多维分解加噪算法在智能电网隐私保护中的优化

在智能电网的数据采集监测中,针对用户隐私泄露安全隐患问题,采取加噪为主的方式来实现隐私保护。提出一种基于多维分解的拉普拉斯噪声算法(MDLN,multidimensional laplacian noise algorithm),该算法将原始测量值分解成多维数据,并根据各维度的隐私敏感度,自适应决定需添加的拉普拉斯噪声幅度,通过有效的噪声扰动方式实现差分隐私。通过与SLN(simple laplacion noise algorithm)算法ULN(uniform laplacian noise algorithm)算法相比较,仿真表明,MDLN算法的隐私保护强度较高,且效能更高。

Optimization of multi-dimensional decomposition and plus noise algorithm in intelligent grid privacy protection

To address the security problem of user privacy leak in the data acquisition and monitoring of smart grid, noise is usually added to achieve privacy protection. In this paper, a Laplacian noise algorithm based on multidimensional decomposition(MDLN) is proposed. The algorithm decomposes the original measured value into multidimensional data, and adaptively determines the Laplacian noise amplitude to be added according to the sensitivity of each dimension, achieving differential privacy by effective noise perturbation. The simulation results show that the MDLN algorithm has higher privacy protection and higher performance compared with the SLN(simple Laplacian noise algorithm) algorithm and ULN(uniform Laplacian noise algorithm) algorithm.