An Artificial Intelligence Framework for Dynamic Selection and Resource Allocation for EVs in Vehicular Networks

Wireless power transfer for charging electric vehicles (EVs) using the inductive charging mechanism enables EVs to recharge their batteries wirelessly while in motion or during halts at traffic signals. Additionally, resonant inductive charging (RIC) can wirelessly transfer energy with high efficiency and over longer distances. To achieve these objectives, we propose a RIC-enabled system for traffic signal scenarios capable of selecting EVs based on their required energy needs for travel and the remaining traffic signal time. We formulate an optimization problem that allows the selected EVs to maximize their battery energy during traffic signal halts, consequently optimizing the defined utility function. Given the dynamic nature of the problem, it falls under the category of NP-hard problems, and to address this, we propose a novel artificial intelligence framework. Our approach utilizes both the halt time and energy demand information, resulting in the development of the Traffic Signal-Aware Electric Vehicle Selection and Resource Allocation algorithm. We employ long short-term memory based deep learning model to predict battery energy needs and generate energy demand score information. The generated demand score, along with the remaining traffic signal time, serves as conditions for the final selection and allocation of charging resources to EVs. Finally, experimental results confirm the effectiveness of our proposed method, demonstrating superior performance compared to the deep neural network model. Furthermore, in terms of battery energy, our approach achieves a 55.1% increase compared to baseline-B and an 8.4% increase compared to baseline-C.