WSN中基于多分辨率和压缩感知的数据融合方案

当前基于压缩感知的传感器网络数据融合方案中,不论数据字段有何特征,均假设网络具有固定而均匀的压缩阈值,从而导致数据通信量过高,能耗浪费较大。提出一种基于多分辨率和压缩感知的数据融合方案。首先,对传感器网络进行配置,以生成多个层次类型不同的簇结构,用于过渡式数据收集,在该结构上,最低层的叶节点只传输原始数据,其他层的数据收集簇进行压缩采样;然后将其测量值向上发送,当母数据收集簇收到测量值时,利用基于反向DCT和DCT模型的CoSaMP算法恢复原始数据;最后,在SIDnet-SWANS平台上部署了该方案,并在不同的二维随机部署传感器网络规模下进行了测试。实验结果表明,随着分层位置的变化,大部分节点的能耗均显著降低,与NCS方案相比,能耗下降50%~77%,与HCS方案相比,能耗下降37%~70%。     

基于压缩感知的无线传感器网络数据收集方案

提出一种应用于大规模环境监测领域无线传感器网络(WSN)中改进的基于压缩感知(CS)数据收集方案。该方案改进联合稀疏模型(JSM)进行数据分析;并采用分簇路由采集并传输数据,即每簇簇内使用基于CS的数据收集方法,簇头之间采用最短路径路由到达汇聚节点。仿真结果表明,该方案不仅缩小了数据处理的分布范围,降低了恢复误差,而且大大减少了数据传输次数,维持了整个网络的能耗平衡。     

基于簇的块稀疏压缩感知的60GHz信道估计

基于压缩感知设计适用于60 GHz毫米波通信系统的信道估计方案,深入研究了正交匹配追踪(OMP)算法和正则正交匹配追踪(Regularized OMP)算法的60 GHz信道估计性能;在此基础上,充分发掘60 GHz无线多径信道所呈现出的分簇特性,提出一种新颖的基于簇分级的稀疏压缩感知重构算法。新算法在有效减少重构迭代次数的前提下,亦能显著降低信道估计误差。综合对比分析了基于簇分块稀疏压缩感知重构算法和现有压缩感知算法在60 GHz信道估计应用中的重构性能,仿真结果表明,压缩感知算法可有效应用于60 GHz系统信道估计,而新设计的基于簇分级的稀疏压缩感知算法则在估计精度和实现复杂度方面具更优越性能。     

基于位置感知的无线传感器网络聚类算法

针对以能量有效的方式收集传感器网络空间相关性数据的问题,本文提出了一种新的基于位置感知的无线传感器网络聚类算法。算法根据用户查询误差门限和基于位置信息的节点感知数据相异度矩阵,进行无监督数据挖掘,将监测区域划分成信息等价域。每个等价域选取城内当前剩余能量最大的节点作为簇头,网络通过移动代理收集簇头感知信息,从而减少了传输数据量,有效节省了网络能量。     

一种基于El Gamal加密算法的压缩数据收集方法

压缩感知技术在信号处理、图像处理、数据收集与分析等方面有很大优势,是近年来的研究热点。研究了如何安全高效地运用压缩感知技术来收集无线传感器网络中的数据。传统的基于压缩感知技术的数据收集方法并不考虑数据收集的安全性,而且网络内的所有节点都会参与每个测量值的收集。将El Gamal加密算法和基于稀疏随机矩阵的压缩感知技术相结合,提出了一种基于El Gamal加密算法的稀疏压缩数据收集方法（El Gamal based sparse compressive data gathering,ESCDG）。理论分析和数值实验表明,ESCDG不仅能降低网络资源的消耗而且能抵御多项式算力的内部攻击和外部攻击。     

