AUTHOR=Zhang Hao , Wang Yue , Park Hyeong Kwang Benno , Sung Tae Hyun 

TITLE=Design and investigation of small-scale long-distance RF energy harvesting system for wireless charging using CNN, LSTM, and reinforcement learning

JOURNAL=Frontiers in Physics

VOLUME=Volume 12 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/physics/articles/10.3389/fphy.2024.1337421

DOI=10.3389/fphy.2024.1337421

ISSN=2296-424X

ABSTRACT=This paper introduces a novel small-scale long-distance RF energy harvesting system tailored for wireless charging, employing a fusion of CNN, LSTM, and reinforcement learning. The primary aim is to combat the energy supply challenges in wireless charging applications. The hybrid model combines CNN for RF signal feature extraction and LSTM to capture temporal dependencies within the RF signal data. Additionally, reinforcement learning algorithms are integrated to optimize energy harvesting and wireless charging processes, boosting overall system performance. Data collection involved gathering RF signal data at varying distances and documenting the correlated RF energy harvesting outcomes. Subsequently, rigorous data cleaning and preprocessing led to the creation of distinct training and testing datasets. The CNN-LSTM model underwent training on the prepared data, optimizing its hyperparameters for enhanced performance. The evaluation involved scrutinizing the model's accuracy in predicting RF energy harvesting and wireless charging efficiency using the dedicated testing data. Further enhancements were achieved by integrating reinforcement learning algorithms. Defining a reward function incentivized efficient wireless charging and maximal energy harvesting, empowering the system to dynamically adapt strategies in real-time conditions. This adaptability maximized energy harvesting efficiency and charging effectiveness. The experimental validation substantiated the system's superior performance, showcasing significant advancements in long-distance RF energy harvesting and wireless charging. The system's scope extends beyond its current applications, exhibiting potential in various wireless charging domains. Its reliability in providing energy support for upcoming wearable devices, IoT, and mobile devices is notably promising. The significance of this research lies in its pioneering integration of deep learning and reinforcement learning methodologies into a compact long-distance RF energy harvesting system. This innovative approach could catalyze advancements in wireless charging technology, offering fresh perspectives and avenues for intelligent and user-friendly wireless charging solutions.