Finding dense locations in symbolic indoor tracking data: modeling, indexing, and processing

Finding the dense locations in large indoor spaces is very useful for many applications such as overloaded area detection, security control, crowd management, indoor navigation, and so on. Indoor tracking data can be enormous and are not immediately ready for finding dense locations. This paper presents two graph-based models for constrained and semi-constrained indoor movement, respectively, and then uses the models to map raw tracking records into mapping records that represent object entry and exit times in particular locations. Subsequently, an efficient indexing structure called Hierarchical Dense Location Time Index (HDLT-Index) is proposed for indexing the time intervals of the mapping table, along with index construction, query processing, and pruning techniques. The HDLT-Index supports very efficient aggregate point, interval, and duration queries as well as dense location queries. A comprehensive experimental study with both real and synthetic data shows that the proposed techniques are efficient and scalable and outperforms RDBMSs significantly.