Processing location prediction queries by pattern indices

Predicting the possible next location of moving objects is a helpful task in database systems and wireless networks. Extracting mobility patterns from past trajectories is very suitable and different methods have been studied for this purpose. But most of them, except a few, focus on data mining techniques and not on database requirements such as proper indexing techniques and management of discovered patterns. Some of the existing methods do not store timings for patterns and none of them can handle long-term predictions. In this paper, we propose an indexing technique called Trajectory Pattern Indexing for Prediction or TPIP. In complexity analysis, it shows lower searching time than just one similar solution. It also makes it possible to process long-term predictions. With TPIP, we can store patterns with same location sequence but different timing; with no redundancy.