D2D power control based on supervised and unsupervised learning

With the coming of the era “Big Data”, Machine Learning (ML) is one of the most significant issues to draw people's attention. This paper is an attempt to conflate a machine learning technique with Device-to-device (D2D) communications. D2D communications underlying the cellular infrastructure are a technology that have been proposed recently as a promising solution to enhance cellular network capabilities. However, since the same resources are shared by both systems, interference is a major challenge. Therefore, interference management techniques such as power control (PC) are needed to control the interference. In this paper, we consider two PC algorithms based on ML algorithms, distributed Q-learning and CART Decision Tree. And then through comparing the two ML algorithms with traditional PC methods in terms of computational complexity, communication throughput and energy efficiency, we provide insight into the potential of fusion of machine learning and wireless communications.