A Convoy Discovering Algorithm for Passengers in the Cruise Based on UWB Positioning

To accurately discover the convoy pattern among passengers in the interior space of cruise, UWB positioning is employed in the cruise to provide on-board personnel positioning. An improved Haussdorff-DBSCAN based scheme is proposed to find trajectory clustering of the cruise passengers, considering the characteristics of UWB location data and the semantic sequence of passenger-wise location. In the first phase, traditional Haussdorff distance calculation method cannot precisely express the similarity between two trajectories. The location semantic information is added to the calculation so that it will be more accurate when discover the convoy pattern of trajectories. In the second phase, the input of the improved Hausdorff-DBSCAN algorithm is the passenger trajectories data set, and the clustering radius is determined according to the overall similarity threshold of trajectories. The output are the emerging clusters of passenger trajectories in the same companion group. Due to the reason that there is no condition to install UWB positioning devices in a real cruise, the validity of the presented algorithm is verified by the trajectory data obtained from the passengers simulation on one deck of the cruise, which is modeled in Anylogic. The results indicate that the precision of the algorithm reaches 0.92, the recall value reaches 0.93, substantially higher than the existing algorithm.