不确定性多维传感器数据的有效存储与查询方法

Efficient storage and query method for multidimensional uncertain sensor data

为解决传统数据库管理技术无法有效管理不确定性数据的问题,该文设计了一种多维数组树(MB树)。MB树是一种基于贝叶斯网络的图数据结构,以贝叶斯网络作为概率图模型解决存储和查询问题。对海量数据建模并响应查询。证明了可预测性和结构关联性。利用真实数据集和合成数据集对MB树的性能进行了测试。验证了具有潜在联合分布的MB树的编码准确度。与相似的图模型比较,采用MB树的查询处理效率平均可提升约3倍。

To solve the problem that traditional database management technology can’ t manage uncertain data efficiently,a multidimensional array B-tree（ MB-tree） is designed here. The MB-tree is a graph data structure based on Bayesian network. Bayesian network is used as a probabilistic graphical model to solve the storage and query problem of uncertain data. Mass multidimensional sensor data is modeled and responds to query. The predictability and relevance of multidimensional data structure are proved. The performance of the MB-tree is tested using real data sets and synthetic data sets. The coding accuracy of the MB-tree with potential co-distribution is verified. The query efficiency of the MB-tree is about 4 times as fast as those of alike graphical models.