An Energy-Efficient Downlink Resource Allocation In Cellular IoT H-CRANs

In order to maximize the efficiency with which wireless resources are put to use by many users in an OFDMA network, we investigate the intertwined problems of subchannel assignments and power allocation in this research. To address this non-convex as well as dynamic optimization issue in the presence of co-channel interference, we introduce multi-agent reinforcement learning (MADRL) oriented techniques in contrast to traditional single-agent algorithms. First, we use a multi-agent deep predictable policy gradient (MADDPG) system to dynamically results the transmission strength for each user based on the current channel gain, then we update the weights of everyone neural network in order to optimize the distribution of resources policy and achieve the maximum possible rate for each cell. The convergence and effectiveness of the suggested algorithms are shown by simulation results.