Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Constrained Network Slice Assignment via Large Language Models

Sagar Sudhakara, Pankaj Rajak

TL;DR

The paper investigates leveraging LLMs to improve 5G network slice resource allocation under capacity and latency constraints. It contrasts zero-shot LLM-based slice assignment with a hybrid approach where LLM-derived request similarity informs an ILP solved by GLPK, enforcing feasibility. Findings show that while LLMs can produce reasonable drafts, combining semantic clustering with exact optimization yields 100% complete and balanced allocations, offering a scalable AI-assisted workflow for network slicing. This hybrid strategy reduces search space and manual tuning, enabling efficient, constraint-satisfying resource distribution in beyond-5G networks.

Abstract

Modern networks support network slicing, which partitions physical infrastructure into virtual slices tailored to different service requirements (for example, high bandwidth or low latency). Optimally allocating users to slices is a constrained optimization problem that traditionally requires complex algorithms. In this paper, we explore the use of Large Language Models (LLMs) to tackle radio resource allocation for network slicing. We focus on two approaches: (1) using an LLM in a zero-shot setting to directly assign user service requests to slices, and (2) formulating an integer programming model where the LLM provides semantic insight by estimating similarity between requests. Our experiments show that an LLM, even with zero-shot prompting, can produce a reasonable first draft of slice assignments, although it may violate some capacity or latency constraints. We then incorporate the LLM's understanding of service requirements into an optimization solver to generate an improved allocation. The results demonstrate that LLM-guided grouping of requests, based on minimal textual input, achieves performance comparable to traditional methods that use detailed numerical data, in terms of resource utilization and slice isolation. While the LLM alone does not perfectly satisfy all constraints, it significantly reduces the search space and, when combined with exact solvers, provides a promising approach for efficient 5G network slicing resource allocation.

Constrained Network Slice Assignment via Large Language Models

TL;DR

The paper investigates leveraging LLMs to improve 5G network slice resource allocation under capacity and latency constraints. It contrasts zero-shot LLM-based slice assignment with a hybrid approach where LLM-derived request similarity informs an ILP solved by GLPK, enforcing feasibility. Findings show that while LLMs can produce reasonable drafts, combining semantic clustering with exact optimization yields 100% complete and balanced allocations, offering a scalable AI-assisted workflow for network slicing. This hybrid strategy reduces search space and manual tuning, enabling efficient, constraint-satisfying resource distribution in beyond-5G networks.

Abstract

Modern networks support network slicing, which partitions physical infrastructure into virtual slices tailored to different service requirements (for example, high bandwidth or low latency). Optimally allocating users to slices is a constrained optimization problem that traditionally requires complex algorithms. In this paper, we explore the use of Large Language Models (LLMs) to tackle radio resource allocation for network slicing. We focus on two approaches: (1) using an LLM in a zero-shot setting to directly assign user service requests to slices, and (2) formulating an integer programming model where the LLM provides semantic insight by estimating similarity between requests. Our experiments show that an LLM, even with zero-shot prompting, can produce a reasonable first draft of slice assignments, although it may violate some capacity or latency constraints. We then incorporate the LLM's understanding of service requirements into an optimization solver to generate an improved allocation. The results demonstrate that LLM-guided grouping of requests, based on minimal textual input, achieves performance comparable to traditional methods that use detailed numerical data, in terms of resource utilization and slice isolation. While the LLM alone does not perfectly satisfy all constraints, it significantly reduces the search space and, when combined with exact solvers, provides a promising approach for efficient 5G network slicing resource allocation.

Paper Structure

This paper contains 15 sections, 1 equation, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Network Slicing Optimization.
  4. LLMs and Foundation Models in Networks.
  5. LLMs for Structured Decision-Making.
  6. Methodology
  7. Problem Formulation
  8. LLM-Based Zero-Shot Assignment
  9. LLM-Guided Integer Programming
  10. Experiments
  11. Synthetic Dataset and Experimental Setup
  12. Results and Analysis
  13. Conclusion
  14. Prompt Design
  15. Integer Programming Formulation

Figures (2)

  • Figure 1: Bandwidth and Density Utilization for Claude variants.
  • Figure 2: Performance metrics for the ILP+LLM allocation method. Left: Completeness and homogeneity scores across 10 runs. Center: Distribution of bandwidth utilization across Slice A, Slice B, and Slice C. Right: Distribution of density utilization across Slice A, Slice B, and Slice C.