一种高覆盖率和安全性的动态传感器网络部署策略

对于资源受限的移动传感器网络的部署问题,提出了采用多目标遗传算法的动态和安全部署策略。具体实现是将随机部署的移动传感器划分和放置到一个具有簇头和最佳路由的独立簇的最优数目中,簇头从其成员传感器收集数据,并通过最具成本效益的路由以压缩和安全的方式将收集到的数据发送到接收器。一旦部署完毕,通过移动(或重定位)自身来最大化它们的覆盖范围,从而实现目标的安全覆盖和监测。实验结果表明,相比于静态部署,本文提出的动态部署不仅提高了传感器的覆盖率,而且在保证安全性的同时降低了节点丢失百分比。     

一种基于随机滤波的神经动作电位信号压缩感知采样方法

神经元细胞通过动作电位进行通信,这些动作电位包含了神经元活动的最关键信息。在论述动作电位在离散小波变换(DWT)域的稀疏特性和分析压缩感知(Compressive Sensing, CS)测量动作电位的基础上,该文提出了一种采用随机滤波对动作电位进行压缩感知测量的方法。从信号恢复和物理可实现性两方面,对比了3种压缩测量的实现方法,从对实测数据的处理结果说明基于随机滤波CS的动作电位测量方案是一种能对动作电位进行压缩,系统复杂度低且最易实现的压缩采样方案。     

无线传感器网络中基于压缩感知和GM(1,1)的异常检测方案

针对现有的异常事件检测算法准确率低和能量开销较大等问题,该文提出一种基于压缩感知(CS)和GM(1,1) 的异常事件检测方案。首先,基于分簇的思想将传感器节点的数据进行压缩采样后传输至Sink,针对传感器网络中数据稀疏度未知的特点,提出一种基于步长自适应的块稀疏信号重构算法。然后,Sink基于CM(1,1)对节点发生的异常进行预测,并对节点的工作状态进行自适应调整。仿真实验结果表明,相比于其它异常检测算法,该算法的误警率和漏检率较低,在保证异常事件检测可靠性的同时,有效地节省了节点能量。     

基于移动Agent和WSN的突发事件场景数据收集算法研究

该文针对无线传感器网络应用于突发事件监测场景的能量消耗和网络延迟问题,提出了基于移动Agent的无线传感器网络簇式数据收集算法.动态成簇过程基于事件严重程度,并由其决定簇的生命周期和覆盖范围.Sink和簇头之间形成以Sink节点为簇头的虚拟簇.移动Agent迁移路径规划过程中下一跳节点的选取基于节点剩余能量、路径损耗及受刺激强度.移动Agent通过节点遍历的方式完成对所有簇内成员节点信息的收集.仿真结果表明,相对于C/S数据收集模型,基于移动Agent的模型具有更好的节能效果,并能一定程度地减少网络延迟,尤其适用于大规模无线传感器网络应用.     

基于二维压缩感知的定向遥感和变化检测

基于高斯测量矩阵的一维压缩感知测量数据不仅能很好地保持稀疏信号的能量信息, 也能够很好地继承稀疏信号的方向信息.但是在一维压缩感知模型中方向信息无法应用于稀疏信号的重构和检验.针对遥感影像中变化区域稀疏的特点提出了二维压缩感知模型.并利用能量和方向信息构建了基于二维压缩感知的稀疏信号重构算法(2DOMP).理论分析和实验结果证明, 2DOMP算法的信号重构能力更强.同时根据压缩感知恢复稀疏信号只需要很少测量数据的特性提出了定向遥感和定向变化检测的概念.     

基于CS的机载分簇型WSN数据采集方法

提出一种适用机载分簇型WSN的数据采集方案。该方案一方面采用随机压缩采样的方式,有效地减少了硬件资源受限的簇成员节点的采样数据量,降低了簇成员节点对硬件资源的要求;另一方面,提出一种基于复合混沌—遗传算法的MP重构方法,将混沌理论良好的局部寻优特性与遗传算法强大的全局搜索能力相结合,有效提高了簇头或Sink中信号重构的性能。实验结果表明,该方案在有效减少簇成员节点采样数据量,且采样频率降为原采样频率1/8的基础上,仍能保证10-7数量级的重构精度,为实用型WSN的实现提供了有效借鉴。     

