Online reliability optimization for URLLC in HetNets: a DQN approach

摘要: Heterogeneous cellular networks (HetNets) have been proven as a promising approach to deal with ever-growing data traffic. Supporting ultra-reliable and low-latency communication (URLLC) is also considered new feature of the upcoming wireless networks. Due overlapping structure mutual interference between cells in HetNets, existing resource allocation approaches cannot be directly applied for real-time applications, especially URLLC services. As novel unsupervised algorithm, Deep Q Network (DQN) has already many online complex optimization models successfully. However, it may perform badly due tiny state change large-scale action space characteristics. In order cope them, we first propose an auto-encoder disturb similarity adjacent states enhance features then divide whole decision process into two phases. DQN solve each phase, respectively, iterate find joint optimized solution. We implement our algorithm 6 scenarios different numbers user equipment (UE), redundant links, sub-carriers. Simulations results demonstrate that good convergence objective. Moreover, by further optimizing power allocation, 1–2 nines reliability improvement obtained bad conditions. Finally, experiment result shows reaches 8-nines common scenarios. method, proposed this paper takes only 0.32 s on average.