| The Wireless Sensor Network (WSN), one of the ten most influential high-tech technologies in the 21 century, is a novel platform to acquire and process information, and is constructed by the convergence of sensor, micro-electro-mechanical system and wireless communication technologies. Being large-scale, self-organized and data- centred, it is widely used in applied fields. Expecially, as a key technology in Internet of Things, it collects and transfers information, and processes relay information. However, as the WSN is usually deployed in harsh environments, the fault rate of the sensor is relatively high; and it is very difficult to have the disabled sensor replaced physically. Besides, the data gained from the sensor node is error-prone and insecure, as the communication network suffers from such adverse factors as interference, attenuation, multi-path and dead-zone. Therefore, the sensor network software and hardware must have strong fault-tolerance to ensure the system's high reliability and robustness.The research content for the data fault-tolerance, the data error caused by different reasons in the WSN, has been investigated in order to execute data fault-tolerant function in the aggregation operation and achieve precise event value, then gives fault judgment and reduces influence from data errors. Consequently, based on existing fault-tolerant protocols and algorithms, we designed novel data fault-tolerant algorithms and models to fulfill the requirement of data fault-tolerance in the WSN.In the first part of this dissertation, the background, application demand and basic concepts of fault-tolerance are introduced, followed by an outline of those important research fields, related projects, and research trends in the WSN. Moreover, the research goal and outline are proposed, and a comprehensive and detailed review on relevant research of data fault-tolerance for the WSN is given.Based on the above discussion, some key issues concerning the building method of a data fault-tolerant model, a self-adaptive algorithm, and the construction of the fault-tolerant transmittal schema based on reliable spanning tree are discussed and researched respectively in this dissertation. Some of the specific tasks are listed below:First, an EFSA is put forward to obtain correct data from the event cluster and to whittle data error, aiming to solve the problem of estimating data error and executing data aggregation in the WSN. Starting from the fact that the sensor networks are based on events, an algorithm is given by using residual node energy and confident rate to form event clusters, thus solving the problem of energy efficiency and the problem that the error node is selected as cluster header. To achieve self-adaptive data fault-tolerance in the event cluster and obtain accurate event value and whittle the influence of fault data, we treat node confident rate as weight and extract appropriate data set by means of k-means algorithm. Meanwhile, the adjustment of the node's confident rate using iterative method ensures correctness of data fault judgment and algorithm convergence.Second, as to the problem of data fault-tolerance and data compensation for multi- events, a fault-tolerant and compensation mechanism of multi-event cluster across over- lapping regions is proposed. The distributed multi-event cluster is first constructed to form different event clusters in case of multiple events. Then, the treatment strategies for the sensor node in the multi-event overlapping region are considered, and the data fault-tolerant compensation mechanism in the event cluster is established; the process- ing scheme for data fault-tolerance in the event cluster is found using different node confident rates. The fault-tolerant strategy in multi-event overlapping regions not only saves node energy, but also reduces the influence of data error on various event clusters.Third, a data fault-tolerant transmission schema based on the reliable spanning tree (RST for short) is put forward to improve its data reliability on the transmission path in the WSN. In this dissertation, the generative process of the tree structure is improved according to the characteristics of space correlation on transmission path. Firstly, we construct a net structure with increasing dimensionality, and define link reliability using adjacent node confident rate, signal interference ratio and etc. Data transmission by selecting the high reliability link ensures transmittal dependability, and then structures a reliable spanning tree with the Sink node as its root. At the same time, the provision of measures of adjusting the node confident rate and ways of confirming dynamically the reliable link reach the expected purpose of achieving reliable transmission from the source nodes to the Sink node and enhancing fault-tolerance in data transmission.Finally, starting from the demand of adaptive data fault-tolerance in the WSN, a data fault-tolerant model based on the hybrid structure is presented. Based on existing data aggregation models in the WSN and combining data aggregation with data fault-tolerant measures, a hybrid fault-tolerant model relying on the confident rate is structured. In this model, the various confident rates, including the node confident rate, the group confident rate and the feedback confident rate, are processed and transferred respectively. The feature from the clustering and tree structure can guarantee the convergence and correctness of the aggregation data, and enhance the capability of data fault-tolerance in large-scale and high-density sensor networks.To sum up, we propose an appropriate model over data fault- tolerance in the WSN and discuss ways of ensuring reliability in data aggregate operations in high-density and low node fault rate sensor network. We also explore approaches to solve fault-tolerant problems self-adaptively in monitoring, querying and fusing in the WSN.
|
PDF Full Text Request