Zhenping Chen

Federated learning enables privacy-preserving short-term household load forecasting by keeping fine-grained electricity consumption data on local devices. However, its deployment over wireless networks is constrained by communication overhead, time-varying channel conditions, and client heterogeneity, particularly the scarcity and uneven distribution of peak-period samples. This paper proposes a wireless federated learning framework coordinated by over-the-air computation and Q-learning for peak-aware household load forecasting. First, a peak-aware data construction strategy together with a risk-sensitive evaluation scheme is introduced to explicitly characterize peak underestimation and peak-interval reliability. Second, to accommodate link dynamics and imbalance in peak-related informativeness, a Q-learning-guided adaptive weighting mechanism is developed to adapt client participation and aggregation-weight concentration according to channel quality and peak-related indicators. Third, weighted model aggregation is achieved through synchronous uplink signal superposition, thereby reducing uplink transmission burden and communication latency. Experiments on smart-meter datasets under time-varying wireless channels show that, compared with conventional federated learning baselines, the proposed method achieves more stable convergence, improves prediction accuracy during peak periods, reduces underestimation risk, and maintains competitive overall forecasting performance.

Smart agriculture is rapidly evolving in response to growing global demands for food security and sustainable resource management. Cloud–edge–device collaborative computing has emerged as a transformative paradigm, addressing the limitations of traditional centralized architectures by enabling distributed intelligence, real-time processing, and adaptive decision-making. This review provides a comprehensive overview of the architectures, technical characteristics, and application scenarios of cloud–edge–device collaboration in agriculture. Key domains covered include environmental monitoring, intelligent irrigation, UAV–machinery coordination, livestock health management, and pest and disease control. Major challenges such as device heterogeneity, data consistency, resource constraints, and privacy concerns are identified and discussed. Furthermore, six critical research directions are outlined, including intelligent scheduling algorithms, lightweight edge AI, hierarchical data fusion, federated learning, interoperability frameworks, and digital twin technologies. This review aims to serve as a practical reference and theoretical foundation for advancing the design and implementation of next-generation smart agriculture systems.

Accurate parameter estimation is vital in instrumentation and measurement, especially for modeling dynamic systems based on measurement data. Recurrent networks have been widely adopted for this purpose, with the output-input feedback (OIF) Elman network offering enhanced modeling through its dual-feedback structure. However, most existing parameter estimation algorithms for such a network underutilize historical measurements and exhibit limited adaptability to its dual-feedback mechanism. This paper proposes a dynamic factor and multi-innovation-based gradient parameter estimation algorithm for the OIF Elman network. The algorithm is derived from a multi-innovation cost function augmented with a weight decay factor, enabling dynamic balancing between current and past measurements to improve estimation accuracy. A dynamic factor adjustment strategy is introduced to adaptively tune both the weight decay factor and the self-feedback factors of the network based on the evolving weight parameter estimates. Unlike batch-based methods, the proposed algorithm supports recursive updates, enabling real-time parameter estimation from sequential measurement data. Convergence and computational complexity analyses confirm the theoretical properties of the proposed algorithm, while numerical simulations and a real-world electric load forecasting case study demonstrate its superior modeling performance compared to existing methods.

Image classification is crucial for various applications, including digital construction, smart manufacturing, and medical imaging. Focusing on the inadequate model generalization and data privacy concerns in few-shot image classification, in this paper, we propose a federated learning approach that incorporates privacy-preserving techniques. First, we utilize contrastive learning to train on local few-shot image data and apply various data augmentation methods to expand the sample size, thereby enhancing the model’s generalization capabilities in few-shot contexts. Second, we introduce local differential privacy techniques and weight pruning methods to safeguard model parameters, perturbing the transmitted parameters to ensure user data privacy. Finally, numerical simulations are conducted to demonstrate the effectiveness of our proposed method. The results indicate that our approach significantly enhances model generalization and test accuracy compared to several popular federated learning algorithms while maintaining data privacy, highlighting its effectiveness and practicality in addressing the challenges of model generalization and data privacy in few-shot image scenarios.

With the rapid development of autonomous driving and edge computing, vehicular edge computing (VEC) has become an emerging paradigm that allows vehicles with abundant computational resources to work as edge nodes. By introducing vehicles as infrastructures, VEC has the potential to improve users’ quality of experience and decrease operator’s deployment expenditure, especially for hot spots. In this article, a novel VEC-based resource reservation framework is designed to handle the time-varying computation requests. To articulate realistic scenarios, the online durations of provider vehicles (PVs) are assumed to be different. Besides, the PVs will not always be online to wait for the reservation assignment for the limited revenue, i.e., the PVs are dynamic and the computational resource reservation points (CRRPs) are static. In this way, dynamic matching is leveraged to model the interaction between the PVs and CRRPs. To prevent the CRRPs from manipulating their preferences for better partners, a strategy-proof and stable resource reservation algorithm is proposed to ensure all CRRPs are truthful during the resource reservation procedure. Finally, numerical simulation results are presented to validate the proofness, truthfulness, and performance of our proposed resource reservation algorithm.

In wireless sensor networks, terminal devices with restricted cost and size have limited battery life. Meanwhile, these energy-constrained devices are not easy to access, especially when the terminal devices are located in severe environments. To recharge the energy-constrained devices and extend their network service time, unmanned aerial vehicles (UAVs) equipped with wireless power chargers are leased by the third-party control center. To incent the participation of UAVs with different charging capabilities and ensure the strategy-proofness of the incentive mechanism, a hidden information based contract theory model, specifically adverse selection, is introduced. By leveraging individual rationality and incentive compatibility, a contract theory based optimization problem is then formulated. After reducing redundant constraints, the optimal contract items are derived by Lagrangian multiplier. Finally, numerical simulation results are implemented to compare the prepared algorithm with three other baselines, which validates the effectiveness of our proposed incentive mechanism.

According to the development and research of sliding mode variable structure control (SMVSC) in recent years. This paper mainly reviews the development and applications of some popular sliding mode control (SMC), such as adaptive SMC, discrete-time system SMC, fuzzy SMC, and neural network SMC. With the research and development of switched system (SS), the research scholars can deal with the complex and changeable system more effectively by combining the SMC with the SS, so as to improve the efficiency and stability of the system. The theoretical research and applications of SMC combined with fuzzy SS, predictive SS and uncertainty SS are also becoming more and more attractive. The purpose of this paper is to analyze and organize the current research status of the popular directions of SMC, and provide the readers with the latest research results on the application directions and new trends in the field of SMC. By considering these theoretical studies and numerous applications, the research and development of intelligent control techniques and switching systems based on sliding mode control are reviewed. Finally, some prospects for related research are presented.