【作者】 李洪兵；

【导师】 熊庆宇；

【作者基本信息】 重庆大学 ， 控制理论与控制工程， 2014， 博士

【摘要】 无线传感器网络(Wireless Sensor Networks，WSNs)是由部署在检测区域内大量传感器节点以无线通信方式组成的一个多跳自组织网络系统，实时地监测、感知和采集监测对象的信息。无线传感器网络是物联网推广的重要载体，未来延伸Internet覆盖范围以及支持普适计算的关键技术。无线传感器网络迅速发展，促进了物联网实现社会生产生活中信息感知能力、信息互通性和智能决策能力的全面提升，并在诸多领域得到广泛应用，已显示出巨大的应用价值并成为现阶段研究的热点。作为一种新兴的信息获取和处理技术的分布式计算平台，无线传感器网络有着无中心自组网、网络拓扑动态变化、通信与计算能力有限、电源能量受限等自身特性。工作环境存在不可预测性，因振动、电磁、系统噪声和随机噪声等环境因素干扰导致错误感知数据。或因节点的移动超出了通信范围、物体阻挡、信道干扰和数据碰撞堵塞等原因，网络链路质量变差，出现暂时或永久性的故障，导致数据传输的失败，影响无线传感器网络传输可靠性和运行稳定性，对网络的自组织能力、自适应性和鲁棒性带来更大的挑战。高可靠性和稳定性目前仍然是无线传感器网络技术的难点。故障容错策略能提高无线传感器网络运行的鲁棒性和传输可靠性。其本质是在检测网络出现异常或故障时，能及时寻找合理的容错控制方案，自适应地处理多种网络异常现象，继续提供高可信的计算服务。但基于预先构成的网络结构，无法满足无线传感器网络大规模可扩展的需求。且因网络复杂性与故障多样性并受限于实际应用条件，现有容错模型简单对故障进行抽象和特征提取，对新出现的故障特征缺乏自主在线学习能力，严重影响网络故障容错性能。针对新出现的故障特征，在动态自适应构造故障容错结构框架、设计自主的在线容错等方面还有待深入研究。针对无线传感器网络节点或链路故障以及工作环境等因素会影响到网络传输的稳定性和可靠性问题，本文主要在无线传感器网络网络层上，通过采取主动故障预防或网络异常后故障容错的方式提高网络传输可靠性。建立并优化多路径路由算法，或运用网路层多种容错策略，或垮层联合控制优化，或引入仿生免疫机理和现代仿生智能算法开展网络层可靠传输和故障容错研究。具体工作如下：（1）根据蚁群仿生智能算法对构建无线传感器网络容错路由的启示，提出非均匀等级分簇的无线传感器网络故障容错路由算法。根据骨干网特性，建立数学模型和网络拓扑结构，运用改进粒子群算法(IPSO)对网络节点进行非均匀等级静态分簇，构建非均匀等级分簇拓扑结构。引入最优最差蚂蚁系统(BWAS)在相邻等级节点间建立多条传输链路，并根据蚂蚁信息素归一化值作为传输路径的选择概率建立能故障容错的网络路由。同时对容错性与复杂度进行理论分析，对数据包接收率、平均传输时延及能耗均衡等进行仿真分析。（2）针对编码机制影响无线传感器网络传输可靠性问题，提出基于纠删编码的无线传感器网络多路径可靠传输策略。对网络节点进行等级标定，根据蚁群算法确定路径信息素值。根据反映链路质量的最优最差蚂蚁系统的信息素归一化值，在相邻等级节点间建立互不交叉的多路径路由传输容错机制，优化纠删编码并建立基于纠删编码的多路径负载均衡机制，将源数据包经纠删编码的数据片沿多条路径分配和传输，建立数学模型并进行理论分析，并通过仿真测试负载均衡和故障容错性能。（3）针对梯度策略影响网络传输可靠和故障容错问题，提出基于梯度的无线传感器网络多路径可靠传输容错策略。首先进行二次k均值非等概率簇分裂构建非均匀拓扑结构，按质量评价函数计算节点的综合度量信息并建立等高线，建立基于梯度的互不交叉的多路径路由，实施负载均衡的线性纠删编码多路径传输。建立负载均衡机制下的多路径传输数学模型，对数据成功传输率与能效性等网络多项性能指标进行理论分析和仿真评价。（4）根据免疫系统机理对构建无线传感器网络多路径容错路由的启示，提出基于免疫系统机理的无线传感器网络多路径容错路由算法。研究免疫系统机理、人工免疫系统模型与无线传感器网络故障检测容错的属性关联。定义基于免疫分簇与免疫多路径等基本问题，运用免疫系统机理对网络进行分簇以构建紧致性较好的分簇拓扑结构，运用免疫系统机理对初始建立的互不交叉的多条传输路径进行多次变异优化并形成最优传输路径，建立数学模型并对算法性能进行理论分析，对分簇算法的紧致性、免疫路由算法的收敛性和能耗性等进行仿真测试。综上所述，本文围绕提高无线传感器网络的稳定运行和可靠传输的目标，构建网络故障容错的相关理论和方法体系，建立故障仿生智能容错平台，提高无线传感器网络的传输可靠性和运行稳定性，为工业监控、矿井安全监测和农业生物环境保护等对网络可靠性要求较高的无线传感器网络预警系统提供理论与技术支撑。更多还原

