项目名称: 基于时空相关性的无线传感器网络节能问题研究
项目编号: No.61272449
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 严建峰
作者单位: 苏州大学
项目金额: 80万元
中文摘要: 无线传感器网络观测数据的时空相关性揭示了无线传感器网络部署环境的动态性与不确定性,对指导和优化无线传感器网络的节能设计有重要意义。本项目以概率图模型为主要工具,分析和建模无线传感器网络中普遍存在的时空相关性,在此基础上研究高效的节能策略,主要包括:研究传感器节点的时空相关性双聚类方法,同时从时间和空间两个维度将观测数据进行聚类分析;基于时空相关性聚类结果,研究相应的拓扑控制策略来降低活跃节点的数量,达到显著节能的效果;利用动态条件随机场对观测数据的时空相关性进行建模,研究相应的参数估计方法,通过对活跃节点的观测来精确推断其他节点的状态。本项目的预期成果拟在保证实际应用性能的前提下,有效的降低必需的活跃节点数量,达到无线传感器网络显著节能的目标。
中文关键词: 无线传感器网络;时空相关性;主题模型;参数推理;
英文摘要: The spatial and temporal correlations of observations gathered by Wireless Sensor Networks(WSN) reveal the dynamic and uncertain nature of the deployment environment. Analysis and modeling of these spatial and temporal correlations brings significant advantages to design and optimization of energy conservation for WSN. In this project, we use probabilistic graphical models as the main tools to analyze and model the spatial and temporal correlations universally existing in WSN, based on which an efficient energy conservation scheme is proposed. We first study the biclustering problem for the spatial and temporal correlations and analyze the observations from both time dimension and space dimension to cluster the observation matrix. Based on the spatial and temporal correlation clusters achieved, we reduce the necessary number of active sensor nodes to effectively conserve energy using topology control theory. We then model the spatial and temporal correlations using Dynamic Conditional Random Field (DCRF) and study parameters estimation methods, according to which we can infer status of other nodes by observations of the active nodes. Our project aims to achieve significant energy conservation by reduction of number of necessary active nodes with the same performance as that of none node-reduction scheme.
英文关键词: wireless sensor networks;spatio-temporal correlation;topic model;parameter inference;