基于无线传感器网络的目标检测与跟踪研究
发布时间：2018-11-24 08:01
【摘要】：随着微机电系统技术、无线通信技术、计算技术与传感器技术的迅猛发展,集信息感知、信息处理与信息传输于一体的多学科交叉技术——无线传感器网络——成为国际上备受关注的前沿热点研究领域。因其独有的特点,无线传感器网络被广泛应用于国防军事、环境监测、智能家居与医疗护理等多个领域,被认为是21世纪最有影响的21项技术之一和改变世界的10大技术之一。 移动目标检测与跟踪作为无线传感器网络的重点应用领域之一,有着广泛的前景,尤其在安全防卫领域发挥着重大作用。本文考虑无线传感器网络大规模和资源受限的特点,针对目标检测与跟踪,研究网络节点部署、节点跟踪调度与信息的实时汇报问题,并通过仿真与实际系统验证算法的有效性。本文的主要贡献有如下几个方面： 1.针对静态传感器网络随机部署后难以有效形成栅栏覆盖的问题,提出了利用移动节点再部署以弥补静态节点间漏洞,从而形成栅栏覆盖。提出了加权栅栏图的概念,将最小代价移动节点问题转化为加权栅栏图的最短路径问题；将最小能耗栅栏覆盖构建问题映射为最小成本二部分配问题；在此基础上,提出了有向栅栏覆盖构建算法以最小的代价部署和调度移动节点与静态节点形成栅栏覆盖。理论分析和仿真结果验证了算法的有效性。 2.从实际角度出发,首次考虑节点位置误差存在对栅栏覆盖的影响。理论分析了节点位置误差如何影响所需的真实移动节点数目,并提出了最优的递进式移动节点部署算法指导节点移动弥补静态节点间漏洞。此外,考虑位置误差的影响,提出了容错加权栅栏图,从图论角度得到确保栅栏覆盖构建所需的最少移动节点数目。实验结果验证了误差的影响以及理论分析和算法的正确性。 3.针对大规模网络中移动目标有效跟踪及位置实时汇报问题,提出了基于簇的目标跟踪算法。提出利用簇头唤醒和休眠节点跟踪目标,并通过簇头构建的骨干网实时汇报目标位置。实验结果表明,基于簇的目标跟踪算法在准确跟踪目标的同时,不仅节省了能耗,也减少了位置汇报的延时。 4.针对基于簇的目标跟踪算法所面临的边界问题,将动态簇思想融入到静态簇网络结构中,提出了基于混合簇的目标跟踪算法。在目标移动过程,不断的利用静态簇与动态簇轮流跟踪目标。实验结果表明,基于混合簇的目标跟踪算法有效解决了基于簇的跟踪算法所面临的边界问题,并在性能方面优于其他经典的目标跟踪算法。 5.针对已有目标跟踪算法缺乏系统验证的问题,在基于簇的目标算法框架下,设计实现了真正的移动目标跟踪系统。此系统包含36个传感器节点,一个汇聚节点与一个基站。多次的静止目标定位实验与移动目标跟踪实验,充分验证了目标跟踪系统的有效性。
[Abstract]:With the rapid development of MEMS technology, wireless communication technology, computing technology and sensor technology, collecting information perception, Wireless sensor network (WSN), a multidisciplinary and integrated information processing and information transmission technology, has become a hot research field in the world. Because of its unique characteristics, wireless sensor network is widely used in many fields, such as national defense, environment monitoring, intelligent home and medical care, etc. It is considered to be one of the 21 most influential technologies in the 21st century and one of the top 10 technologies to change the world. As one of the key application fields of wireless sensor networks, mobile target detection and tracking has a wide range of prospects, especially in the field of security and defense plays an important role. Considering the characteristics of large scale and limited resources in wireless sensor networks, this paper studies the deployment of network nodes, the scheduling of nodes and the real-time reporting of information for target detection and tracking. The effectiveness of the algorithm is verified by simulation and practical system. The main contributions of this paper are as follows: 1. In order to solve the problem that it is difficult for static sensor networks to effectively form fence coverage after random deployment, this paper proposes to use mobile nodes to redeploy to make up for the loopholes between static nodes, so as to form fence coverage. In this paper, the concept of weighted palisade graph is proposed, the minimum cost moving node problem is transformed into the shortest path problem of weighted palisade graph, and the minimum energy consumption fence cover construction problem is mapped to the minimum cost bipartite assignment problem. On this basis, a directed fence coverage construction algorithm is proposed to deploy and schedule moving nodes and static nodes to form fence overlay at the minimum cost. Theoretical analysis and simulation results verify the effectiveness of the algorithm. 2. From the practical point of view, the effect of node position error on fence coverage is considered for the first time. This paper theoretically analyzes how node position error affects the number of real mobile nodes, and proposes an optimal progressive mobile node deployment algorithm to guide node movement to make up for the static node loophole. In addition, considering the influence of position error, a fault-tolerant weighted palisade graph is proposed, and the minimum number of moving nodes needed to ensure the construction of fence coverage is obtained from the point of view of graph theory. The experimental results verify the influence of error and the correctness of theoretical analysis and algorithm. 3. Aiming at the problem of effective tracking and real time reporting of moving targets in large scale networks, a clust-based target tracking algorithm is proposed. A cluster head wake-up and dormant node is proposed to track the target, and a backbone network constructed by cluster head is used to report the target position in real time. The experimental results show that the cluster-based target tracking algorithm not only saves energy consumption but also reduces the time delay of location reporting. 4. Aiming at the boundary problem faced by the cluster-based target tracking algorithm, the idea of dynamic cluster is integrated into the static cluster network structure, and a hybrid cluster-based target tracking algorithm is proposed. In the process of moving the target, the static cluster and the dynamic cluster are used to track the target in turn. Experimental results show that the target tracking algorithm based on mixed clusters can effectively solve the boundary problem faced by the cluster-based tracking algorithm and is superior to other classical target tracking algorithms in performance. 5. In view of the lack of system verification of existing target tracking algorithms, a real moving target tracking system is designed and implemented under the framework of cluster-based target algorithms. The system consists of 36 sensor nodes, a convergent node and a base station. The effectiveness of the target tracking system is fully verified by many static target localization experiments and moving target tracking experiments.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN929.5;TP212.9
