Energy-Efficient User Scheduling and Resource Allocation for NOMA-Enabled URLLC in Industrial IoT Systems

Abstract

B5G/6G wireless systems are driving dense uplink access in Industrial Internet of Things (IIoT) networks, where massive sensors need to transmit short packets under stringent ultra-reliable and low-latency communication (URLLC) requirements and limited spectrum resources. This paper proposes a Lyapunov-guided joint user scheduling and resource allocation (LG-JSRA) mechanism for uplink NOMA transmission with URLLC traffic in IIoT systems. We formulate a long-term average energy minimization problem for uplink short-packet communications, in which user scheduling, blocklength allocation, and power allocation are jointly optimized under stochastic queue evolution, subject to long-term delay violation constraints. By leveraging Lyapunov optimization, the stochastic problem is transformed into a deterministic per-slot mixed-integer nonconvex problem, which is decomposed into user scheduling and resource allocation subproblems solved by branch-and-bound (BnB) and Transformer-enhanced TD3, respectively. Simulation results show that LG-JSRA achieves lower average uplink energy consumption than the benchmark schemes.