Power data publishing method based on differential privacy considering spatiotemporal correlation

With the continuous construction of smart grid, more and more power data can be easily collected through smart meters. These collected power data can be used for analysis and research, so as to improve the level of Saudi power grid construction. However, if these power data are directly released, it may reveal the privacy of power users. Therefore, this paper presents a new power data publishing method to protect the user’s privacy. The attacker can use the background knowledge to obtain the user’s privacy by analyzing the published power data. According to the attack model, this paper designs the probability density function of the corresponding disturbance data, and randomly disturbs the power data according to the probability density function of the disturbance data. Secondly, when designing the noise density function, we consider that the power data between the two dimensions may have spatiotemporal correlation. In order to avoid destroying the correlation when adding random disturbances, we design the probability density function directly from the perspective of two dimensions. Finally, in order to ensure the availability of data, the range of random disturbance is given. Experiments show that our method can not only ensure the privacy of users, but also ensure the reliability of data.