TRindex: Distributed Double-Layer Road Network Trajectory Index

Abstract

With the wide application of mobile device positioning technology, the scale of traffic trajectory data generated is becoming larger and larger. How to store this massive data is a hot research topic in recent years. At present, most stand-alone road network trajectory indexes process large-scale spatio-temporal data with low efficiency, and most distributed indexes either support few query methods or have low efficiency. Therefore, this paper proposes a distributed double-layer trajectory data indexing technology (TRindex). The index adopts a global-local two-layer structure. First, the global index is divided into upper and lower layers. The upper layer is the time shards based on the time attributes of massive trajectory data, and the lower layer is the STR partition built for each time shard. Next, a two-layer local index is constructed for each STR partition. The upper-level time index is constructed based on the linear order partition algorithm, and the lower-level R*-tree index is constructed based on the spatial attributes of the data in the partition; Secondly, the hot data cache scheduling algorithm and Redis are used to reduce the disk query overhead; In addition, to support the trajectory query method of moving objects, HBase is used to store trajectory data; At the same time, to ensure the load balance of nodes, the pre-partition strategy and consistent hash algorithm are adopted; Finally, the incremental update of the index is realized. The index's performance is evaluated by designing relevant experiments, and its efficiency and feasibility are verified by comparing it with the existing related work.