基于混沌压缩感知的传感网序贯式数据收集关键理论与技术研究

There are three shortcomings in the existed compressive sensing theory based data collection mechanism in wireless sensor networks. They are the insufficiency of spatial correlation utilization in the sensed data set, sensing and transmitting redundant data values within sensor nodes, and the absence of reuse mechanism in the data reconstruction process. Aiming at the above problems and with the main goal of improving the efficiency of data collection and reducing the resource consumption, we concentrate on the following there key problems based on the chaotic compressive sensing theory and sequential compressive sensing theory combined with the usage of Lasso optimization model and homotopy method. They are (1) the sparse data source selection problem in wireless sensor networks, (2) the data reconstruction problem of sequential compressive sensing, (3) the reconstruction stopping rule of sequential sensing data collection problem in the noise condition. In the project, we expect to build chaotic compressive sensing theory based data collection model, sequential compressive sensing theory based data reconstruction model and the reconstruction stopping rule of sequential sensing data collection problem in the noise condition. Eventually, we will reduce the required energy consumption of computation, storage, and transmission in the data collection process and prolong the survival lifetime of wireless sensor network. Meanwhile, we can also reuse the data reconstruction results in the server end and improve the efficiency of data recovery.

现有的基于压缩感知理论的无线传感器网络数据收集机制中存在着缺乏利用采集数据集合的空间关联性，传感器节点采集和传输过多的数据，数据重构过程缺乏复用机制这三项不足之处。本项目针对上述问题，以提高数据收集操作效率，降低系统资源消耗为主要需求，基于混沌压缩感知理论和序贯压缩感知理论，结合利用Lasso优化模型和同伦方法，重点研究解决三个关键问题：（1）无线传感器网络中稀疏数据源的选取问题；（2）序贯式压缩感知数据的重构问题；（3）噪声条件下序贯测量数据收集操作的停止规则问题。项目预期建立基于混沌压缩感知理论的数据收集模型、序贯压缩感知数据重构模型以及噪声条件下序贯测量数据收集操作停止规则，最终达到即能有效降低整个传感器网络数据收集过程所需的计算、存储和传输的能量消耗，延长无线传感器网络的生存时间，又能在服务器端复用数据重构过程的计算结果，提高数据恢复的效率。

本项目针对无线传感器网络数据收集机制中存在着缺乏利用采集数据集合的空间关联性，传感器节点采集和传输过多的数据，数据重构过程缺乏复用机制这三项不足之处，以提高无线传感器网络数据收集操作的效率，降低系统资源消耗为主要目标，基于压缩感知理论，设计了相应的数据收集模型与重构算法，开展了无线传感器网络中关于数据收集关键理论与技术的研究工作。项目具体从无线传感器网络的高效数据收集方案、基于压缩感知的无线传感器网络数据重构方案、无线传感器网络中数据稀疏特征提取模型与恢复方案等方面进行了深入细致的研究，所设计的一系列方案既能有效地降低整个无线传感器网络数据收集过程所需的计算、存储和传输的能量消耗，延长无线传感器网络的生存时间，又能在服务器端高效、精确地重构无线传感器网络感知的监测数据。同时，本项目的研究成果还将会对压缩感知理论在其它相关应用领域中的大规模普及应用提供基础理论和技术支撑作用。