融合压缩感知与张量理论的多参量异构WSN高维数据感知重建研究

With the continuous improvement on the strategic position of the Internet of Things, Wireless Sensor Networks ushered in a new research boom. However, there are still some obstacles for heterogeneous WSN with multiple types of sensors in application deployment, such as high network power consumption, limited global information acquisition, and poor robustness. This project presents an innovation of utilizing the compressed sensing and tensor theory in high-dimensional data collection and reconstruction for multi-attribute heterogeneous WSN. It fully exploits and utilizes the correlation between high-dimensional data based on tensor to solve the problem in multi-attribute heterogeneous WSN application. Data in multi-attribute heterogeneous WSN can be distributed in a high-order tensor according to different nodes, collection time, and sensed information types. The data features based on high-order tensor are analyzed with different tensor decomposition methods (e.g., Tucker decomposition); The collection model and missing features of high-dimensional data are studied for sparse sampling aggregation networks, global information collection networks, and data loss networks, respectively; The high-dimensional data reconstruction methods in multi-attribute heterogeneous WSN are established based on compressed sensing and tensor theory to achieve high-precision data reconstruction with low measurement; The real-time data collection and reconstruction model in multi-attribute heterogeneous WSN is constructed to ensure real-time performance with high-precision reconstruction. The successful development of this research is of significance to the research of high-dimensional data collection and reconstruction in multi-attribute heterogeneous WSN.

无线传感器网络随物联网战略地位的不断提升迎来了一场新的研究热潮，然而，具有多类型传感器的异构WSN在应用部署中仍面临网络能耗高、全局信息感知受限及系统鲁棒性低等问题。本项目创新地将压缩感知及张量理论引用到多参量异构WSN高维数据感知重建中，基于张量形式充分挖掘并利用该高维数据间相关性，拟解决多参量异构WSN应用部署中的难题。针对多参量异构WSN感知数据，按感知节点、采集时刻及信息类型分布在高阶张量中，基于Tucker等张量分解方法研究分析高阶张量形式下数据特征；分别针对稀疏感知汇聚、全局感知受限及数据传输丢失网络，研究其高维数据感知模型及缺失特征；基于压缩感知与张量理论构建多参量异构WSN高维数据重建方法，实现低测量值下的高精度数据重建；研究构建多参量异构WSN实时数据感知重建模型，确保实时性的同时实现高精度数据重建。本项目的研发成功对多参量异构WSN高维数据感知重建研究发展具有重要意义。

无线传感器网络随物联网战略地位的不断提升迎来了一场新的研究热潮，然而，具有多类型传感器的异构WSN在应用部署中仍面临网络功耗高、全局信息感知受限及系统鲁棒性低等问题。本项目将压缩感知及张量理论引用到多参量异构WSN高维数据感知重建中，基于张量形式充分挖掘并利用该高维数据间相关性，以解决多参量异构WSN应用部署中的难题。项目围绕多参量异构WSN高维数据感知重建、基于张量理论的高维数据重建和去噪、WSN数据实时汇聚和全局感知三个方面展开研究，研究分析高阶张量形式下多参量异构WSN高维数据分布及特征，建立网络数据感知模型；融合压缩感知与张量理论构建多参量异构WSN高维数据重建方法，实现低测量值下的高精度数据重建；实现WSN高精度实时数据稀疏感知模型和重建方法。先后发表高质量论文16篇，其中在Signal Processing等SCI检索的期刊上发表论文10篇；IEEE通信领域知名会议IEEE GLOBECOM发表会议论文并做分组报告；授权专利5项。项目研究成果为大规模多参量WSN高维数据监测的应用部署提供了理论支撑，并可进一步扩展到物联网海量异构数据处理领域。