LAUSR

In a cellular system with distributed MU-MIMO, an application of cluster-wise distributed MU-MIMO reduces the computational complexity. However, both the intracell interference and the intercell interference are produced. Considering the scalability of the system, in this letter, we propose a fully decentralized interference coordination (IC) which jointly applies the graph coloring algorithm (GCA) and the deep reinforcement learning (DRL). Based on online training with consideration of the time-varying wireless environment, our proposed joint IC can adapt quickly to the changing environment. The simulation reveals that our proposed joint IC can significantly improve the capacity compared to the no IC case.