Resource allocation and NFV placement in resource constrained MEC-enabled 5G-Networks

The fifth-generation (5G) of mobile communication networks are expected to support a large number of vertical industries requiring services with diverging requirements. To accommodate this, mobile networks are undergoing a significant transformation to enable a variety of services to coexist on the same infrastructure through network slicing. Additionally, the introduction of distributed user-plane and multi-access edge computing (MEC) technology allows the deployment of virtualised applications close to the network edge. The first part of this dissertation focuses on end-to-end network slice provisioning for various vertical industries with different service requirements. Two slice provisioning strategies are explored, by formulating a mixed integer linear programming (MILP) problem. Further, a genetic algorithm (GA)-based approach is proposed with the aim to improve search-space exploration. Simulation results show that the proposed approach is effective in providing near-optimal solutions while drastically reducing computational complexity. In a later stage, the study focuses on building a measurement-based digital twin (DT) for the highly heterogeneous MEC ecosystem. The DT operates as an intermediate and collaborative layer, enabling the orchestration layer to better understand network behavior before making changes to the physical network. Assisted by proper AI/ML solutions, the DT is envisioned to play a crucial role in automated network management. The study utilizes an emulated and physical test-bed to gather network key performance indicators (KPIs) and demonstrates the potential of graph neural network (GNN) in enabling closed loop automation with the help of DT. These findings offer a foundation for future research in the area of DT models and carbon footprint-aware orchestration.

Grantor dc:publisher

Università degli studi di Trento

Year dc:date

2023

Author dc:creator

• Fedrizzi, Riccardo

Contributors dc:contributor

• Granelli, Fabrizio

• 5G, MEC, Network Slicing, VNE, Resource Allocation, Optimization, Machine Learning, Network Emulation

Statement dc:rights

• info:eu-repo/semantics/openAccess
• license:Creative commons
• license uri:http://creativecommons.org/licenses/by-nc-nd/4.0/

Language dc:language

eng

Identifier

http://dx.doi.org/10.15168/11572_381269
10.15168/11572_381269

OAI identifier oai:identifier

oai:iris.unitn.it:11572/381269