面向路网轨迹的自适应数据模型与索引结构

针对现有路网轨迹数据模型与时空索引结构自适应调节能力低的问题，提出了一种面向路网轨迹的自适应数据模型与时空索引结构，以支持路网时空轨迹的高效存储与查询。所提出的自适应时空数据模型为多层CLR数据模型的扩展，该模型以从时空轨迹群中挖掘的高频路网路径为主要网络线性元素建立自适应线性基准，并根据自适应线性基准对路网时空轨迹进行转换，转换后的时空轨迹其时空子实体数量变少，可以通过更高的效率进行存储；所提出的自适应时空索引结构为基于LRS的时空索引结构的扩展，该索引结构根据自适应线性基准构建自适应线性参考系统，基于自适应线性参考系统的索引结构其保存的时空子实体数量变少，可以通过更高的效率进行时空查询。为了验证所提出方法的有效性，本文最后采用真实开源T-Drive出租车轨迹数据集与人工合成轨迹数据集进行了充足的实验。实验以2种常见的时空相交查询类型为例，将所提出的方法与原始数据模型以及时空索引结构进行了存储效率和查询效率的对比。对比分析结果表明，所提出的自适应数据模型与索引结构最高能够提升40%的存储效率以及50%的查询效率，为路网轨迹数据的管理提供了新的解决方案。

Adaptive Data Model and Index Structure for Network-constrained Trajectories

An adaptive spatiotemporal data model and a spatiotemporal index structure are proposed to support efficient storage and querying of network-constrained trajectories. The proposed adaptive spatiotemporal data model extends the hierarchical Compressed Linear Reference (CLR) data model by establishing the adaptive route-based linear datum in road network with high-frequency network routes mined from the trajectory dataset. Network-constrained trajectories can be transformed from the link-based linear datum to the adaptive route-based linear datum, and the transformed trajectories consist of fewer sub-entities that can be stored with lower storage capacity. The proposed adaptive spatiotemporal index structure is an extension of the LRS-based index structure, which is constructed based on the adaptive route-based Linear Reference System (LRS). Fewer spatiotemporal sub-entities are saved in the adaptive spatiotemporal index structure, which allows for efficient spatiotemporal querying of network-constrained trajectories. In order to verify the effectiveness of the proposed adaptive data model and index structure, adequate experiments are conducted at the end of this paper using the real open-source T-Drive taxi trajectory dataset and the synthetic trajectory dataset. The experiments take two popular spatiotemporal intersection queries as an example, and the proposed adaptive data model and index structure with the conventional hierarchical CLR data model and the LRS-based spatiotemporal index structure are compared in terms of storage efficiency and query efficiency. The analysis results show that the proposed adaptive data model and index structure can improve storage efficiency by 40% and query efficiency by 50%, which confirms that the proposed method can provide a new solution for the management of network-constrained trajectory data.