Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Fast and Scalable Network Slicing by Integrating Deep Learning with Lagrangian Methods

Tianlun Hu, Qi Liao, Qiang Liu, Antonio Massaro, Georg Carle

TL;DR

This work tackles dynamic radio resource allocation for network slicing under inter-slice constraints by marrying constrained optimization with deep learning. It introduces IDLA, a neural-assisted framework that learns a general per-slice QoS estimator and solves a decomposed, per-cell Lagrangian optimization via automatic differentiation and multi-start gradient search. Empirical results from a system-level simulator show near-optimal QoS satisfaction and strong generalization across varying slice configurations, with superior scalability compared to DRL baselines. The approach offers a practical pathway to fast, adaptable, and scalable network slicing in 5G-and-beyond deployments, capable of handling large networks and evolving slice mixes.

Abstract

Network slicing is a key technique in 5G and beyond for efficiently supporting diverse services. Many network slicing solutions rely on deep learning to manage complex and high-dimensional resource allocation problems. However, deep learning models suffer limited generalization and adaptability to dynamic slicing configurations. In this paper, we propose a novel framework that integrates constrained optimization methods and deep learning models, resulting in strong generalization and superior approximation capability. Based on the proposed framework, we design a new neural-assisted algorithm to allocate radio resources to slices to maximize the network utility under inter-slice resource constraints. The algorithm exhibits high scalability, accommodating varying numbers of slices and slice configurations with ease. We implement the proposed solution in a system-level network simulator and evaluate its performance extensively by comparing it to state-of-the-art solutions including deep reinforcement learning approaches. The numerical results show that our solution obtains near-optimal quality-of-service satisfaction and promising generalization performance under different network slicing scenarios.

Fast and Scalable Network Slicing by Integrating Deep Learning with Lagrangian Methods

TL;DR

This work tackles dynamic radio resource allocation for network slicing under inter-slice constraints by marrying constrained optimization with deep learning. It introduces IDLA, a neural-assisted framework that learns a general per-slice QoS estimator and solves a decomposed, per-cell Lagrangian optimization via automatic differentiation and multi-start gradient search. Empirical results from a system-level simulator show near-optimal QoS satisfaction and strong generalization across varying slice configurations, with superior scalability compared to DRL baselines. The approach offers a practical pathway to fast, adaptable, and scalable network slicing in 5G-and-beyond deployments, capable of handling large networks and evolving slice mixes.

Abstract

Network slicing is a key technique in 5G and beyond for efficiently supporting diverse services. Many network slicing solutions rely on deep learning to manage complex and high-dimensional resource allocation problems. However, deep learning models suffer limited generalization and adaptability to dynamic slicing configurations. In this paper, we propose a novel framework that integrates constrained optimization methods and deep learning models, resulting in strong generalization and superior approximation capability. Based on the proposed framework, we design a new neural-assisted algorithm to allocate radio resources to slices to maximize the network utility under inter-slice resource constraints. The algorithm exhibits high scalability, accommodating varying numbers of slices and slice configurations with ease. We implement the proposed solution in a system-level network simulator and evaluate its performance extensively by comparing it to state-of-the-art solutions including deep reinforcement learning approaches. The numerical results show that our solution obtains near-optimal quality-of-service satisfaction and promising generalization performance under different network slicing scenarios.
Paper Structure (12 sections, 14 equations, 5 figures, 1 algorithm)

This paper contains 12 sections, 14 equations, 5 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. System Model
  3. Problem Formulation
  4. Proposed Solutions
  5. Slice-based QoS Prediction using DNN
  6. Lagrangian Method for Slicing Resource Partitioning
  7. Efficient Implementation to Improve the Performance of Lagrange Multiplier Method
  8. Performance Evaluation
  9. Network Setting
  10. Sample Collection and Network QoS Estimator Training
  11. Performance Comparison
  12. Conclusion

Figures (5)

  • Figure 1: Dynamic slicing resource partition
  • Figure 2: Traffic mask to imitate the dynamic slice traffic
  • Figure 3: Network QoS estimator MAE histogram
  • Figure 4: Comparison of average user throughput among schemes
  • Figure 5: Comparison of network utility

Theorems & Definitions (2)

  • Remark 1
  • Remark 2