Topology and Energy Aware Approximate Algorithm for QoS-Based Resource Slicing in 5G Core Networks

This paper proposes a multi-objective optimization framework for network function virtualization (NFV) resource assignment in 5G core networks, targeting the minimization of both resource cost and server energy consumption subject to stringent quality of service (QoS) constraints for diverse services. This problem considers the assignment of CPU and bandwidth resources to servers and links, subject to network resource constraints. Given the dynamic nature of 5G core networks, where bandwidth and CPU processing demands fluctuate due to time-varying demand, user mobility, diverse QoS requirements, potential link failures, server outages, and the coexistence of multiple virtual networks (VNs), a dynamic resource assignment strategy is adopted in this work as opposed to static approaches in previous works. The majorization-minimization approach is used to handle the resulting mixed integer nonlinear programming problem, and an approximation-based algorithm for core network resource assignment (ACNRA) with an affordable computational complexity is proposed. The simulation results demonstrate the effectiveness of the proposed algorithm in reducing both network resource costs and server energy consumption. Statistical analysis of these results shows that our proposed method outperforms benchmark algorithms, achieving average reductions of 5.92% and 10.28% in resource cost, and 5.43% and 10.10% in energy consumption.