Page 31 - Layout 1

Department of Engineering, ICT and Technologies for Energy and Transport Patent Title Method for the fusion of spatio-temporal trajectories. Ref. CNR 10485 Assignee (s): CNR CNR Institute: IEIIT Main Inventor: Marco Fiore Countries: IT Priority date: 11/07/2016 Abstract Digital transaction data are logs of timestamped, georeferenced events associated to the digital activities of individuals. Typical examples are mobile traffic datasets collected by cellular network operators. These data yield a wealth of information about the movements and undertakings of large mobile user populations, and have rapidly established as a paramount source of knowledge for multiple applications. However, disclosure of mobile traffic datasets is still largely withhold by privacy concerns, due to the uniqueness of user trajectories that makes individuals univocally recognizable even in very large populations. The object of this invention is k-merge, an algorithm that makes any number of sparse spatiotemporal trajectories identical to each other, through generalization. k-merge is a fundamental building block for the implementation of a vast number of anonymity criteria that preserve mobile traffic datasets from trajectory uniqueness. Background The common practice for privacy preservation consists in replacing personal with pseudo-identifiers (i.e., random or non-reversible hash values). Whether this is a sufficient measure is called into question, especially in relation to the possibility of tracking user movements. It has been repeatedly proven that pseudo-identifiers do not protect against user trajectory uniqueness. Although uniqueness is not a privacy threat per-se, but it is a vulnerability that can lead to re-identification, as shown by works in the literature. Technology Given a number k of spatiotemporal trajectories, k-merge merges them, i.e., it returns one spatiotemporal trajectory that is valid for all of the input trajectories. To that end, k-merge leverages a novel approach to spatiotemporal generalization, i.e., it reduces the precision of the input trajectories in space and time, so as to make any input trajectory indistinguishable from the other k-1 trajectories. The algorithm guarantees that the cost of the operation, i.e., the loss of accuracy, is minimized. Advantages and Applications Unlike other approaches in the literature, k-merge generalizes sparse spatiotemporal trajectories while obeying Privacy-Preserving Data Publishing principles and minimizing accuracy loss. It guarantees that any group of trajectories is merged while retaining a good level of accuracy in the data. It does not delete or create samples in the trajectories. The algorithm is the foundational block for the development of anonymisation solutions, and, as such, has applications in trajectory data analytics. Development stage The k-merge algorithm has been implemented and tested on real-world large-scale datasets provided by mobile network operators. The indicative TRL is 6. 21