【作者】 王瑞锦；

【导师】 秦志光；

【作者基本信息】 电子科技大学 ， 信息与通信工程， 2013， 博士

【摘要】 在无线传感器网络中，信息采集、目标跟踪、信息管理、环境监测和基于地理位置的消息路由等许多应用中都需要准确的节点位置信息作为保障。因此，定位技术是无线传感器网络的关键支撑技术。在现实应用环境中，网络呈现出三维分布、链路非确定、应用环境异构、部署区域非结构化等复杂特征，这对定位理论和算法提出了更高的挑战。由于现有定位技术的研究大部分是假设理想化的通信模型和网络部署环境，而忽略了实际应用时各种复杂因素对定位算法的影响，从而导致定位误差控制难、定位误差大、部署成本高及可扩展性差等问题，使之无法直接用于实际系统。本文在对现有定位技术进行研究分析的基础上，针对已有工作的不足，围绕复杂环境下的定位关键技术展开研究，通过系统地研究，提出更加接近实际应用环境的高精度、低开销、高能效、低成本的分布式定位方案，以满足复杂环境下的实际应用。主要研究内容和创新点包括：1．基于泛洪边界控制误差校正的定位策略针对将基于传统的定位方法直接扩展到复杂环境中时存在“模型失用”、误差控制困难等问题，设计了一个新的基于泛洪边界控制误差校正的定位策略3DPHDV-Hop。通过投影平面选取和目标区域划分，进行泛洪边界控制，采用最小均方差准则及归一化加权方法来进行跳距误差校正，从而实现高精度、低误差的细粒度定位。实验结果表明，与APIS、3D DV-Hop和3D-MDS等经典策略相比，3DPHDV-Hop策略的定位率分别提高了45.2%、35.4%和32.3%。2．基于空间几何分割的凹/凸分解定位机制针对存在障碍物的复杂3D凹/凸不平表面网络拓扑几何特性所带来定位误差大且能耗高的问题，提出了一种新的基于空间几何分割的凹/凸分解定位机制3D-CCD。按照节点的海拔高度先逻辑水平分层，再在层内按照凹/凸分解的方法将网络合理的划分为多个子区域，用于降低因地形分布凹/凸而造成的定位误差和能耗，定位过程中只需依靠被目前大多数定位机制所利用的RSSI来实现高精度、低开销的细粒度定位。实验结果表明，3D-CCD与SV和COLA相比，定位误差分别降低了10%、17%，计算成本分别降低了31%、22%。3．基于网络拓扑分形的三角划分定位算法针对存在障碍物的复杂3D凹/凸不平表面网络中锚节点部署难且成本高的问题，提出了一种新的基于网络拓扑分形的三角划分定位算法3DT-ST。该算法仅利用网络连通特性和特殊节点（特殊渡口节点），进行三角划分和建模，在每一个三角区域上采用MDS-MAP方法建立起局部相对位置地图，通过合并每个三角子区域，建立起整个网络全局的位置地图。实验结果表明，3DT-ST算法与目前使用的SV方法相比，定位误差降低了85.3%，且定位过程无需锚节点和迭代，仅通过节点间的连通性进行定位，这提高了定位的精度、降低了计算开销的同时节省了部署成本。4．基于图论的移动锚节点定位路径规划针对存在障碍物的复杂3D凹/凸不平表面网络中静止锚节点可扩展性差且能耗高的问题，提出一种新的基于图论的移动锚节点定位路径规划3DT-PP，它利用移动锚节点的路径规划在复杂的3D地形中实施定位节点，规划锚节点的避障路径，用最优路径遍历整个监测区域，达到快速定位及全网定位的目的。实验结果表明，3DT-PP方法比MDS-MAP方法、Landscape-3D方法位置误差分别降低了91％、8.7％，计算开销分别降低了75％、1.3％。该方法在最短时间内高性能地定位所有未知节点，减少了网络中的孤立节点和锚节点的数量，降低了障碍物对定位精度的影响，有效降低网络构建成本。综上，本文围绕复杂环境下的无线传感器网络定位关键技术展开了研究，提出了四个解决方案，并通过大量的实验进行验证。实验结果表明，提出的四个方案满足了精度高、开销低、效率高、成本低的实际需求。实验结果不仅具有重要的研究意义，更具有广泛的实际应用价值。更多还原

