consists of 400 trajectories and can be used for the ground truth verification of sub-trajectory clustering tasks. The creation scenario of the synthetic dataset is the following: Objects move upon a simple graph that consists of the following destination nodes (points) with coordinates: A(0, 0), B(1, 0), C(4, 0) and D(2, 1). It is assumed that half of the objects are moving with normal speed (2 units per second) and the rest of them are moving with high speed (5 units per second). The dataset's figureillustrates the 2-D map of the SMOD with the three one-directional roads and one bidirectional road. The objects are moving under the following scenario (rules), for a lifetime of one hundred seconds:
-There exist three one-directional roads (A → B, B → D, D → C) and one bidirectional road (B ⇆ C).
-At t = 0 sec, all objects of MOD start from (in a real application, very close to) point A. Thus, the first destination of all objects is point B.
-When an object arrives at a destination point, it ends its trajectory with a probability of 15%. Otherwise, it continues with the same speed to the next point. If there exist more than one possible next point, it decides randomly the next destination.
-A small number of objects (outliers, four in our experiment) randomly moved in space ignoring roads. In addition, the speed of outliers randomly changes.
Finally, the dataset comes in two versions. We have added Gaussian white noise of different Signal to Noise Ratio (SNR) levels, measured in db, to spatial coordinates of the SMOD. The two versions have additive noise of SNR = 50 db and SNR = 30 db, respectively.
This is a synthetic dataset created by InfoLab (http://infolab.cs.unipi.gr).