【作者】 陆璐；

【导师】 杨云；

【作者基本信息】 扬州大学 ， 计算机应用技术， 2008， 硕士

【摘要】 网络在运行过程中不可避免地会发生故障。受制造工艺的限制、使用寿命以及工作条件的影响,有多种原因导致网络故障,如网络线路破坏、终端系统或路由设备故障,甚至人为的攻击等。无论是元器件还是电路系统的故障,都需要引起网络管理人员的重视。因此,网络故障诊断技术应运而生,并随着无线通信网络的发展,人们对故障诊断技术提出了新的技术要求。近年来,随着研究领域的广泛、技术的深入,无线传感器网络具有良好的应用前景。无线传感器网络在实际应用过程中十分容易受到故障的干扰,从而严重影响其工作的效率和质量。传感器节点是随机部署在各种特殊区域的,人类很多情况下无法亲自对其设备性能进行维护。因此,针对传感器网络故障的诊断技术有别与传统有线网络。传感器设备的物理特性决定了它对能量依赖性,节约能量的故障诊断技术不仅要保证网络的运行效率,而且能维持网络的长期运行。能量效率是无线传感器网络故障诊断技术设计的首要考虑因素。本文根据无线传感器网络在采集信息、处理信息、传输数据这三个主要步骤中产生的数据出错故障,尤其是数据不一致性的问题。从信息处理和数据传输两个方面,对数据质量进行监督,据此设计了两种故障诊断方法,在诊断故障的同时注重对能量的合理利用。文章主要研究内容及创新点如下:(1)针对无线传感器网络数据传输过程中产生的非一致性问题,提出了一种面向无线传感器网络的非一致性故障检测机制(Inconsistent Failures Detection Mechanism: IFDM)。IFDM通过采用节点不相交多重路径,比较数据包内容,判断故障路径,并实施故障节点的定位。IFDM算法能有效的检测出非一致性故障路径及故障节点,具有检测速度快、功耗低、时延小等特性。(2)针对无线传感器网络中节点采集环境特征数据,并转化为二进制决策值过程中出现的异常问题,提出了一种基于测量值评估机制的WSN分布式故障检测(Mechanism of Distributed Failures Detection by Assess Measurements: DFD-AM)。DFD-AM利用相邻节点数据值的空间相关性,计算节点的评估值,并通过评估值与测量值的比较,判断节点是否出错。同时采用故障识别机制对出错节点标记从而排除故障节点。DFD-AM算法能有效地纠正错误测量数据,检测故障节点,具有检测效率高、误检测率低、功耗低等特性。网络故障诊断是一个极其复杂的问题,是一门多学科融合的技术。故障诊断的切入点很多。目前,还没有一种能解决所有网络问题的故障诊断方法。一般只是就某种或某些故障类型而提出的诊断方法。要想完全解决各种故障,需要从多个科研领域,多方位,多角度全面地考虑并加以研究,同时考虑功耗、时效等多个因素,未来面对的困难还有很多。更多还原

【Abstract】 Faults are inevitable in the running process of network. Due to the restricts of fabricated technical, the using time and the working condition, network faults come from many reasons, such as the destroyed internet wire, the fault of terminal systems or route equipments, and even some attacks by human doing. No matter which the fault comes from parts of apparatus or the circle system, it needs to be paid attention to by the networks administrator. Therefore, the technologies of fault diagnoses emerge as the times require. With the development of wireless communication network, new technique requirements are needed for the fault diagnosis.Nowadays, with the broad and deep research of domains and techniques, WSN (wireless sensor network) has a bright future in applications. However, it is very easily interfered by faults in the process of practice applications, and then the working efficiency and quality can be affected. Sensor nodes are deployed stochastically in many especial areas, in the places which human can’t reach to doing some maintenance to the apparatus’performance. Thus, the diagnoses of WSN are different from traditional wired networks. Because of the physical traits, the sensor equipments are strongly dependent on its energy resource. A fault diagnosis method can not only assure the efficiency of network but also survives the long running of network. Energy using efficiency is the chiefly considered fact of fault diagnosis in WSN.In this article, according to faulty of data errors, particularly to the data inconsistence in the three importance steps of information collect, information process and data transmit in WSN. Then we pay a close attention to the two steps of information process and data transmit, and supervise on data quality and then proposed two fault diagnosis methods. Meanwhile we make effort on the reasonable utilize of energy. The primary works and innovations of this paper represent as follows:(1) For the problem of data inconsistence in transmission of Wireless sensor networks, we propose a mechanism of IFDM (Inconsistent Failures Detection Mechanism). It compares the packet data, decides the faulty path and locates the faulty node by adopting node-disjoint multi-path routing. Simulation shows that the mechanism of IFDM can detect the inconsistent faulty paths and nodes efficiently and it has high detecting speed, low power cost and less time delay.(2) According to the abnormity which arises in the process of transforming environments Ground-true value into binary determinant value in wireless sensor networks, we propose a Mechanism of DFD-AM (Distributed Failures Detection by Assess Measurements) in WSN. It takes advantage of spatial correlated data from neighbor sensor nodes to calculate the assessment of nodes. Then determines nodes wrong or not by comparing the assessment with the measurement. The mechanism identifies the faulty nodes, and then removes them. Simulation shows that DFD-AM can correct the error measurements and detect faulty nodes efficiently, and it also has characteristics of high detection success, low wrong detection probability and low power consumption.Network faults diagnosis is a complicated problem and also a technology which combined of multi-discipline. At the present time, there not exists a method which can figure out all of the faults in the network. Most of the diagnosis methods can only deal with a certain fault or some faults. We have to integrate many domain sciences and consider full-scales from various orientations and angles to solve all the faults completely. Meanwhile, we should consider the factors of power cost, time waste, etc. We will face a large number of difficulties in the future.更多还原