无线传感器网络中基于压缩感知的数据收集方法研究

发布时间：2018-07-05 01:07

本文选题：无线传感器网络 + 压缩感知　；参考：《湘潭大学》2017年硕士论文

【摘要】：随着无线传感器网络(Wireless Sensor Network,WSN)在军事防御、环境监测、医疗诊断、智能交通等领域的越来越广泛使用,利用WSN实现目标对象实时监测采集信号的研究也备受关注。WSN通常由大量体积小、成本低、不受环境限制的传感器节点密集的布置在感知区域。但是由于单个节点的资源有限,而网络中的通信量巨大,导致节点能源耗费过快,网络寿命不长。而节点间数据通信所消耗的能量占节点总能源耗费的比重最大,因此减少网络通信量可实现能量有效的数据收集。压缩感知(Compressive Sensing,CS)技术利用相邻节点在一段时间内采集的数据存在时空相关性的特点,减少网络冗余数据的采集和收发,不仅能够减少网络通信量还能平衡通信负载。但是,普通压缩感知技术造成了节点早期的通信负载过高,据此,混合压缩感知提出仅在通信量高于压缩感知瓶颈的节点中使用CS技术处理数据,否则传输原始数据。因此,针对如何减少WSN数据通信量、提升WSN寿命期限,本文提出一种结合混合CS技术的分簇式WSN数据收集方法。首先,按地理位置划分感知区域成若干簇,并假设各簇区域中心存在一个虚拟簇头节点,且选取虚拟簇头节点一跳通信范围内的节点为候选簇头节点;其次,使用Prim算法以sink为根节点连接各虚拟簇头节点生成一棵最小生成树;然后,从sink节点开始,为最小生成树各分支中的簇从候选簇头节点中动态规划选出簇头节点;最后,构造以sink节点为根节点且按最小生成树顺序连接各簇头节点的数据传输骨干树。仿真结果证明,当压缩率为10时,本文算法通信量比clustering without CS、SPT without CS、SPT with hybrid CS、以及clustering with hybrid CS,分别减少了65%、55%、40%和10%。
[Abstract]:With the increasing use of Wireless Sensor Network (WSN) in the fields of military defense, environmental monitoring, medical diagnosis, intelligent transportation, etc. The research of using WSN to realize real-time monitoring and collecting signal of target object is also concerned. WSN is usually arranged in the sensing area by a large number of sensor nodes which are small in volume, low in cost and not restricted by environment. However, due to the limited resources of a single node, and the huge amount of communication in the network, the node energy consumption is too fast, the network life is not long. The energy consumed by the data communication between nodes accounts for the largest proportion of the total energy consumption, so the energy efficient data collection can be realized by reducing the network traffic. Compressed-sensing (CS) technology can reduce the collection and transmission of redundant data in the network by utilizing the spatio-temporal correlation of the data collected by adjacent nodes over a period of time. It can not only reduce the network traffic but also balance the communication load. However, the common compression sensing technology causes the high communication load in the early stage of the nodes. Therefore, the hybrid compression sensing technology only uses CS technology to process the data in the nodes where the traffic is higher than the bottleneck of compression perception, otherwise, the original data is transmitted. Therefore, in order to reduce the data traffic and increase the lifetime of WSN, a clustering data collection method based on hybrid CS technology is proposed in this paper. Firstly, the perceptual region is divided into several clusters according to geographical location, and a virtual cluster head node is assumed to exist in the center of each cluster region, and the node in the one-hop communication range of the virtual cluster head node is selected as the candidate cluster head node. Using Prim algorithm to connect each virtual cluster head node with sink as the root node to generate a minimum spanning tree; then, starting from the sink node, the cluster heads in each branch of the minimum spanning tree are dynamically programmed from the candidate cluster header node. Finally, the cluster head node is selected from the candidate cluster head node. The data transmission backbone tree with sink node as the root node and connected each cluster head node in the order of minimum spanning tree is constructed. The simulation results show that when the compression ratio is 10:00, the traffic of our algorithm is 65% and 10% less than that of clustering without with hybrid CSand clustering with hybrid CSS, respectively.
【学位授予单位】：湘潭大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212.9;TN929.5

【参考文献】