DRL-Leveraged and RIS-Assisted Hybrid Network Slicing for eMBB and URLLC Co-existence in 6G Systems

In this paper, we design a Reconfigurable Intelligent Surface (RIS)-assisted down-link cellular network for the co-existence of enhanced Mobile Broadband (eMBB) and Ultra Reliable Low Latency Communication (URLLC) services in 6G. A novel hybrid slicing technique comprising Non-orthogonal Multiple Access (NOMA) and puncturing is proposed to schedule the URLLC transmission on top of the already scheduled eMBB traffic. We propose a slot-based eMBB scheduling to schedule the eMBB flows at the beginning of the slot, and a mini-slot-based URLLC scheduling to insert the mini-slots on top of the scheduled eMBB traffic. The overall objective is to maximize the URLLC packets admission rate and minimize the eMBB rate loss while satisfying the Quality of Service (QoS) demands of both eMBB and URLLC services. The eMBB allocation problem is solved using Alternating Optimization (AO) by optimizing the RIS phase shift and eMBB power. The URLLC allocation is achieved using a Deep Reinforcement Learning (DRL)-based Proximal Policy Optimization (PPO) algorithm. The performance of the proposed technique with hybrid network slicing is compared with the NOMA-based and puncturing-based slicing, and other state-of-the-art (SOTA) techniques for URLLC allocation. The comparison results show that the proposed hybrid network slicing outperforms these techniques. Specifically, it is observed that with the worst-case URLLC load, the proposed solution provides 5.31% and 144% improvements in the URLLC packet admission rate and eMBB sum rate, respectively, compared to an optimization-based SOTA URLLC allocation technique. Similarly, there is an improvement of 78.36% in execution time compared to the SOTA URLLC allocation technique based on the heuristic algorithm.