Mobile Crowdsensing Location Aggregation Data Release with Differential Privacy Protection

In the scenario of mobile crowdsensing, the release of location aggregation data has supported the development of many domains, however, personal sensitive privacy information of mobile users implicit in location aggregated data may be disclosed as a result, which greatly discourages people from sharing their own location data. In this paper, we propose a differential privacy-based mobile crowdsensing location aggregation data release scheme Re-LDCR. Specifically, to make the privacy budget application adaptable to data changes while avoiding excessive budget consumption, we use the defined data change rate as the basis for budget allocation, and combine the recycle factor to limit the allocation proportion. Then, to improve the noise resistance of individual data, we comprehensively consider the data change characteristics and privacy protection ability, using BIRCH clustering to group the data with similar features. Finally, the combination of prediction, sampling, perturbation, and filtering mechanisms ensures the data utility of privacy protection results. Experimental results show that the proposed Re-LDCR outperforms the existing scheme and achieves a balance between privacy protection effectiveness and data utility.