【Abstract】 Localization is one of the most fundamental techniques and services in WirelessSensor Networks (WSNs). Many applications in WSNs, such as information collection,targets tracking, information management, environment monitoring, geographic routingand so on, all require accurate position information of sensors. Under complexenvironment, networks are characterized by3D distribution, link indeterminacy,environment heterogeneity and unstructured deployment properties. These factors bringgreater challenges to localization theories and algorithms. Currently, most localizationstudies assume that the communication model and network deployment environment areidealized, and they ignore the effects of various factors on the localization algorithms.As a result, these models and algorithms will become inefficient or useless in practicalapplications. To overcome the defects of existing work and to relax the challenge ofcomplex environment localization, four distributed localization solutions whichsatisfy the practical requirements of high precision, low overhead, high efficiency andlow cost in complex environment, are proposed in this dissertation. The maincontributions of this dissertation are listed as follows:1. Flood boundary control based error-correction localization strategyTo solve the problem of model impropriety, large accumulated errors and matrixmulticollinearity form directly extending the traditional hop-based localization schemeto3D complex concave/convex surface with obstacles, a new hop-basederror-correction localization scheme called3DPHDV-Hop is proposed.3DPHDV-Hopuses the scheme of projection plane selection, hop control in N subareas from interestarea, hop distance correction and measurement correction to achieve fine-grainedlocalization with high accuracy and low error. Experiment results show that3DPHDV-Hop has higher localization rate than APIS,3D DV-Hop and3D-MDS, withthe ratio of45.2%,35.4%and32.3%, respectively.2. Concave/convex decomposition localization mechanism based on spatialgeometric segmentation.To solve the problem of large localization error and energy consumption on thetopology of3D complex concave/convex surface with obstacles, a new concave/convexdecomposition localization mechanism called3D-CCD, which is based on spatial geometric segmentation, is proposed.3D-CCD utilizes the logical delamination andconcave/convex decomposition to divide the complex, irregular area to subareasaccording to the concave/convex topology. It only utilizes RSSI used by mostlocalization schemes to reach high accuracy, low consumption fine-grained localization.Experiments show that, compared with SV and COLA,3D-CCD reduces thelocalization error by10%and17%respectively, and reduces the energy consumption by31%and22%respectively.3D-CCD offers a brand new research method on3Dcomplex concave/convex surface localization.3. Triangulation based localization algorithm based on fractal networktopologyIn order to solve the problem that it is expensive and difficult to deploy anchornodes on3D complex concave/convex surface, a new triangulation localizationalgorithm (called3DT-ST) which is based anchor free and applies to complex3Dterrainis presented.3DT-ST only utilizes network connection and special nodes (specialport nodes, SPN) to triangulate and model. It establishes local relative maps in everytriangle areas, then combine triangle together to get global map of the network.Experiments show that when comparing with SV,3DT-ST reduces the localization errorby85.3%, and localization is an iteration-free process with only the information ofnetwork connection. It improves the localization accuracy, lowers the localization error,andsaves the cost of network deploying as well. It provides a new method on energysaving localization research over complex networks.4. Mobile anchor sensor routing planning of localization based on graphtheory.To address the problem that fixed anchor sensors has the disadvantage of lowexpansibility and high cost in3D concave/convex uneven surface networks, a newmobile anchor sensor routing planning of localization3DT-PP is proposed. It leveragesthe routing planning of mobile anchor sensor, locating sensors in complex3D terrain,planning the accessible route and traversing the entire monitored area with optimal routeto quickly locate the sensors in entire network. Experiments show that compared withMDS-MAP and Landscape-3D, the proposed method lowers the localization error by91%and8.7%respectively and reduces the computation overheads by75%and1.3%respectively. It can localize all nodes in a very short time with high performance, reducethe number of anchor nodes and isolated nodes, relieve the impact of obstacles and lower the cost of network establishment.In summary, this dissertation takes efforts on WSNs localization in complexenvironments. A series of experiments show that the proposed four solutions satisfy thepractical requirements of high precision, low overhead, high efficiency and low cost.Therefore they not only have important theoretical value, but also have extensiveapplication potential.更多还原