基于时空相关性的无线传感器网络节能策略研究

发布时间：2018-03-13 00:01

本文选题：无线传感器网络　切入点：时间相关性　出处：《集美大学》2017年硕士论文　论文类型：学位论文

【摘要】：在智能交通系统(Intelligent Transportation System,ITS)的交通设施中增加一种无线传感器网络技术,能够从根本上缓解困扰现代交通的安全、通畅、节能和环保等问题,同时还可以提高交通工作效率。无线传感器网络(Wireless Sensor Network,WSN)通过无线传感器节点获取环境信息,自组织地进行无线通信和组网,在无线传感器网络中,单个传感器节点周期性采集到的数据在时间上可能是相关的,地理位置相邻的传感器节点收集到的数据在空间上往往也是相关的,于是,可以使用某种变换来去除其中的冗余信息,达到数据压缩、节省传输能耗的目的。本文对基于时空相关性的WSN节能策略进行研究。首先,在对周期性采集到的数据进行相关性分析,研究数据采集频率和数据失真度之间的关系,对各种有损压缩及无损压缩的性能进行比较的基础上,提出一种改进的LTC(Lightweight Temporal Compression)算法,该算法基于判断门限,将重构精度与采集频率进行折衷。MATALB仿真结果表明,算法在保证数据失真度的同时,可有效地抑制数据发送频率,减少传输冗余,从而达到节约能耗的目的。其次,讨论了节点间距离、信道衰耗等空间相关因素对失真度的影响,证明了在满足空间相关性的条件下,可以通过选用代表节点的方法,减少全网数据传输。同时引入相关半径概念,利用相关半径构建成相关簇,根据相关性系数和节点位置信息确定失真函数。合理选择簇头节点和发送数据方式,既有效地利用数据之间的空间相关性,保证数据失真在一定范围内,又避免了因数据传输量过大而能量消耗过大。并深入分析了GCC(Greedy Corrected Clustering,GCC)和K-Means两种与空间相关性结合的节点分簇算法,经MATLAB仿真比较得知,在对WSN中节点进行相关分簇时,这两种算法都有效的抑制了数据传输量,降低了数据冗余,从而网络能耗得到优化。另外,经数据对比可知,K-Means算法比GCC算法可以得到更加均匀的分簇,且在相同簇数量的情况下获得较小的平均失真。最后,利用相关性分析结果,在典型的LEACH(Low Energy Adaptive Clustering Hierarchy)算法中,将相关性分簇的K-Means和GCC算法应用于LEACH,仿真实验表明,在不同的场景中,不同的算法融合适用于不同的需求,算法融合可进一步实现网络能耗的节约,提高数据精度,减小失真,延长网络寿命。
[Abstract]:Adding a kind of wireless sensor network technology to the transportation facilities of Intelligent Transportation system (ITS) can fundamentally alleviate the problems of safety, smooth, energy saving and environmental protection that beset modern traffic. At the same time, it can also improve traffic efficiency. Wireless Sensor Network (WSNs) can obtain environmental information through wireless sensor nodes, self-organize wireless communication and network, in wireless sensor networks, The data collected periodically by a single sensor node may be related in time, and the data collected by a sensor node adjacent to a geographical location are often also spatially relevant. Some transformation can be used to remove the redundant information and achieve the purpose of data compression and energy saving. In this paper, the WSN energy-saving strategy based on temporal and spatial correlation is studied. Based on the correlation analysis of periodically collected data, the relationship between data acquisition frequency and data distortion, and the comparison of various lossy compression and lossless compression performance, an improved LTC(Lightweight Temporal compression algorithm is proposed. Based on the threshold, the reconstruction accuracy and acquisition frequency are compromised. The simulation results show that the algorithm can effectively suppress the data transmission frequency and reduce the transmission redundancy while ensuring the data distortion. In order to save energy consumption, the influence of spatial correlation factors, such as distance between nodes, channel decay and so on, on the distortion degree is discussed. It is proved that the method of representing nodes can be used to satisfy the condition of spatial correlation. At the same time, the concept of correlation radius is introduced, the correlation radius is used to construct the correlation cluster, and the distortion function is determined according to the correlation coefficient and node position information. It not only makes use of the spatial correlation between the data effectively, but also ensures the distortion of the data within a certain range. It also avoids the excessive energy consumption due to too much data transmission. Two node clustering algorithms, GCC(Greedy Corrected clustering (GCC(Greedy) and K-Means (K-Means), which are combined with spatial correlation, are analyzed in detail. The results of MATLAB simulation show that, when the nodes are clustered in WSN, These two algorithms can effectively suppress the amount of data transmission, reduce the data redundancy, and optimize the network energy consumption. In addition, the K-Means algorithm can get more uniform clustering than the GCC algorithm. In the case of the same number of clusters, the average distortion is small. Finally, in the typical LEACH(Low Energy Adaptive Clustering hierarchy algorithm, the K-Means and GCC algorithms of correlation clustering are applied to read. In different scenarios, different algorithms can be used to meet different requirements. Algorithm fusion can further achieve energy saving, improve data accuracy, reduce distortion and prolong network life.
【学位授予单位】：集美大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495

【参考文献】