【作者】 周颖；

【导师】 肖德宝；

【作者基本信息】 武汉理工大学 ， 信息与通信工程， 2008， 博士

【摘要】 无线传感器网络(Wireless Sensor Network)是随机分布的传感器通过分布式协议自组织方式形成的网络,它作为一个全新的研究领域,在基础理论和工程技术方面向科技工作者提出了大量的挑战性研究课题,其中一个重要研究课题就是无线传感器网络的网络管理。无线传感器网网络管理是保障无线传感器网络稳定、可靠、高效运行的重要条件。基于此,本文深入讨论研究了现有无线传感器网络管理技术,并结合无线传感器网络的特点和实际需求,在对基于策略的无线传感器网络管理框架、策略发布架构的分类、无线传感器网络的分簇管理机制几方面进行研究设计及实现的基础上,提出了一种基于策略和移动代理的无线传感器网络管理体系结构。本文主要研究工作如下:(1)总结了现有无线传感器网络管理的需求和遇到的主要挑战,并将概念上传统的集中式的PBNM(Policy-based Network Management)思想应用到分布式分散式的无线传感器网络中去。在此基础上,为无线传感器网络提出了一种基于策略的网络管理框架,同时将此框架分解成为几个重要的组成部分分别进行了描述和分析。该网络管理框架能较好的体现无线传感器网络管理系统所期望的理想管理需求和关键管理特征,本文着重研究了其中用于无线传感器网络环境下的一些关键组成部分。(2)对基于策略的网络管理系统来说,其策略发布架构的选择是影响系统性能好坏的关键因素之一。在对策略发布架构相关的一组特性进行系统分析的基础上,提出了一种适用于低带宽的无线传感器网络管理的混合型策略发布架构。其基本方法是将策略发布架构按各个特性的不同进行分类设计,通过对各类架构进行详细研究,阐明了各种策略发布架构的相应操作特性和性能;分析和比较了在各种可能的网络环境下,不同的策略发布架构配置的优势和不足。最后通过仿真实验得出混合型策略发布架构对于低带宽的无线传感器网络来说是一个比较好的选择,并用多跳的策略发布架构的仿真结果阐明了无线传感器网络管理中保持最小跳数分簇管理机制的必要性。(3)分簇机制是无线传感器网络自组织管理框架的关键机制之一,在研究分析了基于能量考虑的传感器网络分簇算法的基础上,从管理的角度提出并实现了一种随机分布式的无线传感器网络分簇管理机制,并将策略用到能量受限的分簇的无线传感器环境中。其基本思想是限制每一个策略服务器的服务覆盖范围,即限制策略服务器与客户端之间的无线跳数,大大地提高分簇的灵活性以及网络的管理效率和自组织管理能力。同时针对网络的动态特性,并将能量的使用扩展到整个网络,提出了基于服务自适应性和资源定位的簇头复选方法,以实现簇结构的重复选择和调整,用于增强上述分簇算法,及增加基于策略服务的可用性。(4)在比较了现有无线传感器网络管理体系的优劣,研究剖析了移动代理技术在网络管理中应用的优越性的基础上,针对无线传感器网络节点能量有限、网络流量不均衡等网络管理的特殊性,将策略管理技术和移动代理技术相结合,给出了策略代理(PA)的概念,并提出了一种基于策略和移动代理的无线传感器网络管理体系结构(PABNM)。该网络管理体系结构充分融合了策略管理和移动代理的优势,管理任务采用策略规则定义的形式,使用移动代理作为运载工具来执行,结合静态代理和移动代理合作的方法完成本地策略决策,以协作的方式对传感器的节点进行管理,并结合本文的分簇算法和具体的路由协议在该体系结构中的应用和性能分析,证明了该网络管理体系结构具有策略和代理的灵活性以及管理的合理有效性的特点。更多还原

【Abstract】 Wireless Sensor Network (WSN) is composed of a large number of sensor nodes distributed randomly by self-organization distributed protocols. As a completely new research field, WSN provides the scientific researcher with amount of challengeble research subjects in terms of basic theory and engineering technology, one of which is the network management of wireless sensor network. The network management of WSN is the important requisites to guarantee the stability, reliability and high efficiency in WSN. Based on the above mentioned, deeply discussion and research on the existing technology of network management of WSN are given in this paper, combined with the characteristics and practical requirements of the wireless sensor network, with the research in such aspects as the policy-based network management framework of WSN, the classification of policy distribution architecture and the clustering management mechanism of WSN, we finally propose a new architecture of policy and mobile agent-based network management architecture for WSN. The main research work of this paper are as follows:(1) The requirements and main challenges of the existing WSN management are summarized firstly, and then the conceptually traditional and centralized PBNM (Policy-based Network Management) idea is applied to the distributed and decentralized WSN. Based on that, we propose a policy-based network management framework for WSN, and divide the framework into several parts and make description and analysis respectively. This framework can better embody the ideal requirements and the key characteristic of management that are expected for the network management system of WSN. In this paper, we mainly focus on the several critical parts used under the environment of WSN.(2) As for policy-based network management system, the choice of its policy distribution architecture is one of the key factors that affect the system operation. After the systematical analysis on a group of specific properties related to the policy distribution architecture, we present a hybrid policy distribution architecture that is suitable for the low wideband WSN network management. The basic method is to design and classify the policy distribution architectures according to the different characteristics, and clarify the corresponding operational features of each kind of the policy distribution architectures; and make analysis on and comparison between the advantages and disadvantages of different configuration of the policy distribution architecture under the every possible network environment. Finally, the simulation results show that the hybrid policy distribution architecture is a comparative better choice for the low wideband WSN network management, and the comparison of multi-hop policy distribution architectures indicate the necessarity of smallest hops in the clustering mechanism of WSN management.(3) The clustering mechanism is one of the key mechanisms in the self-organization management framework of WSN. Based on the research and analysis of energy-based clustering algorithms in WSN, we propose a distributed clustering management mechanism in WSN from the management point of view, and apply the policy technology to the energy-limited clustered WSN environment. The basic idea of this clustering mechanism is to limit the service coverage of each policy server, that is, to limit the number of wireless hops between the policy server and the client, so as to greatly improve the clustering flexibility and the management efficiency as well as the ability of self-organization management. At the same time, aiming at the dynamic of WSN and in order to extend the energy usage to the whole network, we also put forward a clusterhead re-selection method based on service adaptivity and resource discovery to realize the re-selection and adjustment of cluster configuration, which can be used to enhance the performance of this algorithm and increase the usability of policy service.(4) After a comparison of existing network management models for WSN, then the advantages of mobile agent technology in the applications of network management is analyzed. On the basis of features of WSN such as limited power and imbalances traffic among sensor nodes, we combine the technologies of policy management and mobile agent and present the concept of policy agent, and then propose a new architecture of policy and agent-based network management PABNM (Policy Agent-Based Network Management) for WSN. This network management architecture of WSN fully combine the advantages of the technologies of policy management and mobile agent, in which the management tasks are defined by the form of policy rules and are enforced by the mobile agent as the carrier tool, to make local policy decisions and network management of WSN with cooperation of static agent and mobile agent. The PABNM architecture is proved to have the characteristics of flexibility of policy agent and validity of management for WSN by describing the applications and analysis of our clustering algorithm and objective parameters selection of routing method in the architecture.更多还原