Reliable Two-Timescale Scheduling in a Multi-User Downlink Channel with Hard Deadlines

Ultra-Reliable Low-Latency Communications (URLLC) is an important part of emerging 5G and 6G networks which enables mission-critical applications like autonomous driving. These novel applications depend on the error-free delivery of short messages before an application-specific deadline, which is challenging in a fast-changing environment. In this work, we consider a wireless fading downlink channel shared for the transmission of periodically arriving messages for multiple mobile units (MUs). The message sizes, deadlines and the period of message arrival are MU-specific. The message for a MU can be split into smaller data packets, so that multiple unreliable transmissions can be combined to achieve a reliable transmission. We formulate an infinite time horizon Markov Decision Process (MDP) for the average timely throughput, and show that the MDP is periodic. We propose a novel two-timescale scheduling solution, which incorporates the uncertainty of the channel in an inter-frame problem and errors caused by short-packet coding in an intraframe problem. Through numerical simulations, we show that the proposed approach outperforms State-of-the-Art scheduling algorithms in terms of timely throughput.