物联网中压缩感知算法云加速方法.pdfVIP

第43 卷 第3 期 电 子 科 技 大 学 学 报 Vol.43 No.3 20 14年5月 Journal of University of Electronic Science and Technology of China May 20 14 物联网中压缩感知算法的云加速方法 张永平，张功萱，朱昭萌 (南京理工大学计算机科学与工程学院 南京 210094) 【摘要】为了减少采集的数据量，提出在物联网中引入“边采样边压缩”的新型采样方法——压缩感知。针对压缩感知 理论中信号重建算法计算复杂度较高的问题，设计并实现了一个基于云平台和代码迁移的算法加速方案；该方案解决了代码 并行化的自动翻译、算法向云端迁移、本地和云端执行同步等问题，对可并行化的算法，仅需要增加几个新定义的接口及插 入一些描述性的注释，就可以利用云资源实现算法的加速；实验表明，该方案是可行的、有效的。该文还研究了基于物联网 资源的云加速方法，提出了基于云加速方案、结合多核/ 多CPU方法和GPGPU方法，能充分利用已有物联网资源的混合压缩感 知算法加速框架，并初步设计了理论运行流程。 关 键 词 算法加速; 云计算; 压缩感知; 物联网; 并行化 中图分类号 TP391 文献标志码 A doi:10.3969/j.issn.1001-0548.2014.03.017 Cloud Acceleration Method for Compressed Sensing in Internet of Things ZHANG Yong-ping, ZHANG Gong-xuan, and ZHU Zhao-meng (School of Computer Science and Engineering, Nanjing University of Science and Technology Nanjing 210094) Abstract In order to reduce the amount of data collected, the compressed sensing (CS), a new sampling method, is used for data acquisition and processing in Internet of Things (IoT). To overcome high computational complexity of CS algorithms, an acceleration scheme based on cloud platform and code migration is introduced in this paper. The scheme solved the automatically translated problem of parallelization code, the migration problem of algorithm, and the synchronization problems of local and cloud. It can use the resources from cloud environment to speed up algorithms by adding several interfaces and inserting some comments. In addition, this paper studies the method of cloud acceleration based on computing resources in Internet of Things, and put forwards an acceleration fram