Mitigating consumer privacy breach in smart grid using obfuscation-based generative adversarial network.

Area of Science:

• Computer Science
• Electrical Engineering
• Cybersecurity

Background:

• Smart meters enable real-time power monitoring, but expose granular consumption data.
• This fine-grained data can lead to privacy breaches by revealing household appliance usage patterns.
• Existing privacy-preserving methods like perturbation and anonymization have significant limitations.

Purpose of the Study:

• To propose a novel privacy-preserving architecture for fine-grained smart grid power data.
• To enhance consumer privacy against behavior leakage from smart meter data.
• To ensure the generated synthetic data maintains minimal energy difference from the ground truth.

Main Methods:

• Utilized a Generative Adversarial Network (GAN) combined with an obfuscator.
• Developed an architecture to generate synthetic time-series power consumption data.
• Replaced real appliance signatures with synthetic ones, simulating inactive appliances.

Main Results:

• The proposed architecture effectively generates synthetic power data.
• Demonstrated enhanced privacy protection compared to existing methods.
• Validated effectiveness using real-world power consumption datasets and non-intrusive load monitoring (NILM) algorithms.

Conclusions:

• The GAN-based architecture offers a robust solution for smart meter data privacy.
• The method successfully masks appliance usage while preserving data integrity.
• This approach significantly improves consumer privacy in smart grid environments.