Deep Learning-Based Multi-Domain Framework for End-to-End Services in 5G Networks

Abstract

Over the past few years, network slicing has emerged as a pivotal component within the realm of 5G technology. It plays a critical role in effectively delineating network services based on a myriad of performance and operational requirements, all of which draw from a shared pool of common resources. The core objective of 5G technology is to facilitate simultaneous network slicing, thereby enabling the creation of multiple distinct end-to-end networks. This multiplicity of networks serves the paramount purpose of ensuring that the traffic within one network slice does not impede or adversely affect the traffic within another. Therefore, this paper proposes a Deep learning-based Multi Domain framework for end-to-end network slicing in traffic-aware prediction. The proposed method uses Deep Reinforcement Learning (DRL) for in-depth resource allocation analysis and improves the Quality of Service (QOS). The DRL-based Multi-domain framework provides traffic-aware prediction and enhances flexibility. The study results demonstrate that the suggested approach outperforms conventional, heuristic, and randomized methods and enhances resource use while maintaining QoS.