DTCluster: A CFSFDP Improved Algorithm for RFID Trajectory Clustering Under Digital-twin Driven

In the field of indoor monitoring, to analyze the track data collected by RFID readers effectively and intuitively is the critical issue. The indoor environment is usually crowded with a large number of flexible and mobile obstacles. The movements of obstacles always change the road network structure. To analyze the trajectories under the labile road network constrains is the key problem. To solve this problem, we propose a trajectory clustering method based on digital-twin technology called DTCluster. The proposed digital-twin model is a virtual digital model, which can extract and display the changing road network structure through real-time mapping and virtual simulation functions. In our paper, the original CFSFDP algorithm is improved to reduce the influence of the track missing points and to enhance the clustering effect in the face of unequal trajectories in length. We also take two parameters: the speed and the direction of moving targets into account. The trajectory clustering results in our DTCluster follow the changing road network for synchronous display on the proposed digital-twin model. It is convenient and effective for researchers to use DTCluster to find the problems which cannot be observed in the clustering results displayed in original planned road network. These problems can only be observed in the current changed road network environment. In our paper, we demonstrate the validity and intuition of DTCluster byexprimental results.