Performance Optimization Based on Compressive Sensing for Wireless Sensor Networks

A routing algorithm, based on a dual cluster head redundant mechanism combined with compressive sensing data fusion algorithm, is proposed to improve reliability and reduce data redundancy of the industrial wireless sensor networks. The Dual cluster head alternation mechanism is adopted to balance the energy consumption of cluster head nodes. Through the compressive sensing data fusion technology to eliminate redundancy, effectively improve the network throughput of the sensor network. The simulation results show that the proposed algorithm is able to enhance the networks performance, significantly reduces the number of lost packets and extend the network’s lifetime.     

A novel approach for designing delay efficient path for mobile sink in wireless sensor networks

In the recent years, the use of mobile sink has drawn enormous attention for data collection in wireless sensor networks (WSNs). Mobile sink is well known for solving hotspot or sinkhole problem. However, the design of an efficient path for mobile sink has tremendous impact on network lifetime and coverage in data collection process of WSNs. This is particularly an important issue for many critical applications of WSNs where data collection requires to be carried out in delay bound manner. In this paper, we propose a novel scheme for delay efficient trajectory design of a mobile sink in a cluster based WSN so that it can be used for critical applications without compromising the complete coverage of the target area. Given a set of gateways (cluster heads), our scheme determines a set of rendezvous points for designing path of the mobile sink for critical applications. The scheme is based on the Voronoi diagram. We also propose an efficient method for recovery of the orphan sensor nodes generated due to the failure of one or more cluster heads during data collection. We perform extensive simulations over the proposed algorithm and compare its results with existing algorithms to demonstrate the efficiency of the proposed algorithm in terms of network lifetime, path length, average waiting time, fault tolerance and adaptability etc. For the fault tolerance, we simulate the schemes using Weibull distribution and analyze their performances.     

无线传感器网络中基于压缩感知的动态目标定位算法

传统的动态目标定位算法需要采集、存储和处理大量数据,并不适用于能量受限的无线传感器网络。针对该缺陷,该文提出一种基于压缩感知的动态目标定位算法。该算法利用目标的运动规律设计稀疏表示基,从而将动态目标定位问题转化为稀疏信号恢复问题。针对传统观测矩阵难以实现的缺陷,该算法设计可实现且与稀疏表示基相关性低的稀疏观测矩阵,从而保证了算法的重构性能。该算法的特点是可利用较少的数据采集实现动态目标定位,从而大大延长无线传感器网络的寿命。仿真结果表明,该文所提出的基于压缩感知的动态目标定位算法具有较好的定位性能。     

基于最小生成树的分簇协作频谱感知改进算法

为减少分簇过程中的时延,基于最小生成树的单向比较优势提出簇首快速推举方法,并提出改进的分簇协作频谱感知算法,分析了算法的时间复杂度。算法首先基于最小划分对所有次用户节点进行分簇,簇内节点根据设置的评价条件进行性能比较,推举簇首。由簇首进行本地簇内频谱检测,并上传检测结果,最后融合中心在簇首间实现协作的频谱检测。在瑞利信道条件下,仿真显示在大信噪比时,融合中心应用AND规则,系统具有较小的虚警率,所提算法检测性能优;小信噪比时,应用OR规则能扩展系统的有效检测区间,所提算法在满足系统要求的前提下检测性能较差,但簇内信道效率提高了n-1倍。     

基于压缩感知的免携带设备双目标定位算法

免携带设备的目标定位(DFL)不需要目标携带任何设备就能获取位置信息,针对现有算法在多目标定位中存在的因射频信号时变特性引起的问题,该文结合指纹法,提出了基于压缩感知的免携带设备双目标定位算法。该算法采用中心概率覆盖模型建立单目标射频地图到双目标射频地图的映射关系,解决指纹法由于目标数的增加引起的离线训练量骤增的问题。并采用K-means聚类方法对双目标射频地图进行分类,通过类匹配缩小定位区域的范围,降低定位算法的复杂度。最后利用压缩感知的方法,将定位问题转化成稀疏信号的重构问题,提高了定位精度。实验结果表明,与基于无线层析成像的压缩感知定位算法相比,该算法能达到较高的定位精度,且实时性更高。     

