Intelligent Cache-Assisted Mobile Edge Computing via Deep Learning

Abstract

Mobile edge computing (MEC) and caching stand out as pivotal technologies for the future of wireless networks. Efficiently anticipating users' demands holds paramount importance in meeting the surge in user requests. Leveraging the high prediction accuracy of deep learning (DL) and the recent enhancements in computational capabilities, DL algorithms are seamlessly integrated into wireless systems. Hence, in this letter, in a wireless heterogeneous environment comprising multiple small base stations (SBS) linked to a central base station (BS), we propose a joint integration of K-means clustering and DL framework to predict popular contents. The central BS collects the information from all the SBSs to learn the data. Then, the popularity rank for each content is obtained, and the contents are cached in real-time. Tensorflow and Keras libraries are used for the prediction model on the MovieLens dataset. Simulation results are provided to show that the proposed model significantly outperforms the recent prediction models in terms of average cache hit rate and mean squared error. © 2025 IEEE.