Smart City Gnosys
Smart city article details
| Title | Toward Privacy-Aware Federated Analytics Of Cohorts For Smart Mobility |
|---|---|
| ID_Doc | 57713 |
| Authors | Gjoreski M.; Laporte M.; Langheinrich M. |
| Year | 2022 |
| Published | Frontiers in Computer Science, 4 |
| DOI | http://dx.doi.org/10.3389/fcomp.2022.891206 |
| Abstract | Location-based Behavioral Analytics (LBA) holds a great potential for improving the services available in smart cities. Naively implemented, such an approach would track the movements of every citizen and share their location traces with the various smart service providers—similar to today's Web analytics systems that track visitors across the web sites they visit. This study presents a novel privacy-aware approach to location-based federated analytics that removes the need for individuals to share their location traces with a central server. The general approach is to model the behavior of cohorts instead of modeling specific users. Using a federated approach, location data is processed locally on user devices and only shared in anonymized fashion with a server. The server aggregates the data using Secure Multiparty Computation (SMPC) into service-defined cohorts, whose data is then used to provide cohort analytics (e.g., demographics) for the various smart service providers. The approach was evaluated on three real-life datasets with varying dropout rates, i.e., clients not being able to participate in the SMPC rounds. The results show that our approach can privately estimate various cohort demographics (e.g., percentages of male and female visitors) with an error between 0 and 8 percentage points relative to the actual cohort percentages. Furthermore, we experimented with predictive models for estimating these cohort percentages 1-week ahead. Across all three datasets, the best-performing predictive model achieved a Pearson's correlation coefficient above 0.8 (strong correlation), and a Mean Absolute Error (MAE) between 0 and 10 (0 is the minimum and 100 is the maximum). We conclude that privacy-aware LBA can be achieved using existing mobile technologies and federated analytics. Copyright © 2022 Gjoreski, Laporte and Langheinrich. |
| Author Keywords | federated analytics; location-based services; mobile computing; mobility modeling; privacy |
Similar Articles
| Id | Similarity | Authors | Title | Published |
|---|---|---|---|---|
8067  | 0.868 | Badu-Marfo G.; Farooq B.; Mensah D.O.; Al Mallah R. | An Ensemble Federated Learning Framework For Privacy-By-Design Mobility Behaviour Inference In Smart Cities | Sustainable Cities and Society, 97 (2023) |
| 55733 | 0.864 | De Mattos E.P.; Domingues A.C.S.A.; Santos B.P.; Ramos H.S.; Loureiro A.A.F. | The Impact Of Mobility On Location Privacy: A Perspective On Smart Mobility | IEEE Systems Journal, 16, 4 (2022) |
| 43109 | 0.859 | Montori F.; Bedogni L. | Privacy Preservation For Spatio-Temporal Data In Mobile Crowdsensing Scenarios | Pervasive and Mobile Computing, 90 (2023) |
| 59153 | 0.853 | Schneider M.; Schneider J.; Löffelmann L.; Christen P.; Rahm E. | Tuning The Utility-Privacy Trade-Off In Trajectory Data | Advances in Database Technology - EDBT, 26, 3 (2023) |
| 35525 | 0.85 | Yang, DQ; Qu, BQ; Cudre-Mauroux, P | Location-Centric Social Media Analytics: Challenges And Opportunities For Smart Cities | IEEE INTELLIGENT SYSTEMS, 36, 5 (2021) |