SRS-LM: differentially private publication for infinite streaming data

Abstract

Recently, differential privacy achieves good trade-offs between data publishing and sensitive information hiding. But in data publishing for infinite streams, along with the increasing release of streaming data, the privacy budget consumption of every data continues grow, leading to a low-level data utility. To remedy this, this paper proposes a Laplace Mechanism based on Simple Random Sampling (SRS-LM) to differentially private publishing infinite streaming data. Specifically, we generate a Laplace series, which has a finite length, as the fundamental noise series and when an update comes, we randomly simple the noise from the Laplace series to add to the update data. Experimental results show that SRS-LM outperforms state-of-the-art differential privacy mechanisms in terms of security and mean absolute error for large quantities of queries.