【作者】 刘娟娟；

【导师】 刘文远；

【作者基本信息】 燕山大学 ， 计算机技术， 2012， 硕士

【摘要】 覆盖是无线传感器网络的一个基本问题。基于全向传感器模型的覆盖控制研究已经取得了一系列的成果。随着有向传感器节点的应用，现有的覆盖控制方法不再适用。本文针对随机部署的有向传感器网络，对网络的覆盖质量和监控目标质量进行了深入研究。首先，研究了基于暴露路径的网络覆盖质量度量算法。定义了有向传感器网络的感知度模型和暴露模型，通过格网分割将连续路径问题离散化，在此基础上设计了一个基于马尔科夫链的路径模拟算法，查找传感器网络中的暴露路径，特别是最小暴露路径。当目标移动模式未知时，用HMM完成自学习。提出两种覆盖质量度量标准，协作k覆盖和序列k覆盖，更好地评价了网络覆盖性能。其次，研究了有向传感器网络中面向移动目标的全方位覆盖问题，保证获得监测目标的正面图像，提高监测识别率，并克服了现有方法部署开销过大、灵活性差的缺点。推导确定了随机部署网络中节点的感知方向与监测目标表面矢量的观测角度下界θ，并证明满足θ的全方位覆盖指派是NPC问题，随之提出一个集中式算法和一个分布式算法调整节点的感知方向，使得利用少量节点全方位覆盖最多目标。最后，设计了完整的仿真实验来验证上述算法的正确性。仿真表明暴露路径可以有效地对传感器网络的覆盖质量进行度量，对有向传感器节点感知角度进行局部调整有效提高了监测对象的全方位覆盖率。更多还原

【Abstract】 The coverage in wireless sensor networks is a fundamental problem. A series ofachievements have been made in the study of coverage control for omnidirectional sensornetworks. However, with the development of directional sensor nodes, the existingcoverage control method fails to apply. In this paper, we address the problem of coveragecontrol in directional sensor networks with random distribution, and have in_depthresearch on the quality of coverage and targets under monitored.First, we study the quality of coverage based on exposure path in directional sensornetwork. We define the sensing model of directional sensor and exposure model. Then thecontinuous path is transformed into a discrete problem by the grid segmentation. On thebasis of this, an algorithm simulating the path of invaders based on Markov Chain isproposed, looking for the exposure path of the sensor network, especially the minimumexposure path. We also design a model of HMM for self learning on movement pattern oftargets. The algorithm is applicable under the condition of the given any distribution,density and network features of the sensor nodes.Then we study the full-view coverage problem in directional sensor network. Thefull-view coverage ensures that the positive images of targets are obtained. In existingworks, the topics only focus on whether a region is full-view covered or not, which needstoo many nodes, and is less flexible for the change of surveillance environment. Therefore,full-view coverage for the targets is highlighted in this paper. Firstly, we derive theoptimal observation angle between the sensing direction of nodes and the vector of targetsurface. We also prove that the full-view coverage is a NPC problem. Then the centralizedand the distributed algorithms are proposed to maximize the number of full-view coveredtargets by scheduling the sensing directions of some sensors in the network.Finally, we design a simulation experiment to verify the correctness of the algorithm.The results show the exposure path can effectively to metric the coverage quality of thesensor network, which is important for the network deployment, and is helpful to improvethe monitoring quality of targets. Results also show that local adjusting of the directionalnodes will increase the full-view coverage for the global monitored objects with random deployment.更多还原