Multi-Service Network Slicing 5G NR Orchestration via Tailored HARQ Scheme Design and Hierarchical Resource Scheduling

In the radio access network domain, novel applications' requirements and network dynamics claim scalable processing capabilities, decentralized architecture, and on-demand resource allocation. Due to the intrinsic heterogeneous nature of the network resources, a fully context-aware fronthaul management is extremely complex to design. Operators have a high level-of-abstraction of the infrastructure's resources, which limits their decision-making capabilities. In this paper, a novel radio access network orchestrator is proposed, which jointly combines network function placement and resource allocation techniques in a single optimization framework. Given the infrastructure and users' delay constraints, we formulate a constrained linear maximization problem to determine the optimal timing parameters of the downlink hybrid automatic repeat scheme for each user. At inter-slice layer, our proposed resource allocation method is modelled as a Markov decision process, where the resources' sharing policy is tailored according to the real-time slice traffic load. At the intra-slice level, under mild conditions the radio resources are strategically prioritized, following a game theoretic model. Simulation results show that our method performs the hybrid automatic repeat request response 37%–88% compared to the classical LTE schemes. The dynamic allocation method at the inter-slice level improves the resource's utilization up to 50% compared to static network slicing. Moreover, at the intra-slice, the under and over resource provisioning has been decreased by 21.35% and 60.08%, respectively, compared to the classical fair resource allocation schemes.