【Abstract】 Wireless sensor network (WSNs), composed by a large number of randomlydeployed sensor nodes in the detection area, is a multi-hops self-organization networksystem in wireless communication way. It can real-time monitor, perceive and acquirethe monitored object’s information. Wireless sensor network is an important carrier ofInternet of Things (IOT) and the key technology of future extension of Internetcoverage and pervasive computing. Its rapid development leads to the comprehensivepromotion of Internet of Things in information awareness, information interoperabilityand intelligent decision-making. It shows great widespread application in many fieldsand becomes the hot spot in present study.As a new distributed computing platform of information acquisition and processingtechnology, wireless sensor network has characteristics of no center distributed network,dynamic topology changes, restriction of communication, computing power and energysupply. Additionally, unpredictability exists in the working scenario of network due tothe environmental interference of vibration, electromagnetism, system noise, randomnoise and so on which can lead to incorrect perception packets. Owing to movingbeyond the scope of communication, objects blocking, channel interference and datacollision, it will lead to link quality variation, temporary or permanent network fault orfailure in data transmission. Therefore, the transmission reliability and working stabilitywill be affected in wireless sensor networks. More challenges are brought toself-organization ability, adaptability and robustness of the network. High reliability andstability are still the difficult technical issues in the present study on wireless sensornetworks.Fault tolerant strategy can improve the operation robustness and transmissionreliability of wireless sensor networks. The nature of fault tolerance is to seek thereasonable fault-tolerant control scheme, adaptively process a variety of networkanomalies and continues to provide high credible computing services when somethinggoes wrong with the network system. However, the fixed structure based network cannot meet the large-scale extensible requirements of wireless sensor networks. Inaddition, the network has characteristics of complexity and diversity and is restricted bythe conditions of practical application. Existing fault-tolerant models simply extract fault features and lack of the ability of independent online learning to new failurecharacteristics. This seriously affects the accuracy and robustness of fault tolerance.Aiming at the new failure characteristics, further study should be carried out based onthe framework structure establishment of dynamic-adaptive and online fault-tolerantdesign.Therefore, for the issues that the fault of nodes or links and the working scenariowill affect the transmission stability and reliability in the network layer, This studyadopts the initiative fault prevention strategies before the abnormity and fault tolerantstrategies after the abnormity to improve the network reliability mainly in network layer.These strategies include the optimization of multi-path transmission routing algorithm,joint-control optimization on cross layers, introduction of modern intelligent bionicalgorithms which are based on the mechanism of bionics and immune to study the faulttolerance and reliable transmission in network layer. The details are as follows:(1) According to the inspiration of bionic intelligent algorithm of ant colony to thefault routing establishment in wireless sensor networks, a non-uniform clusteringrouting algorithm is presented with the function of fault tolerance. According to thecharacteristics of vascular network, mathematical model and network topology areestablished. Improved particle swarm optimization (IPSO) is studied and adopted to dothe non-uniform gradient static clustering. Best and worst ant system (BWAS) isintroduced to create fault toerant transmission paths between the neighbor hierarchicalclustering heads. Normalized values of ant pheromone are adopted to be as the pathselecting probability of transmission path. Performance of fault tolerance and algorithmcomplexity are theoretical analyzed. Packet receving rate, average transmission delayand energy consumption balance are simulated.(2) For the issues that the coding mechanism can affect the transmission reliabilityin wireless sensor networks, multi-paths and erasure encoding based reliabletransmission strategy (MPE2S) is presented. Nodes in the network are hierarchicaldetermined and pheromone values are calculated by BWAS. According to thepheromone normalized values by BWAS which reflect the links quality, a transmissionfault-tolerant mechanism of disjoint multi-paths is established among the adjacenthierarchical clustering heads. The algorithm optimizes erasure coding and establisheserasure coding based multi-path load balance mechanism. It allocates the erasure-codedfragments from source packet to multiple paths for transmission. Mathematical model is established to do the theoretical analysis. Simulations are carried out to test performanceof load balance and fault tolerance.(3) For the issues that the gradient can affect the performance of transmissionreliability and fault tolerance, a gradient based multi-path reliable transmission strategywith fault tolerance is presented for wireless sensor networks. Firstly non-uniformclustering topology is established in non-equal clustering probability based on twiceK-means algorithm. Then it calculates the comprehensive measurements (CM) ofclustering heads by the quality evaluation function and establishes the contour lines.Finally gradient based multiple disjoint transmission paths are established by themechanisms of load balance and linear erasure coding. It establishes the transmissionmathematical model to analyze the network performance. Simulation shows that thisstrategy has good performance of fault tolerance. It improves the transmission reliabilityof the network and energy efficiency.(4) For the inspiration of immune system mechanism to improve fault toleranceand transmission reliability for improving the overall performance of wireless sensornetworks, mechanism of immune system based multi-path fault tolerant routingalgorithm is presented. It studies mechanism of the immune system, artificial immunesystem model and properties associated with fault-tolerance in wireless sensor networks.Basic issues are defined such as immune clustering, immune multi-path and so on.Mechanism of immune system is adopted to establish the hierarchical clusteringtopology with optimal performance of clustering compactness. Mechanism of immunesystem is also applied to do the variation and optimization on the initial antibodypopulation, namely the multiple disjoint paths, to establish the final optimaltransmission paths. Mathematical model is established to do the theoretical analysis onthe performance of the algorithm. Through the simulation on the packet receiving rate,accuracy rate and energy efficiency, the algorithm improves the transmission reliabilityandenergy efficiency of the network. It shows superior performance of fault tolerance.In conclusion, this study aims at improving the transmission reliability and runningstability in wireless sensor networks. It establishes theories and methods related withfault tolerance based on the previous studies. Furthermore, bionic intelligentfault-tolerant system is established to improve transmission reliability and stability. Itprovides theoretical and technical support to warning system by wireless sensornetworks in the domains requiring high reliability performance such as industrial monitoring, mine safety monitoring and agricultural biological environmental protection,etc.更多还原