Comparative Study of Grid-Inverted List Hybrid Indexing Techniques for Moving Objects and Queries

Abstract

Advancement in GPS technologies and availability of a variety of devices to capture location and other attributes of the objects has led to an enormous development in geo-textual data. Searching through these objects for relevant objects as per the requirement needs efficient indexing technique and searching algorithm. Past queries, Present queries and Future queries are the three types of geo-textual queries. Past queries are responded based on the historical locations of the moving objects stored in the database. Future queries can be answered if the velocity vector of the object is known in advance. But in many real-time applications the position of the objects in future cannot be predicted. In such a scenario capturing movement of the objects and queries in real time and answering queries or updating query answer sets in real time is essential. In this paper three different techniques based on grid index, modified to handle geo-textual queries using hybrid index, YPK-CNN, SEA-CNN and CPM to handle real time queries, are presented. The methods to find kNN based on these three techniques are proposed in this paper and are also compared. Conceptual partitioning along with hybrid index improve the query performance by 30 to 40%.