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

• Published in: IEEE transactions on vehicular technology Vol. 72; no. 4; pp. 5021 - 5034
• Main Authors: Maule, Massimiliano, Vardakas, John S., Verikoukis, Christos
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.04.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: 5G mobile communication; 5G NR; Automatic repeat request; Constraints; Decision making; Delays; functional split; game theory; HARQ timing; Infrastructure; Long Term Evolution; Markov processes; network design optimization; Network slicing; Optimization; Provisioning; Real-time systems; Resource allocation; Resource management; Resource scheduling; Ultra reliable low latency communication
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2022.3223252

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.