Abstract:
Smart grids are evolving towards a more decentralized, distributed, prosumer-centric, and autonomous energy exchange framework. Data-driven models are key for informed decision-making in such frameworks, yet bring significant privacy concerns on the prosumer part. End-users in the smart grid are becoming increasingly vigilant about unauthorized usage of their personal information, necessitating the demand for enhanced control over the processing of their data. In this paper, we introduce a pricing framework that can incorporate our proposed user-specific privacy levels for consumers, and based on that, propose a Privacy-Preserving Personalized Pricing Scheme (P4S) tailored for smart grids. In particular, we integrate the concept of differential privacy into the federated learning framework to facilitate and enable pricing discounts on electricity bills for customers based on their privacy budgets and preferences. In this way, end-users are empowered with the freedom to customize their privacy choices, while guaranteeing transparency in the form of the cost associated with privacy. We conduct simulations, and the results show that our scheme ensures robust convergence performance across diverse privacy settings. We anticipate that P4S will promote trust enhancement and user empowerment, encouraging consumers to actively participate in the emerging distributed energy market.