Preventive Audits for Data Applications Before Data Sharing in the Power IoT

Bohong Wang, Yingying Wang, Qinglai Guo, Yanxi Lin, Yang Yu

Abstract

With the increase in data volume, more types of data are being used and shared, especially in the power Internet of Things (IoT). However, the processes of data sharing may lead to unexpected information leakage because of the ubiquitous relevance among the different data, thus it is necessary for data owners to conduct preventive audits for data applications before data sharing to avoid the risk of key information leakage. Considering that the same data may play completely different roles in different application scenarios, data owners should know the expected data applications of the data buyers in advance and provide modified data that are less relevant to the private information of the data owners and more relevant to the nonprivate information that the data buyers need. In this paper, data sharing in the power IoT is regarded as the background, and the mutual information of the data and their implicit information is selected as the data feature parameter to indicate the relevance between the data and their implicit information or the ability to infer the implicit information from the data. Therefore, preventive audits should be conducted based on changes in the data feature parameters before and after data sharing. The probability exchange adjustment method is proposed as the theoretical basis of preventive audits under simplified consumption, and the corresponding optimization models are constructed and extended to more practical scenarios with multivariate characteristics. Finally, case studies are used to validate the effectiveness of the proposed preventive audits.

Introduction: The Data Sharing Dilemma in Power IoT

In the modern energy landscape, the Power Internet of Things (IoT) generates massive volumes of granular data. While electricity retailers require this data for demand forecasting and market management, its sharing introduces significant privacy risks. Data patterns in electricity consumption often possess "ubiquitous relevance," meaning they are statistically linked to sensitive private traits such as employment status, household size, or income level, even after traditional anonymization.

Currently, data sharing often grants "usufructuary rights"—the right to use and profit from an asset—without transferring ownership. The primary risk is that a data buyer might use the data for "unexpected applications." For instance, a load profile purchased for forecasting could be exploited by a credit agency to infer personal details. There is a critical need for formal mathematical tools that allow data owners to vet and adjust data before it is shared.

Figure 3: Data sharing ecosystem in Power IoT, highlighting the conflict between utility requirements for expected applications and privacy requirements against unexpected ones.

The Preventive Audit Framework

To address this, researchers from Tsinghua and Beihang Universities propose a "preventive audit" framework based on semantic information theory. This approach treats data sharing more like leasing a property: the owner grants usage rights but "locks the doors" to private information the tenant does not need.

The framework uses Mutual Information (MI) as a core metric to quantify the correlation between unshared data ($Y$) and two types of characteristics: Private ($X^{(1)}$) and Non-private/Utility ($X^{(2)}$). By measuring the "leakage matrix" between these variables, data owners can conduct audits based on how feature parameters change before and after sharing.

Figure 1: Conceptual model of the preventive audit for multivariate characteristics, showing the relationship between unshared data (Y), shared data (U), and private/non-private traits.

Methodology: Probability Exchange Adjustment

The technical core of the audit is the Probability Exchange Adjustment Method. This involves subtly shifting the joint probability distribution of the data to "exchange" probability mass. The goal is to decrease the MI for private traits while maintaining or increasing it for utility traits.

The authors developed two optimization models:

  1. Constant Probability Distribution: Maintains the original statistical "shape" of the shared data, offering a simpler but less flexible privacy-utility trade-off.
  2. Variable Probability Distribution: Allows the data's distribution to change, enabling the audit to reach the "Pareto optimal bound"—the theoretical limit of maximum utility for a given level of privacy.

Case Study and Results

The framework was validated using the CER Ireland Smart Meter Data, which includes consumption profiles and survey responses for 143 social characteristics. Testing on 10 representative traits (e.g., internet access, family size) demonstrated successful "inverse adjustment":

  • Privacy Protection: MI for "internet access" (private) was reduced from 0.0180 to 0.0002.
  • Utility Enhancement: MI for "family members" (utility) was increased from 0.0294 to 0.2571.

The "Variable Distribution" method successfully hit the theoretical Pareto optimal surface in complex multivariate scenarios, balancing multiple private traits against utility needs.

Conclusion and Future Directions

While the mathematical foundation is robust, challenges remain in computational complexity as the number of characteristics grows. Future work aims to integrate these audits into automated data transaction systems using smart contracts and blockchain, ensuring data is cleared automatically before any trade occurs. This shift from syntactic to semantic privacy protection marks a significant step toward secure data ecosystems in the Power IoT.

Summary updated

Abstract

With the increase in data volume, more types of data are being used and shared, especially in the power Internet of Things (IoT). However, the processes of data sharing may lead to unexpected information leakage because of the ubiquitous relevance among the different data, thus it is necessary for data owners to conduct preventive audits for data applications before data sharing to avoid the risk of key information leakage. Considering that the same data may play completely different roles in different application scenarios, data owners should know the expected data applications of the data buyers in advance and provide modified data that are less relevant to the private information of the data owners and more relevant to the nonprivate information that the data buyers need. In this article, data sharing in the power IoT is regarded as the background, and the mutual information of the data and their implicit information is selected as the data feature parameter to indicate the relevance between the data and their implicit information or the ability to infer the implicit information from the data. Therefore, preventive audits should be conducted based on changes in the data feature parameters before and after data sharing. The probability exchange adjustment method is proposed as the theoretical basis of preventive audits under simplified consumption, and the corresponding optimization models are constructed and extended to more practical scenarios with multivariate characteristics. Finally, case studies are used to validate the effectiveness of the proposed preventive audits.
BibTeX
@article{wang2024preventive,
  author = {Bohong Wang and Yingying Wang and Qinglai Guo and Yanxi Lin and Yang Yu},
  title = {{Preventive Audits for Data Applications Before Data Sharing in the Power IoT}},
  journal = {IEEE Internet of Things Journal},
  year = {2024},
  doi = {10.1109/jiot.2024.3483929},
}
Figure