基于分簇和小波压缩的传感网数据融合

发布时间：2018-03-30 03:43

本文选题：无线传感器网络　切入点：数据融合　出处：《西南大学》2017年硕士论文

【摘要】：传感器网络技术作为21世纪最重要的技术之一,很大程度上改变了我们的生活。通过传感器获取的传感数据,再通过物联网的使用将会使数据变得更加有用,而物联网的应用则比传感器网络更加多样化,因此物联网可以看成是将传感器网络的功能实现了更大的延伸。在大型传感器网络监测环境中会布署大量的传感器节点,在节点密集的区域,传感器之间的感知数据存在多种差异性和时空相关性。这些情况可能会导致多传感器采集的数据之间可以存在着某些对应的关系或者相似的数据。为了有效地提取信息,通常会将数据融合技术加入在数据收集的过程中,该技术主要目的就是在满足一定的应用监测需求下,将多份数据通过去冗余实现更有效的传输。在成本和体积限制的情况下,无线传感器网络中的节点一般都是能源有限的。而无线传感器网络面向应用需求设计可以通过引入数据融合技术获得更大效益。本文通过深入研究典型的时空相关性数据融合算法的特点、原理和相应的性能指标,结合应用需求与无线传感器网络的采集数据之间的时空相关性,提出了相应的传感器网络数据收集分簇模型,针对收集的数据之间仍存在的冗余性,设计了相应的提升小波数据压缩去冗余算法。本文主要研究内容和贡献主要有:(1)本文针对无线传感器网络监测区域环境中临近节点之间采集的感知数据的时间和空间相关性,提出了一种基于应用需求的传感器网络数据收集分簇模型。通过临近区域的数据相关性大小,先选取一定剩余能量高、数据代表性强的节点作为簇成员,参与到传感器网络数据收集中,其次,根据这部分成员节点与其余普通节点的数据相关度比较来选择一部分不可以被替代的普通节点,并将其加入簇内参与数据传输。并且根据剩余能量和实时数据的变化,动态地调整网络分簇的大小和检测异常节点的数据变化。通过模拟仿真实验,结果表明该分簇模型能够在满足一定的数据变化传输要求下节省更多的能量。(2)本文针对区域单个簇头节点所收集到的簇内节点数据的空间相关性和不同时刻传输的数据之间的时间相关性,我们提出了一种有效的消隐式提升小波数据压缩算法,该算法不仅能够去除大量的数据冗余,而且计算速度快,占用内存少,在汇聚节点的恢复数据精度也比较高,从而使无线传感器网络获得了更大的效益。
[Abstract]:Sensor network technology, one of the most important technologies of the 21st century, has greatly changed our lives. Sensor data obtained through sensors, and then the use of the Internet of things, will make the data more useful. The application of the Internet of things is more diversified than that of the sensor network, so the Internet of things can be regarded as a larger extension of the function of the sensor network. A large number of sensor nodes will be deployed in the monitoring environment of a large sensor network. In areas where nodes are dense, There are many differences and spatiotemporal correlations between sensors. These situations may lead to some corresponding relationship or similar data between the data collected by multi-sensor. In order to extract information effectively, Data fusion technology is usually added to the process of data collection. The main purpose of this technology is to achieve more efficient transmission of multiple data through de-redundancy under certain application monitoring requirements. The nodes in wireless sensor networks are generally limited in energy, but the application-oriented design of wireless sensor networks can achieve greater benefits by introducing data fusion technology. In this paper, the typical space-time phase is studied in depth. The characteristics of correlation data fusion algorithm, The principle and the corresponding performance index, combined with the temporal and spatial correlation between the application requirements and the collected data of the wireless sensor network, the corresponding clustering model of the sensor network data collection is proposed, aiming at the redundancy between the collected data. A corresponding lifting wavelet data compression and de-redundancy algorithm is designed in this paper. The main research contents and contributions of this paper are as follows: (1) this paper focuses on the temporal and spatial correlation of perceptual data collected between adjacent nodes in the wireless sensor network monitoring region. A cluster model of sensor network data collection based on application requirement is proposed. According to the data correlation of adjacent region, the nodes with high residual energy and strong data representation are selected as cluster members. Participate in the sensor network data collection, secondly, according to this part of the member nodes and the other common nodes data correlation comparison to select a part of the ordinary nodes can not be replaced, It is added to the cluster to participate in data transmission. According to the change of residual energy and real time data, the size of network clustering and the data change of detecting abnormal nodes are dynamically adjusted. The results show that the clustering model can save more energy under certain data transmission requirements.) in this paper, the spatial correlation of data collected by a single cluster head node and its transmission at different times are discussed. The temporal correlation between the data, We propose an effective lifting wavelet data compression algorithm. This algorithm can not only remove a large amount of data redundancy, but also has the advantages of fast computing speed, less memory, and high accuracy of data recovery at convergent nodes. Thus, the wireless sensor network gets more benefit.
【学位授予单位】：西南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN929.5;TP212.9

【参考文献】