[Abstract]:With the rapid development of MEMS technology, wireless communication technology, computing technology and sensor technology, collecting information perception, Wireless sensor network (WSN), a multidisciplinary and integrated information processing and information transmission technology, has become a hot research field in the world. Because of its unique characteristics, wireless sensor network is widely used in many fields, such as national defense, environment monitoring, intelligent home and medical care, etc. It is considered to be one of the 21 most influential technologies in the 21st century and one of the top 10 technologies to change the world. As one of the key application fields of wireless sensor networks, mobile target detection and tracking has a wide range of prospects, especially in the field of security and defense plays an important role. Considering the characteristics of large scale and limited resources in wireless sensor networks, this paper studies the deployment of network nodes, the scheduling of nodes and the real-time reporting of information for target detection and tracking. The effectiveness of the algorithm is verified by simulation and practical system. The main contributions of this paper are as follows: 1. In order to solve the problem that it is difficult for static sensor networks to effectively form fence coverage after random deployment, this paper proposes to use mobile nodes to redeploy to make up for the loopholes between static nodes, so as to form fence coverage. In this paper, the concept of weighted palisade graph is proposed, the minimum cost moving node problem is transformed into the shortest path problem of weighted palisade graph, and the minimum energy consumption fence cover construction problem is mapped to the minimum cost bipartite assignment problem. On this basis, a directed fence coverage construction algorithm is proposed to deploy and schedule moving nodes and static nodes to form fence overlay at the minimum cost. Theoretical analysis and simulation results verify the effectiveness of the algorithm. 2. From the practical point of view, the effect of node position error on fence coverage is considered for the first time. This paper theoretically analyzes how node position error affects the number of real mobile nodes, and proposes an optimal progressive mobile node deployment algorithm to guide node movement to make up for the static node loophole. In addition, considering the influence of position error, a fault-tolerant weighted palisade graph is proposed, and the minimum number of moving nodes needed to ensure the construction of fence coverage is obtained from the point of view of graph theory. The experimental results verify the influence of error and the correctness of theoretical analysis and algorithm. 3. Aiming at the problem of effective tracking and real time reporting of moving targets in large scale networks, a clust-based target tracking algorithm is proposed. A cluster head wake-up and dormant node is proposed to track the target, and a backbone network constructed by cluster head is used to report the target position in real time. The experimental results show that the cluster-based target tracking algorithm not only saves energy consumption but also reduces the time delay of location reporting. 4. Aiming at the boundary problem faced by the cluster-based target tracking algorithm, the idea of dynamic cluster is integrated into the static cluster network structure, and a hybrid cluster-based target tracking algorithm is proposed. In the process of moving the target, the static cluster and the dynamic cluster are used to track the target in turn. Experimental results show that the target tracking algorithm based on mixed clusters can effectively solve the boundary problem faced by the cluster-based tracking algorithm and is superior to other classical target tracking algorithms in performance. 5. In view of the lack of system verification of existing target tracking algorithms, a real moving target tracking system is designed and implemented under the framework of cluster-based target algorithms. The system consists of 36 sensor nodes, a convergent node and a base station. The effectiveness of the target tracking system is fully verified by many static target localization experiments and moving target tracking experiments.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN929.5;TP212.9
【参考文献】
相关期刊论文 前6条
1 孙其博;刘杰;黎，
本文编号：2352897
本文链接：https://www.wllwen.com/kejilunwen/wltx/2352897.html
