Deep Reinforcement Learning Based Resource Allocation for D2D Communications underlay Cellular Networks

摘要

The present invention relates to a deep reinforcement learning-based resource allocation method and device for D2D communication underlay cellular networks. The present invention includes the steps of dividing a single agent in a base station into a multi-agent structure, each of the multiple agents corresponding to a D2D link of a D2D terminal pair, and each artificial neural network of the multiple agents is configured to determine the transmission power and spectrum channel of the corresponding D2D terminal. It consists of steps including allocating resources.