WSN data gathering algorithm based on compressive sensing and matrix completion technique

The unreliable links and packet losing are ubiquitous in WSN.The performance of data collection algorithm based on compressive sensing is sensitive to packet losing.Firstly,the relationship between packet loss rate and CS-based reconstruction precision was analyzed,and the sparsest block measurement (SBM) matrix was formulated to keep the data gathering consumption smallest and make sure the low-rank property of measurements.Then,combined with the matrix completion (MC) and compressive sensing (CS),the CS data gathering algorithm based on sparsest block measurement matrix (CS-SBM) algorithm was proposed.CS-SBM gathered data in a period and recovered the loss data based on MC to weaken the impact of packet loss on data gathering.CS-SBM reconstructed data based on CS to reduce measurement number and energy consumption and prolong the network lifetime.Simulation analysis indicates that the proposed algorithm reconstruct the whole data with high-accuracy under 50% packet loss rate,resisting unreliable links effectively.     

Channel assignment scheme in clustered multichannel cognitive radio networks with outdated CSI over Rayleigh fading channels

This work investigates channel assignment for cooperative spectrum sensing in multichannel cognitive radio networks, where the heterogeneity of primary user (PU) activity and the effect of varying channel condition on the received signal‐to‐noise ratio during cluster formation are considered. With the objective to minimize interference to the PU while enhancing multiple spectrum utilization of the secondary user (SU), an overlapping cluster‐based assignment is formulated into a nonlinear integer optimization problem. To obtain an efficient solution, the nonlinear integer problem is transformed into a mixed integer linear problem, based on which, this paper proposes an exact solution and then two new heuristic algorithms for suboptimal solutions, respectively. Furthermore, a comparative study of four different cluster head selection schemes with respect to their performance in cooperative spectrum sensing, under cluster's heterogeneity in terms of SUs distribution relative to PU transmitter location is presented. Based on the study, a robust cluster head selection scheme is proposed. Simulation results show that good sensing performance and increased opportunistic spectrum utilization in multichannel cognitive radio networks are two sides of a coin that depend on the ratio of the SUs to the number of PU channels. How far away the PU is from the cluster center is also seen to be key in the optimal selection of cluster heads in cooperative spectrum sensing.     

Constructing a MANET Based on Clusters

This paper proposes a scheme for constructing a mobile ad hoc network (MANET) based on clusters. The proposed MANET architecture is made up of two hierarchies. One hierarchy is the backbone network which is made up of cluster heads and associate nodes, and the other hierarchy is the cluster which is made up of one cluster head and multiple cluster members. In the proposed cluster generation algorithm, the number of potential cluster members is used as a metric, and it is always the new node with the maximum number of potential cluster members that is elected as a cluster head. In this way, the number of cluster heads is minimized. In this scheme, only one associate node is used to achieve the communication between two cluster heads, so the number of nodes included in the backbone network is minimized. This scheme also proposes the cluster merging algorithm in order to maintain the minimum number of cluster heads. In the proposed cluster repair algorithm, if a cluster head fails/moves out of the cluster, then a new cluster head is elected to maintain the cluster stability. From the perspective of the cluster stability, this paper analyzes the performance parameters of the proposed scheme, and the data results show that the proposed scheme improves the MANET stability.     

基于线性混合模型的高光谱图像谱间压缩感知重构

该文针对高光谱数据的线性混合模型,提出一种简单有效的谱间压缩感知下高光谱数据的重构方案。该方案不同于传统的压缩感知重构方法直接重构高光谱数据,而是将高光谱数据分离成端元和丰度分别进行重构,然后利用重构的端元和丰度信息合成高光谱数据。实验结果表明,该方案的重构质量明显优于标准压缩感知重构方法,并且运算速度具有极大提升,同时便于获得端元和丰度信息。