基于并行压缩感知的物联网海量数据处理.docVIP

第 29 卷第 10 期2012 年 10 月计算机应用与软件Computer Applications and Softwar 第 29 卷第 10 期 2012 年 10 月 计算机应用与软件 Computer Applications and Software Vol. 29 No． 10 Oct． 2012 基于并行压缩感知的物联网海量数据处理 张永平 张功萱 朱昭萌 张 巍 郭 箭 ( 南京理工大学计算机科学与技术学院 江苏 南京 210094) 摘 要 在大规模和超大规模物联网中往往需要采集和处理海量的数据，这对智能感知系统的寿命及物联网的性能将产生很大 影响。提出利用压缩感知采样方法减少物联网中的采集数据，压缩感知是一种能够实现采样和数据压缩同时进行的新型数据采样 方法，它通过降低采样率减少采集的数据量。压缩感知算法的计算复杂度较高、自适应性较差，因此提出并实现压缩感知算法的并 行处理以提高算法执行速度并尝试引入冗余字典改善算法的灵活性。实验证明并行处理可以显著提高算法的执行速度、增强数据 处理的实时性。最后，讨论了引入压缩感知理论后可能的物联网结构变化和未来的工作。 关键词 物联网 海量 数据处理 信号重构 并行处理 结构 中图分类号 文献标识码 TP39 A DOI: 10． 3969 / j． issn． 1000-386x． 2012． 10． 016 MASS DATA PROCESSING IN INTERNET OF THINGS BASED ON PARALLEL COMPRESSED SENSING Zhang Yongping Zhang Gongxuan Zhu Zhaomeng Zhang Wei Guo Jian ( School of Computer Science and Technology，Nanjing University of Science and Technology，Nanjing 210094，Jiangsu，China) Mass data are usually collected and processed in large and ultra large-scale Internet of Things，and this will greatly affect the Abstract life of IntelliSense equipment and the performance of Internet of Things． In this paper，we propose to reduce the collected data from Internet of Things by using compressed sensing sampling method． Compressed sensing is a new sampling method that the data sampling and compress- ing can be done simultaneously． Compressed sensing can significantly reduce the collected data size by lowering the sampling rates of sensors， but it is non-adaptive and its algorithm has high computational complexity as well． We put forward and achieved the parallel processing of compressed sensing algorithm for improving algorithm’s execution speed，and tried to introduce redundant dictionary to increase the flexibility of the algorithm． Experiments prove that parallel processing can significantly improve the execution speed of the compressed sensing algo- rithm，enhance the real-time property of data processing． In end of the paper，we discussed the changes in the structure of the Internet of Things