基于联合稀疏模型的无线传感网数据重构算法jointsparsemodel.pdfVIP

第 37 卷第 Z1 期 通 信 学 报 Vol.37 No.Z1 2016 年 10 月 Journal on Communications October 2016 doi:10.11959/j.issn.1000-436x.2016269 基于联合稀疏模型的无线传感网数据重构算法 刘义颖，李国瑞，田丽 (东北大学计算机科学与工程学院，辽宁 沈阳 110819) 摘 要：无线传感器网络中数据具有较强联合稀疏特性，应用压缩感知理论，通过联合编码压缩数据，再使用联 合解码进行还原，可实现低采样代价收集传感数据。提出了一种基于联合稀疏模型与压缩感知理论的同步子空间 追踪算法，以稀疏特性为先验知识，通过回溯迭代方式，判断并选取合适的联合子空间，用更少量观测值实现原 始传感数据的精确重构。与 SCoSaMP 算法、SP 算法在不同稀疏特性和不同采样率下相比较，同步子空间追踪算 法具有较好的恢复性能。 关键词：无线传感器网络；压缩感知；联合稀疏模型；数据重构 中图分类号：TP393 文献标识码：A Joint sparse model based data reconstruction algorithm for wireless sensor network LIU Yi-ying, LI Guo-rui, TIAN Li (School of Computer Science and Engineering, Northeastern University, Shenyang 110819, China) Abstract: The data of wireless sensor network has strong joint sparse characteristics, by utilizing compressed sensing theory, compressed data by joint encoding, and then reconstructed the data by joint decoding, the sensed data can be gathered with low computational cost. A synchronous subspace pursuit algorithm based on joint sparse model and com- pressed sensing theory was proposed. By utilizing the sparsity of the sensed data, it selected the correct joint subspace and reconstruct the original signal group accurately with fewer observations in a backtracking iterative manner. Com- pared with SCoSaMP algorithm and SP algorithm, the proposed algorithm presents better data reconstruction perform- ance under the conditions of different sparsity and sampling rate. Key words: wireless sensor network, compressive sensing, joint sparse model, data reconstruc