Towards Secure and Scalable Energy Theft Detection: A Federated Learning Approach for Resource-Constrained Smart Meters

TL;DR

This paper addresses energy theft detection in smart grids under privacy and resource constraints by proposing a privacy-preserving federated learning framework that uses a lightweight MLP and Gaussian differential privacy. The approach enables on-device training for resource-constrained smart meters and is evaluated on real-world SGCC data under IID and non-IID distributions, achieving competitive detection metrics while reducing communication overhead. Comparisons with state-of-the-art FL methods illustrate favorable privacy-utility-efficiency trade-offs, particularly in non-IID settings. The work offers a practical, scalable pathway for secure energy theft detection in next-generation smart grids, balancing privacy, computation, and performance.

Abstract

Energy theft poses a significant threat to the stability and efficiency of smart grids, leading to substantial economic losses and operational challenges. Traditional centralized machine learning approaches for theft detection require aggregating user data, raising serious concerns about privacy and data security. These issues are further exacerbated in smart meter environments, where devices are often resource-constrained and lack the capacity to run heavy models. In this work, we propose a privacy-preserving federated learning framework for energy theft detection that addresses both privacy and computational constraints. Our approach leverages a lightweight multilayer perceptron (MLP) model, suitable for deployment on low-power smart meters, and integrates basic differential privacy (DP) by injecting Gaussian noise into local model updates before aggregation. This ensures formal privacy guarantees without compromising learning performance. We evaluate our framework on a real-world smart meter dataset under both IID and non-IID data distributions. Experimental results demonstrate that our method achieves competitive accuracy, precision, recall, and AUC scores while maintaining privacy and efficiency. This makes the proposed solution practical and scalable for secure energy theft detection in next-generation smart grid infrastructures.

Figures (4)

• Figure 1: FL Framework
• Figure 2: Performance of 2 clients for 3 epochs and 80 rounds
• Figure 3: Performance of 3 clients for 3 epochs and 80 rounds
• Figure 4: Performance of 5 clients for 3 epochs and 80 rounds