【作者】 唐诗奇；

【导师】 李迅；

【作者基本信息】 国防科学技术大学 ， 控制科学与工程， 2011， 硕士

【摘要】 无线传感器网络(WSNs)的时空相关性为数据融合发展带来了深远的意义。本文主要研究了时空相关性在无线传感器网络数据融合中的应用,分析了点源信号的时空相关性模型,并基于点源信号的时空相关性分别对虚拟采样算法(VSS)、WSNs自适应数据融合算法和WSNs主动缓冲管理算法进行了研究:一.改进了点源信号模型,在数据融合模型基础上推导了点源信号的时间相关性模型、空间相关性模型及时空相关性模型及对应的时间相关性均方误差函数、空间相关性均方误差函数和时空相关性均方误差函数,并根据时空相关性均方误差函数对时空相关性进行了分析。二.基于时空相关性研究了能量高效的数据融合算法——VSS。VSS利用节点间的冗余性,获得节点集的多个子集,让子集中部分节点对环境进行采样,不进行采样的节点处于低能耗的睡眠状态,VSS利用时空相关性进行虚拟分簇,虚拟分簇机制适合以目标追踪为背景的应用研究,对感知对象状态的不确定性有较好的自适应能力。VSS采用分布式虚拟循环采样的方法进行采样,VSS在实现保存有意义的信息的基础上能有效的平衡各节点的能量消耗,减少冗余数据。三.基于时空相关性提出了一种自适应数据融合算法。首先研究了基于时间相关性的自适应采样算法,并根据时间相关性均方误差函数确定节点所处采样状态,再根据节点所处状态自适应调整采样频率。该算法自适应性强,可以有效的捕捉目标速度的变化,进行自适应采样频率调整。其次研究了基于空间相关性的空间融合度自适应调整算法,根据空间相关性均方误差函数确定满足可靠性要求的最小节点数目和满足空间冗余度要求的最大节点数目,然后根据以上数据定义空间融合状态,再依据融合所处的状态自适应调整空间融合度。该算法使得每次融合既可以满足跟踪目标的可靠性,又可以满足设计的空间冗余度要求,降低了能量消耗,保证了采集数据的精确度。四.基于空间相关性研究了WSNs主动缓冲管理算法。从计算数据包丢失概率、数据包的选择和主动缓冲算法三方面对基于空间相关性的主动缓冲管理算法进行了详细介绍。不同于其他利用数据包优先权思想的缓冲管理算法,该算法用真实队列长度代替平均队列长度计算数据包丢失概率,依据感知数据的空间相关性选择丢弃数据包,通过理论分析和仿真证明:该算法可以有计划地丢掉冗余数据,保证了丢包的公平性,且较早的丢掉数据包有效缓冲数据冲突。更多还原

【Abstract】 Spatial-temporal correlation along with the collaborative nature of the Wireless sensor network brings significant potential advantages for the development of data fusion. This paper focuses on the application of spatial-temporal correlation in data fusion for wireless sensor network. We analyze the spatial-temporal correlation models and do research on virtual sampling scheme, adaptive data fusion scheme and a active buffer management algorithm based on the spatial-temporal correlation.Firstly,this paper improves the point source model and analyzes the temporal correlation, spatial correlation and spatial-temporal correlation along with the distortion function of them basing on data fusion model, Also, analysis of the spatial-temporal correlation is carried out basing on the distortion function of it.Secondly,this paper studies VSS based on spatial-temporal correlation. VSS primarily utilizes redundancy in the nodes to get some subsets to sample the environment at any one time. Nodes not sampling the environment are in low-power sleep mode. The virtual cluster technique based on the spatial-temporal correlation fits for target tracking application in researches. The virtual cluster network has better adaptive capabilities for the uncertainty of the object state. Furthermore, VSS can balance the energy consumption amongst nodes by using a round robin method and reduce redundant sensor data to conserve energy while retaining the meaningful information.Thirdly, the paper proposes a novel adaptive data fusion algorithm basing on the spatial-temporal correlation. Firstly, it proposes a adaptive sampling algorithm based on the temporal-correlation. The algorithm determines the state of node sampling according to the distortion function of temporal correlation, and then adjusts the sampling frequency adaptively according to the different states. The algorithm is so strong self-adaptive that it can effectively capture the change of target and adjust the sampling frequency adaptively. Secondly, basing on the spatial correlation, it proposes an algorithm that can adjust the degree of spatial data fusion adaptively. Basing on the distortion function of spatial correlation, the minimum number of reliable nodes and the maximum numbers of redundancy nodes are determined, according to which the spatial state is determined. And the degree of spatial fusion is adjusted. The algorithm makes fusion meet the reliability and redundancy requirement and reduce the energy consumption, and ensure the accuracy of data collection.In the end, this paper designs a buffer management algorithm for wireless sensor nodes basing on the spatial-temporal correlations. It introduces the buffer management algorithm from three parts: calculating queue’s probability to drop a packet, selecting a packet and active buffer management algorithm based on spatial correlation. Differing from other buffer management algorithm based on packet priority, the algorithm uses real queue length instead of average queue length. thus it is easy to calculate queue length and can drop packets on the spatial correlation, both theory analysis and simulation results prove that the algorithm is designed to drop redundant data and drop packets in a early time in order to buffer burst data in conflict and keep fair among the sub-clusters.更多还原