Abstract:In mobile crowdsensing systems, the public crowd are required to report data with actual locations under location privacy vulnerabilities. Moreover, even sensing data itself further deepens location privacy breaches. Existing works allow each worker to consider his own privacy, but the accumulated privacy budget will lower down group data privacy of each sensing region. Moreover, multi-region spatial data correlations indicate that multi-group correlated data privacy may be leaked from each other. To this end, we develop a novel MCS framework, called GDA-Crowd(Group effect-based Data Aggregation), which consists of three parts: location obfuscation and aggregation, group effect-based data privacy and aggregation, and incentive mechanism. We start from individual location privacy guarantee and propose a location aggregation method to cluster workers into groups. Then, we exploit intra-group effect, i.e., data privacy interdependence under the judicious selection of workers’ participation, to enhance privacy-accuracy balance. Moreover, multi-group global histogram incorporates inter-group effect, i.e., correlated privacy loss from spatial data correlations, into inter-group data aggregation. Finally, we design a truthful, individually rational and computationally efficient incentive mechanism for participant selection. The synopsis of contributions includes dual privacy protection, dual group effect for desirable privacy-accuracy tradeoff, synergies among incentive mechanism, privacy-preserving data aggregation for approximate optimality. Theoretical analysis and extensive experiments validate our effectiveness and superiority.

Geometric Data Transfer Method for Privacy Preservation in Clustering Data Mining

Privacy Preserving Distributed DBSCAN Clustering

Effective Privacy Preserved Clustering Based on Voronoi Diagram

Group Effect-Based Privacy-Preserving Data Aggregation for Mobile Crowdsensing

A privacy preserving clustering technique using hybrid data transformation method

Privacy Preserving Method Based on GM(1, 1) and Its Application to Clustering

Privacy Data Preserving Method Based on Discrete Cosine Transform Martrix

Privacy-Preserving DBSCAN Clustering over Vertically Partitioned Data.

Privacy preserving clustering over distributed data

A Differential Privacy-Based Protecting Data Preprocessing Method for Big Data Mining

Clustering Algorithm For Privacy Preservation On Mapreduce

A New Hybrid Approach for Privacy Preserving Distributed Data Mining

A DCT‐based privacy‐preserving approach for efficient data mining

MHD: A New Method Towards Privacy Protecting Datasets Published

Application of Information Security on Data Mining

Privacy Preserving Based on Model Division for Large Data

A New Method on Personalized Privacy Preserving Multi-classification

Tools for privacy preserving Kernel methods in data mining

Big Data Analysis and Perturbation using Data Mining Algorithm

A Density Peaking Clustering Algorithm for Differential Privacy Preservation.

Privacy-Preserving Data Publication for Clustering