Database Reference
In-Depth Information
Figure 9.4 Milan GPS trajectories.
is, the groups of trajectories that share common mobility behavior, such as the
commuters that follow similar routes in their home-work and work-home trips.
An anonymizing transformation of the trajectories consists of the following
steps:
1. Characteristic points are extracted from the original trajectories: starting
points, ending points, points of significant turn, points of significant stop
(Figure
9.5
a);
2. Characteristic points are clustered into small groups by spatial proximity
(Figure
9.5
b);
3. The central points of the groups are used to partition the space by means of
Voronoi tessellation (Figure
9.5
c);
4. Each original trajectory is transformed into the sequence of Voronoi cells that
it crosses (Figure
9.5
d).
As a result of this data-driven transformation, where trajectories are gen-
eralized from sequences of points to sequences of cells, the probability of re-
identification already drops significantly. Further techniques can be adopted to
lower it even more, obtaining a safe theoretical upper bound for the worst case
(i.e., the maximal probability that the linkage attack succeeds), and an extremely
