good topology management, correct processing of the node data and the rules of scheduling node are very important to the node data management and energy consumption optimization. The main studies of node data management and energy consumption optimization are discussed.②By analyzing the advantages and disadvantages of the existent clustering algorithms, a new clustering algorithm based on optimum parameters is presented. The relationships of the optimum one-hop distance and clustering angle with the nodes electronic parameters and the number of the total nodes are given for minimizing the energy consumption between inter-cluster communications. Furthermore, the continuous working mechanism of each cluster head which acts as the local control center and will not be replaced by the candidate cluster head until its continuous working times reach the optimum values is given. The simulation results demonstrate that the presented clustering algorithm can effectively reduce the energy consumption used for intra-cluster broadcasting message and gathering data, and prolong the network lifetime.③With calculation and analysis function of node, a node classification algorithm based on the integrative supportability of sensing data is presented. Each cluster head analyzes its member data correlation using error function and fuzzy function, and gains the integrative supportabilities of its members’sensing data. Based on the integrative supportabilities, the members of the cluster are classified as conflict nodes, complementary nodes and reliable nodes. The sleeping rules are given according to the node’s integrative supportability and its increment, and the controlling rules for the redundant nodes with high integrative supportabilities are given to reduce the energy consumption of the cluster and balance the energy consumption among members. The simulation results demonstrate that the presented algorithm can realize the node classification, reduce the amount of the data transmission and prolong the network lifetime.④In order to reduce the amount of data transmission, a node data prediction algorithm for the data collection is presented. The cluster head predicts some members sensing data with the basic GM (1, 1) prediction model and the mechanism of dynamically updating array parameters. According to the different predicted modes of the cluster head, the two predicted scheduling mechanisms contained ordinal scheduling and selective scheduling are presented and the data fusion algorithm is given. The test and simulation results demonstrate that the presented algorithm can accurately predict the node sensing data, improve the energy efficiency and prolong the network lifetime.The last chapter concludes the presented algorithms and the related work for the study of the node data management and energy consumption, and outlines the further research contents and directions.更多还原