【作者】 许磊；

【导师】 石为人；

【作者基本信息】 重庆大学 ， 控制理论与控制工程， 2009， 博士

【摘要】 无线传感器网络涉及现代传感器、微电子、无线通信、嵌入式计算、分布式信息处理等多个学科领域,以数据为中心,具有自组织、自调节特点,是构建普适计算新环境、实现普适控制的新兴交叉研究领域,可广泛用于军事、工农业、医疗保健、空间探索及其他各种商业应用。节点位置信息对于无线传感器网络至关重要,不仅是监测应用进一步采取措施和做出决策的基础,而且还是目标跟踪、辅助路由、网络管理等其他关键技术研究/实现的基础。本文面向控制学科的前沿研究方向——无线传感器网络所涉及的理论与技术,针对传统典型的Anchor-Dependent定位技术定位精度受锚节点密度影响、在危险/恶劣环境中存在应用局限性等特点,基于几何方法、重复优化技术、图论、粒子群优化等计算方法,从相对定位与基于移动锚节点定位两方面开展Anchor-Independent定位算法与应用技术研究。作为一项具有创新性意义的工作,论文在研究方法与思路上力求有所突破,主要研究成果包括如下四个方面:①基于OK分簇算法,提出了面向相对定位的K跳分簇算法(EOK)。在相对定位算法中,对网络分簇是降低测距累积误差、提高定位收敛速度、增强定位扩展性的有效方法。EOK通过设计综合节点连通度和剩余能量制定加权随机选择簇头选择策略、基于功率控制设计邻居簇头合并机制、节点簇优化措施,满足相对定位应用对簇头数量、节点簇重叠度、节点簇坐标转换等方面的独有要求。相比OK算法,EOK算法提高了簇头选择的针对性,降低了簇头数量,可有效控制簇重叠度,确保节点簇坐标转换率100%。②基于EOK算法,提出了基于分簇的分步求精相对定位算法(SRRLC)。相对定位算法无需锚节点参与,定位精度不受锚节点密度影响,是典型的Anchor-Independent定位算法。SRRLC分为初定位、定位优化、局部坐标系转换三个阶段:在初定位阶段,提出基于RSSI测距的加权质心定位算法(WCL-RSSI)进行节点初定位,基于点着色理论的频点分配算法,消除隐藏终端对簇边界节点测距的影响,并通过参考节点精选机制、参考节点组合精选策略等策略,降低了测距误差对算法的影响,提高定位初值精度。迭代优化阶段中,基于SMACOF技术对节点位置迭代优化,综合测距距离与定位误差等因素构建优化权值函数,加快迭代收敛速度,提高算法定位精度。相比典型的相对定位算法,SRRLC对测距误差具有较高的适应性,具有较高的定位精度,在算法收敛速度、通信开销、定位覆盖率等指标上具有优势,更适合节点大规模布设的无线传感器网络。③提出了基于移动锚节点的粒子群优化定位算法(PLBMA)。在很多危险/恶劣环境中,无线信道损耗模型参数未知,无法直接基于RSSI测距定位。因此,PLBMA将无线信道损耗模型未知环境中的节点定位问题抽象为非线性约束优化问题,构建加权定位目标函数,并采用粒子群优化策略进行节点优化定位。为减小网络定位成本,降低定位精度对锚节点数量的依赖,利用移动锚节点代替传统典型算法中的静态锚节点,将网络覆盖区域网格划分,采用移动锚节点采用扫描线最短完全遍历路径,确保虚拟锚节点完全覆盖网络区域。PLBMA算法简单易实现,可调参数少,通过多次迭代寻优,可有效提高算法定位精度、增强对环境噪声干扰的适应能力,拓展了Anchor-Independent定位技术应用适应性。④面向典型危险/恶劣环境监测应用的共性特点,构建Anchor-Independent定位应用原型系统——基于无线传感器网络的多模态监测原型系统。针对多移动Sink网络拓扑结构,设计了面向混合网络拓扑结构的路由协议;集成PLBMA算法与Pioneer3智能移动平台,设计了基于PLBMA算法的事件定位策略;实现了便携式指挥、事件定位、环境信息感知、移动用户生理信息监护、多模式交互(语音、图像、文字等方式)等5大功能。该系统通过集成802.11g与802.15.4通信协议、构建多移动Sink网络拓扑结构、结合智能移动平台,丰富和拓展了传统的静态无线传感器网络概念,为无线传感器网络技术的进一步实用化做出了探索性尝试。本文深入的开展了无线传感器网络Anchor-Independent定位理论与技术研究,研究提出了三个相关算法,并构建Anchor-Independent定位应用原型系统,仿真、实验结果充分证明了各算法、系统的有效性。进一步研究将从提升性能与增强实用性两方面优化提出的算法;并面向实际应用需求,完善原型系统功能,提升系统性能。更多还原

