WSN中基于Mobile Agent的数据融合算法研究

发布时间：2018-04-22 03:15

本文选题：无线传感器网络 + 数据融合　；参考：《兰州交通大学》2017年硕士论文

【摘要】：随着计算机技术、无线通信技术、传感器技术以及嵌入式系统技术的跨越式发展,一些生产成本低、功耗低的微型传感器应运而生;这些传感器主要拥有对监测目标的感知能力、对信息数据的计算能力以及在传感器设备之间的无线通信能力。无线传感器网络正是由这类微型传感器节点以随机或者人工的方式部署在指定的监测区域,使用无线通信方式形成一个多跳自组织的分布式网络。无线传感器网络具有大规模性、拓扑结构的动态性、自组织性以及高度的应用相关性,使得它在传统的信息收集方式上带来了一场根本性的变革。但是,传感器节点在能量资源、设备存储和计算能力等方面存在的局限性,尤其是来自能量资源约束,极大地阻碍了无线传感网络面向具有更大规模性、多样性等应用领域的发展。在初始阶段,本文对基于传统C/S计算模型的传感器网络随着网络规模增大时所产生的一系列问题进行分析和讨论;随后,以基于Mobile Agent计算模型的无线传感器网络为研究方向展开理论分析和技术研究。本文的主要研究内容如下:首先,在基于Mobile Agent计算模型中,由于Mobile Agent对网络传感器节点的访问路径会直接影响网络的能量开销和数据融合的性能,所以本文对Mobile Agent访问路径的规划方法进行研究和改进。在此基础之上分别提出了两种路径规划方法:(1)WSN中基于数据规模的Mobile Agent路径规划方法,该算法主要根据网络各个簇的规模来确定网络所需Mobile Agent的数量,进而明确Mobile Agent对网络节点的访问规则。(2)WSN中基于迭代局部搜索的Mobile Agent路径规划方法,此算法主要针对多Agent的路径规划方法存在的一些问题,使用迭代局部搜索理论进行优化和改良。其次,Mobile Agent计算模型所使用的数据融合算法的效率对最终融合结果起着显著作用。本文对常见的数据融合算法进行分析和研究,在此基础之上,提出一种基于数据抽样的Mobile Agent数据融合算法;针对一些融合方法在融合效率以及精确性存在的缺陷,此算法主要从减少网络数据的传输量、提高数据融合的精确性方面进行完善。最后,通过在TinyOS仿真平台TOSSIM下,对本文所提算法进行仿真实验与测试;实验结果表明:本文算法在无线传感器网络的高效节能、均衡网络负载、降低网络延迟和网络系统开销、延长网络生命周期方面具有一定的合理性及有效性。
[Abstract]:With the development of computer technology, wireless communication technology, sensor technology and embedded system technology, some micro sensors with low production cost and low power consumption come into being. These sensors have the ability to perceive the target, compute the information data and wireless communication between sensor devices. Wireless sensor networks (WSN) are deployed randomly or manually in the designated monitoring area to form a multi-hop self-organized distributed network using wireless communication. Wireless sensor network (WSN) has the characteristics of large scale, dynamic topology, self-organization and high application correlation, which makes it bring a fundamental change in the traditional information collection method. However, the limitations of sensor nodes in energy resources, device storage and computing capabilities, especially due to energy resource constraints, greatly hinder the wireless sensor network facing to a larger scale. Development of applications such as diversity. In the initial stage, this paper analyzes and discusses a series of problems in sensor networks based on the traditional C / S computing model when the network size increases. The theoretical analysis and technical research of wireless sensor networks based on Mobile Agent computing model are carried out. The main contents of this paper are as follows: firstly, in the Mobile Agent computing model, the access path of Mobile Agent to sensor nodes directly affects the energy cost and data fusion performance of the network. Therefore, this paper studies and improves the planning method of Mobile Agent access path. On this basis, two kinds of path planning methods, Mobile Agent path planning method based on data scale, are proposed respectively. This algorithm mainly determines the number of Mobile Agent required by the network according to the size of each cluster in the network. Furthermore, it is clear that the Mobile Agent path planning method based on iterative local search in Mobile Agent's access rule to network nodes. This algorithm is mainly aimed at some problems existing in the path planning method of multiple Agent. The iterative local search theory is used for optimization and improvement. Secondly, the efficiency of the data fusion algorithm used in Mobile Agent model plays a significant role in the final fusion results. Based on the analysis and research of common data fusion algorithms, this paper proposes a Mobile Agent data fusion algorithm based on data sampling, aiming at the shortcomings of some fusion methods in fusion efficiency and accuracy. This algorithm is mainly improved by reducing the amount of network data transmission and improving the accuracy of data fusion. Finally, through the simulation experiments and tests on the TinyOS simulation platform TOSSIM, the experimental results show that the proposed algorithm can efficiently save energy, balance network load, reduce network delay and network system overhead in wireless sensor networks. It is reasonable and effective to extend the network life cycle.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212.9;TN929.5

【参考文献】