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

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.

Non-Convex Optimization in Federated Learning via Variance Reduction and Adaptive Learning

This paper proposes a novel federated algorithm that leverages momentum-based variance reduction with adaptive learning to address non-convex settings across heterogeneous data. We intend to minimize communication and computation overhead, thereby fostering a sustainable federated learning system. We aim to overcome challenges related to gradient variance, which hinders the model's efficiency, and the slow convergence resulting from learning rate adjustments with heterogeneous data. The experimental results on the image classification tasks with heterogeneous data reveal the effectiveness of our suggested algorithms in non-convex settings with an improved communication complexity of $\mathcal{O}(\epsilon^{-1})$ to converge to an $\epsilon$-stationary point - compared to the existing communication complexity $\mathcal{O}(\epsilon^{-2})$ of most prior works. The proposed federated version maintains the trade-off between the convergence rate, number of communication rounds, and test accuracy while mitigating the client drift in heterogeneous settings. The experimental results demonstrate the efficiency of our algorithms in image classification tasks (MNIST, CIFAR-10) with heterogeneous data.

Anomalous Client Detection in Federated Learning

Federated learning (FL), with the growing IoT and edge computing, is seen as a promising solution for applications that are latency- and privacy-aware. However, due to the widespread dispersion of data across many clients, it is challenging to monitor client anomalies caused by malfunctioning devices or unexpected events. The majority of FL solutions now in use concentrate on the classification problem, ignoring situations in which anomaly detection may also necessitate privacy preservation and effectiveness. The system in federated learning is unable to manage the potentially flawed behavior of its clients completely. These behaviors include sharing arbitrary parameter values and causing a delay in convergence since clients are chosen at random without knowing the malfunctioning behavior of the client. Client selection is crucial in terms of the efficiency of the federated learning framework. The challenges such as client drift and handling slow clients with low computational capability are well-studied in FL. However, the detection of anomalous clients either for security or for overall performance in the FL frameworks is hardly studied in the literature. In this paper, we propose an anomaly client detection algorithm to overcome malicious client attacks and client drift in FL frameworks. Instead of random client selection, our proposed method utilizes anomaly client detection to remove clients from the FL framework, thereby enhancing the security and efficiency of the overall system. This proposed method improves the global model convergence in almost 50\% fewer communication rounds compared with widely used random client selection using the MNIST dataset.