传感数据起源信息的多粒度标注与多分辨压缩

Multi-Granularity Annotations and Multi-Resolution Compression Algorithm for Sensor Data Provenance

为了有效地追溯传感器网络中传感数据的生成与处理历史,提出一种面向无线传感器网络数据重发布的数据起源生成及传播方法.根据传感网数据起源特点分析了3种不同粒度起源的创建和传播.提出mgp-SQL语句实现了查询传播意义上起源的多粒度化;为降低传感网对起源信息的传输成本和存储成本,提出了一种基于离散小波分解理论的多分辨率起源压缩方法;根据传感网络起源信息的生成和发布的过程,结合上述2种技术实现了传感数据起源追溯算法MMPQ.真实数据及合成数据上的测试实验证明,MMPQ可以在查询过程中灵活调节压缩比率,生成适当分辨率的起源标注并高效存储.

In order to trace the producing and processing history of data in sensor network efficiently,a generation and transmission method of data provenance management for data distribution in sensor network is proposed.According to the characteristics of data provenance in sensor network,generations and transmissions of 3 different granularity provenances are introduced and new mgp-SQL query syntax is proposed to realize multi-granularity provenance annotation.Then a method of multi-resolution provenance compression based on discreet wavelet resolution theory is presented to reduce the costs of data transmission and storage in sensor network.Finally,a provenance query algorithm for sensor data-MMPQ is realized in combination with 2 previous technologies.Experiments on real and synthetic data set show that MMPQ can flexibly adjust the compression ratio in the query process and generate the appropriate resolution provenance annotation and gain cost-effective storage capacity.