URLLC and eMBB Service Multiplexing in a CoMP-Enhanced RAN with Imperfect Channel State Information

By providing macro-diversity, coordinated multipoint (CoMP) communications is promising for achieving ultra-reliable low-latency communications (URLLC). Two significant issues are considered in this paper: imperfect acquisition of channel state information (CSI) and coexistence with enhanced mobile broadband (eMBB) services. The former arises from non-negligible channel estimation delays and/or channel estimation errors, while the latter stems from the conflicting requirements of different services sharing network resources. To address these challenges, we explore a resource puncturing scheme for URLLC when the CoMP system provides URLLC and eMBB services simultaneously. Specifically, we propose a pilot-based symbol allocation framework that maximizes the number of symbols for eMBB services after puncturing, subject to constraints on the URLLC block error rate (BLER) and delay. Finally, we develop a constrained deep reinforcement learning (DRL) algorithm to obtain optimal policy for URLLC services. Simulation results demonstrate that the proposed DRL algorithm can converge quickly. Compared to two existing benchmarks, our proposed algorithm reduces the number of symbols punctured by the URLLC service while ensuring its BLER outage probability.