【Abstract】 Integrated modern sensor, micro-electronics, wireless communication, embedded computing, distributed information processing and other technologies, wireless sensor network (WSN) is an emerging research direction. Compared with the traditional networks, WSN is a data-centric wireless network with self-organizing and self-regulating characteristics, and can be widely used in military, health care, space exploration and many other kinds of commercial applications. Location information of each sensor node is essential to WSN. It is not only the basic of taking further measures and making decision-making, but also the research foundation of other key technologies (for example target tracking, routing supporting and network management) in WSN.Most of typical localization algorithms belong to Anchor-Dependent localization, whose localization accuracy decrease quickly once the density of anchor nodes declined significantly, and is not suitable for some harsh or dangerous application environments. Facing on the limitations of Anchor-Dependent localization, based on geometry, probability theory, iterative memorization, graph theory, particle swarm optimization (PSO) and other calculating methods, Anchor-Independent localization research is carried out from two aspects in this thesis: the relative localization and the mobile anchor node based localization. The important research results are as follows:①Based on OK clustering algorithm, a K-hop clustering algorithm for relative localization- Enhanced Overlapped K-hop (EOK) is proposed. In relative localization, accumulating ranging error is one of the main factors affecting localization performance, and can be reduced effectively through clustering. In EOK, some strategy (cluster header selection based on both connectivity and remaining energy of each node, neighbor cluster header mergence based on power control, cluster optimization) are help to satisfy the demands of relative localization on the number of cluster header, overlapping degree of cluster and coordinate transformation rate between neighbor clusters. Compared with OK, EOK not only reduces the number of clusters header and make better uniformity of distribution of clusters, but also can control overlapping degree of cluster and maintain coordinate transformation rate between neighbor clusters at 100%.②A stepwise refinement relative localization algorithm based on clustering (SRRLC) is put forward. Relative localization algorithm is a typical type of Anchor-Independent algorithms. It needs none anchor nodes, and localization accuracy is not affected by the density of anchor nodes. SRRLC is divided into three stages: initial local localization, localization optimization and local coordinate system transformation. In initial local localization stage, a weighted centroid localization algorithm based on RSSI (WCL-RSSI) is put forwarded. In WCL-RSSI, a frequency allocation based on vertex coloring algorithm is proposed to eliminate the influence of hidden terminal phenomenon on ranging error of cluster border nodes. Besides, reference nodes refining strategy and reference node sets refining strategy are used to reduce the influence of ranging error on WCL-RSSI, and enhance the localization accuracy. In localization optimization stage, a SMACOF based strategy is used to optimize location, and a weight function used in this strategy, general considering both distance and localization errors factors, is put forward. This weight function can speed up the convergence rate of this strategy and better the localization accuracy. Compared with typical relative localization algorithm, SRRLC is not only robust to ranging error and achieve higher localization accuracy, but also enjoys advantages on convergence rate, communication overhead and localization coverage.③A PSO localization algorithm based on mobile anchor is put forward. In many harsh or dangerous application environments, parameters in path loss model of wireless channel are unknown, and neighbor nodes are unable to range using RSSI. So localization in this environment which path loss model of wireless channel is unknown, is abstracted into non-linear constraint optimization in PLBMA.A weighted localization target function is designed and the location of each node can be estimate based on PSO. In order to reduce localization cost and the dependence of localization accuracy on the density of anchor nodes, a mobile anchor node is used in PLBMA instead of static anchor nodes in Anchor-Dependent localization algorithm. Network interest area is divided into grids, and the mobile anchor node traversals network interest area based on shortest path traversal strategy, which insure network interest area can be complete covered with virtual anchor nodes. The principle of PLBMA is simple and easy to implement, and there is only a few adjustable parameters. After a few times of iteration, localization accuracy can be improved obviously and is robust to the environment noise interference. PLBMA expands the application adjustability of Anchor-Independent localization technology.④Considering military defense, disaster monitoring and rescue and other harsh or dangerous application environments, an Anchor-Independent localization application prototype system - multi-modal monitoring prototype system based on WSN is built. Facing on mixed network topology with multiple mobile Sinks, a new routing protocol is designed and implemented in this system. Besides that, integrated Pioneer3 intelligent mobile platform and PLBMA, a event localization strategy is put forward and implemented. The conception of conventional static WSN has been enriched and expended because of some new elements integrated in this system: 1) both 802.11g and 802.15.4 wireless communication protocol are coexisted, 2) mixed network topology with multiple mobile Sinks, 3) intelligent mobile platform is used.An in-depth study of Anchor-Independent localization theory and application technology is carried out in this thesis. Three relative algorithms have been studied and put forward, and an Anchor-Independent localization application prototype system is built. The simulation and experiment results full proof the validity of each algorithm and system. In near future, some research about each algorithm proposed will be carried out from two aspects: improve the performance and enhance the practicability; besides some application research on this prototype system will also be carried out.更多还原