Indoor Trajectory Prediction for Shopping Mall via Sequential Similarity

With the prevalence of smartphones and maturation indoor positioning techniques, predicting movement a large number customers in environments has become promising challenging line research recent years. While most current approaches that take advantage mathematical methods perform well outdoor settings, they exhibit poor performance environments. To solve this problem, study, sequential similarity-based prediction approach which combines spatial semantic contexts into unified framework is proposed. We first present revised Longest Common Sub-Sequence (LCSS) algorithm to compute similarity trajectories, then novel considering R-tree proposed similarities; after this, considered group clustered trajectories are used train models. Extensive evaluations were carried out on real-world dataset collected from shopping mall validate our method. The results show markedly outperforms baseline can be scenarios.