WSNs连通性与拓扑容错性问题的研究

As a new generation of information technology, wireless Sensor Networks (WSNs) have been widely used in many fields, and gained worldwide attention. Connectivity and topological fault-tolerance are two important criteria of network performance, and this project focuses on “how to maintain connectivity and improve topological fault-tolerance in WSNs”. Fault diagnosis and localization mechanisms collaborated with connectivity restoration schemes have the ability of maintaining the network connectivity. Meanwhile, the construction of (k,m)-CDS based virtual backbones can improve the fault-tolerance of network topology. Therefore, several topics are investigated, which include: .1) To achieve a great boost for accuracy with a little cost of complexity, an identification code based hierarchical fault diagnosis and localization scheme will be designed;.2) Measured by the ratio between the actual value and the theoretical optimum in resource consumption (also known as approximation ratio), a connectivity restoration scheme will be devised using network geometric properties and graph theory algorithms;.3) Measured by the optimum approximation ratio in data collection cost, a data collection based connectivity restoration scheme in random terrains will be proposed;.4) Measured by the optimum approximation ratio in the construction of (k,m)-CDS, a construction scheme based on (k,m)-CDS will be developed..Results of this project will not only improve the practicability of fault diagnosis and localization, connectivity restoration and the construction of (k,m)-CDS based virtual backbone, but also offer technical references to the study of both connectivity and topological fault-tolerance of WSNs. Therefore, the main topics of this project have both theoretical significances and practical values.

无线传感器网络(WSNs)是新一代信息技术，应用广泛，备受关注。连通性与拓扑容错性是网络两大性能指标。本课题围绕“如何保障WSNs连通性、提高其拓扑容错性”展开研究。故障诊断与定位技术结合连通性修复技术能保障WSNs连通性，而基于(k,m)-CDS的虚拟主干网构造技术能提高其拓扑容错性，因此本课题研究内容包括：①以较小复杂度提高诊断精度为指标，研究基于识别码的分层式故障诊断定位策略；②以资源消耗及其理论最优值的比（近似比）为指标，研究消耗最少资源的连通性修复策略；③以数据采集代价的最小近似比为指标，研究随机地形下基于数据采集的连通性修复策略；④以(k,m)-CDS构造的最小近似比为指标，研究基于(k,m)-CDS的虚拟主干网构造策略。本课题研究可促进故障诊断与定位、连通性修复和虚拟主干网构造这三项技术更加实用化，为WSNs连通性与拓扑容错性研究提供理论和技术参考。

项目从无线传感器网络的连通性和拓扑容错性角度出发，将图理论、故障诊断机理、人工智能与区块链技术相结合，构建连通的、可靠的、容错的无线传感器网络。课题组主要的创新性成果包括研究基于几何结构（直骨架、费马点）的网络连通性修复机理，提出基于机器学习（深度强化学习、卡曼滤波径向基神经网络）和区块链的连通性修复、数据采集策略，提出基于滤波理论（卡曼滤波）的分级故障诊断算法，并研究基于故障诊断度理论的网络可靠性和容错路由机制。以上研究成果具有比较明显的源创性，多项核心成果发表在国际主流权威期刊并受了关注和引用。..项目的最终研究成果及各项指标均达到和超过了预期目标，主要包括在国内外学术刊物和期刊上发表14篇学术论文，核心技术和方法申请了4项专利，项目资助并培养了3位硕士研究生。