Reinforcement learning for interference-aware cell DTX in heterogeneous networks

This paper focuses on Inter-Cell Interference Coordination for small cells implementing cell discontinuous transmission (DTX) to enhance the network energy efficiency. The small cell activity is dynamically orchestrated such that the inter-cell interference is limited, the power consumption of Radio Access Network and backhaul is jointly reduced, and the Quality of Service (QoS) constraints are satisfied. To achieve this goal, we develop a Reinforcement Learning framework that achieves network-wise optimization in stochastic environments. Our solution leads to notable energy saving with respect to the state of the art DTX approaches without affecting the QoS. Moreover, the user performance is close to the one experienced when a centralized scheduler is used to limit the inter-cell interference.