An Efficient Hybrid Index Structure for Temporal Marine Data

摘要

Marine data is a typical big data that features multi-source, multi-class, multi-dimension and massiveness. Rapid query to big marine data is the fundamental request in vast marine applications. To improve query performance, we should devise a complete index structure. In this paper we propose a multi-layer index (ML-index, for short) with regarding to Time Interval B~+-tree and Hybrid Space Partition Tree (HSP-tree, for short). It employs Marine data value function that consists of data time length, data access frequency etc. to optimize the primary key index (i.e. B~+-tree). Moreover, we propose an adaptive space partition method on the basis of data characters, user query habits and data unit capacity particularly. Furthermore we build a secondary index, namely, the HSP-tree over the above partition result. We show the results of experiment that compares ML-index with two state-of-the-art index methods on the real marine data. These suggest that the ML-index enable user to perform marine data query in about 2/3 the time needed by the state-of-the-art tools.