E-DPNCT: an enhanced attack resilient differential privacy model for smart grids using split noise cancellation

Area of Science:

• Cybersecurity
• Smart Grid Technology
• Data Privacy

Background:

• High-frequency energy consumption data in smart grids reveals sensitive consumer lifestyle information, posing significant security and privacy risks.
• Existing differential privacy (DP) models for smart grids, while aiding analysis for billing and load monitoring, are susceptible to collusion attacks by adversaries.
• These attacks exploit collaborations between malicious smart meters and untrusted aggregators to extract private data.

Purpose of the Study:

• To demonstrate the vulnerability of current DP-based privacy models in smart grids against collusion attacks.
• To propose a novel, collusion-resistant privacy model for smart grid data.
• To ensure accurate load monitoring and billing while safeguarding sensitive consumer information.

Main Methods:

• The study first establishes the need for a collusion-resistant model by analyzing DP model vulnerabilities.
• A new model, Enhanced Differential Private Noise Cancellation Model for Load Monitoring and Billing for Smart Meters (E-DPNCT), is proposed.
• E-DPNCT utilizes differential privacy with a split noise cancellation protocol involving multiple master smart meters (MSMs) for collusion resistance.

Main Results:

• The proposed E-DPNCT model demonstrates resistance against collusion attacks, a significant improvement over existing models.
• Simulations using real-time data show substantial enhancements in privacy protection under attack scenarios.
• Comparisons with state-of-the-art models like EPIC confirm E-DPNCT's effectiveness against collusion attacks.

Conclusions:

• The E-DPNCT model effectively protects smart grid data privacy against collusion attacks.
• It provides accurate load monitoring and billing services concurrently with robust privacy preservation.
• The model offers a viable solution for enhancing security and privacy in smart grid data analysis.