• Energy Research

With the prevalence of smart appliances, smart meters, and Internet of Things (IoT) devices in smart grids, artificial intelligence (AI) built on the rich IoT big data enables various energy data analysis applications and brings intelligent and personalized energy services for users. In conventional AI of Things (AIoT) paradigms, a wealth of individual energy data distributed across users’ IoT devices needs to be migrated to a central storage (e.g., cloud or edge device) for knowledge extraction, which may impose severe privacy violation and data misuse risks. Federated learning, as an appealing privacy-preserving AI paradigm, enables energy data owners (EDOs) to cooperatively train a shared AI model without revealing the local energy data. Nevertheless, potential security and efficiency concerns still impede the deployment of federated-learning-based AIoT services in smart grids due to the low-quality shared local models, non-independently and identically distributed (non-IID) data distributions, and unpredictable communication delays. In this article, we propose a secure and efficient federated-learning-enabled AIoT scheme for private energy data sharing in smart grids with edge-cloud collaboration. Specifically, we first introduce an edge-cloud-assisted federated learning framework for communication-efficient and privacy-preserving energy data sharing of users in smart grids. Then, by considering non-IID effects, we design a local data evaluation mechanism in federated learning and formulate two optimization problems for EDOs and energy service providers. Furthermore, due to the lack of knowledge of multidimensional user private information in practical scenarios, a two-layer deep reinforcement-learning-based incentive algorithm is developed to promote EDOs’ participation and high-quality model contribution. Extensive simulation results show that the proposed scheme can effectively stimulate EDOs to share high-quality local model updates and improve the communication efficiency.

In the process of cities informatization, the smart cities use of information technology and service system to handle of urban problems, improving people's lives. We study the problem of indoor signal coverage in smart cities. Indoor signal coverage inside large buildings is important to provide guaranteed communication quality for indoor mobile users. Specifically, traditional cellular coverage is deficit for indoor communications in the following three aspects: the emergence of the mobile signals are weak, or even blind at indoor environment, due to the complicated interior building structure; in some area with high traffic, traditional network is far from sufficient to meet the capacity needs of users; radio frequency interference problem between floors seriously a2642ects the stability of mobile signals and communication quality. Previous research mainly focuses on the outdoor macro cell coverage, and fails to meet the surge demand of indoor communication demand. Indoor coverage system is therefore demanded to solve the issue. In this paper, we investigate on the planning of an indoor distribution system. Based on the analysis of an indoor coverage system, an TD-LTE system is developed to provide indoor signal coverage. We implement our design in a real-world scenario. Using realworld experiments, we verifies the performance of the proposed system.

Energy consumption is the key to restrict the development of electric vehicles (EV), which is heavily affected by complex driving behaviors. In this paper, we propose a classified driving behavior based energy consumption prediction model, as well as recommended mechanisms for energy-efficient driving. Firstly, utilizing six EVs, we collect real data related to driving behaviors and energy consumption of vehicle in one year. After clustering behaviors of drivers, we present an energy consumption predication model, which accurately forecast the energy consumption caused by different driving behaviors. Motivated by the model, energy-saving strategies are proposed to recommend suitable driving behaviors. The simulation results further indicate that the accuracy of the proposed model is up to 98%. Specifically, the proposed model is less dependence on data volume, which guarantees the precision of more than 96% when the data volume is very small.