Resource Allocation for Hierarchical Service Assurance in NOMA-Assisted Network Slicing with Finite Blocklength Communication

As services diversify, providing hierarchical assurance for various service types becomes crucial. This paper proposes an optimization algorithm for resource allocation in NOMA-assisted URLLC and mMTC network slicing, combining deep reinforcement learning with graph theory-based matching to minimize the overall network decoding error rate. URLLC slices require high reliability and low latency, while mMTC slices prioritize massive device connectivity. NOMA technology allows users to share the same channel resources, boosting system capacity and spectrum efficiency. Simulation results show that the proposed algorithm outperforms DDQN and DQN in both convergence speed and value, while exchange matching achieves lower complexity with comparable performance to exhaustive matching.