基于多变元可视化的网络诊断关键技术研究

项目名称： 基于多变元可视化的网络诊断关键技术研究

项目编号： No.61502374

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 自动化技术、计算机技术

项目作者： 李瑞

作者单位： 西安电子科技大学

项目金额： 22万元

中文摘要： 无线传感网是物联网感知现实世界的基础组成部分，是连接现实世界与数字世界的桥梁，被广泛的应用于环境感知、危险境地导航与交通流量监控等诸多领域。然而，由于传感网的节点资源受限，数据传输无线多跳，计算分布式自组织，运行环境复杂多变等因素的影响，传感网自身的健康状况难以得到可靠保证。在大规模传感网中，这样的高不可靠性尤为明显。本课题主要研究基于多变元可视化的网络异常检测方法以及传感网诊断策略的评估选择方法。面对传感网收集的诊断数据高维异构，计算能力较弱等挑战，课题组拟从数据可视化的方法出发，研究网络异常检测，设计高效的检测算法，实现传感网的智能可管理。在诊断策略的评估与选择设计方面，构建限定性条件的诊断策略形式化方法，并通过多目标优化算法指导选择诊断策略，引领未来的诊断策略设计研究。基于本课题的研究为传感网的稳定持久运行提供了必要的保障。

中文关键词： 网络诊断；异常检测；数据可视化；无线传感网；物联网

英文摘要： Wireless sensor network is a fundamental part of IOTs, and it is a bridge connected the real world with the digital world, is widely used in many areas of environmental awareness, danger navigation and traffic monitoring. However, due to the node resource constrained, wireless multi-hop data transmission, the impact of self-organization of distributed computing, complex and volatile operating environment and other factors, the health status of the sensor network itself is difficult to get reliable guarantee. Especially in large-scale sensor networks, such high unreliability is particularly evident. The main research contents of our project are multivariate visualization-based network anomaly detection approach and assessment of diagnostic strategies. Faced with high-dimensional and heterogeneous data and weak computing ability of the node, our project intends to depart from the data visualization methods, and study network anomaly detection, design efficient detection algorithm, in order to guarantee the intelligent management of the sensor network. For the assessment and selection of diagnostic strategies, build restrictive conditions to formalize the diagnostic strategies, and guide the selection of diagnostic strategies through multi-objective optimization, further to lead the future design of diagnostic strategies. What we have done is to provide the protection of durable and stable operation for sensor networks.

英文关键词： Network Diagnosis ;Anomaly Detection;Data Visualization;Wireless Sensor Network;Internet of Things