Sébastien Gambs

Mobility data has become a strategic asset in urban planning, crisis management and smart city operations. However, centralized systems for mobility tracking raise severe privacy concerns as they have the ability to directly link individuals to their movements. To address these issues, we propose LoChain, a decentralized protocol that enables the privacy-preserving collection and processing of mobility data based on blockchain technology. More precisely, LoChain replaces precise coordinates with standardized geoaddresses, associate user movements to disposable identities, communication them via Tor routing and stores the resulting data across a decentralized network built on Hyperledger Fabric. The system also employs a novel geopool and multi-channel architecture to simulate sharding, enabling localized data ingestion, inter-district communication and global statistical aggregation without compromising individual privacy. Localized position obfuscation and pseudo-random identity purging are used to further prevent reidentification. A proof-of-concept prototype, including an Android app, blockchain backend and visualization layer was developed and evaluated using synthetic data from 10,000 virtual users. The experiments results obtained from the simulation highlight the LoChain’s ability to preserve user privacy while maintaining analytical utility. Finally, we also introduce an incentive model as well as a decentralized governance structure to ensure long-term scalability, regulatory compliance and participatory control.

In this paper, we introduce WaKA (Wasserstein K-nearest neighbors Attribution), a novel attribution method that leverages principles from the LiRA (Likelihood Ratio Attack) framework and k-nearest neighbors classifiers (k-NN). WaKA efficiently measures the contribution of individual data points to the model’s loss distribution, analyzing every possible k-NN that can be constructed using the training set, without requiring to sample subsets of the training set. WaKA is versatile and can be used a posteriori as a membership inference attack (MIA) to assess privacy risks or a priori for privacy influence measurement and data valuation. Thus, WaKA can be seen as bridging the gap between data attribution and membership inference attack (MIA) by providing a unified framework to distinguish between a data point’s value and its privacy risk. For instance, we have shown that self-attribution values are more strongly correlated with the attack success rate than the contribution of a point to the model generalization. WaKA’s different usages were also evaluated across diverse real-world datasets, demonstrating performance very close to LiRA when used as an MIA on k-NN classifiers, but with greater computational efficiency. Additionally, WaKA shows greater robustness than Shapley Values for data minimization tasks (removal or addition) on imbalanced datasets.