Research On Key Techniques Of Fault Tolerance In Wireless Sensor Network

Wireless sensor network (WSN) is a new field in information science, serving as the results of cross and merge between the traditional disciplines and modern disciplines. WSN consists of a kind of embedded computers, named"motes"(nodes). These embedded computers are micro-computers with independent energy supplying, computing abilities and communication abilities. The communication of WSN can be built by the self-reorganization of nodes in difference application scenes. However, the faults of nodes occur frequently in application because of the limited energy capacity, computing abilities, being maintenanced abilities of nodes. These existences of faults will reduce the efficiency of communication and quality of service (QoS), and will result in a wide range of network failure in serious cases. As a result, the fault tolerance ability of WSN is desiderated in most applications.In this dissertation, firstly, the division of fault type in WSN is proposed according to four factors including the energy, communication, hardware and software. Based on the fault type, the hierarchical fault model of WSN is proposed and studied, including the nodes level and subnetwork level, and the fault model of key units in nodes level is emphasized. The proposal of hierarchical fault model in WSN makes it possible of fine-grained research in fault area.Secondly, based on the subnetwork fault model, a series of concepts and theorems of key nodes of WSN, including key nodes and key links etc., are proposed, and focusing on the average path length and clustering coefficient in complex network theory, qualitative research and quantitative research are given subsequently about key nodes in sub-network level. The probability of the generation of key nodes and node critical connectivity are discussed and analyzed, and six policies are proposed to improve the fault tolerant abilities of nodes and networks, from the factors of topology distribution, clustering coefficient, degree-sensitive algorithm family, node deployment density, sub-network division and fault tolerance of nodes. Based on the theory of degree-degree correlation, both bridge type key nodes and key nodes of degree correlation in WSN were defined, and quantitative determination and analysis were proposed based on them.Thirdly, the concept of key nodes subset is proposed in WSN. In order to design and divide the subset of key nodes, a kind of algorithm based on topology generation technology, named KNSG algorithm, is proposed and implemented. The determination coefficient is emphasized, mean and simple time series moving average method are applied for degree prediction in KNSG algorithm. Subsequently, the result of simulation and analysis are given based on experiments.Fourthly, the fault tolerance ability of wireless sensor nodes is studied in order to improve reliability of nodes and network. Especially, the research of fault tolerant routing protocols is focused on, and two kinds of fault tolerant routing algorithm are proposed, including FTASNB algorithm and FTA-RN algorithm. The results of simulation show that the success ratio of data acceptance of target nodes increases in both algorithms, in contrast with same type traditional algorithm.In summary, this dissertation focus on the research of a series of key techniques on fault tolerance in WSN, including fault models of the network, key nodes of network fault, key nodes subsets of network fault and fault tolerant routing algorithms. Novel topology generation algorithm and network routing algorithms are proposed to improve the abilities of fault tolerance in WSN, and these feasibility and effectiveness are proved by simulations.