Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Optimum Network Slicing for Ultra-reliable Low Latency Communication (URLLC) Services in Campus Networks

Iulisloi Zacarias, Francisco Carpio, André Costa Drummond, Admela Jukan

TL;DR

The paper addresses the challenge of delivering ultra-reliable low-latency communications (URLLC) in campus networks that reuse public 5G infrastructure. It proposes a MILP framework for vertical RAN slicing across a multi-tier edge-to-core topology, with isolated RU/DU/CU NF chains and end-to-end routing that originates at a known edge source but terminates at a CU placed dynamically along the path. A key novelty is jointly accounting for processing, queuing, and propagation delays, and allowing dynamic UPF placement to meet URLLC requirements while maintaining slice isolation. Results on a realistic 51-node, 70-link topology show the MILP can find optimal placements and that edge-focused deployment helps satisfy URLLC budgets, whereas a heuristic may falter at moderate to high slice counts; these insights guide practical campus-network designs and resource provisioning for URLLC. Future work points to stronger slice isolation, NF replication, and shared CU/DU NF scenarios to further enhance reliability and scalability.

Abstract

Within 3GPP, the campus network architecture has evolved as a deployment option for industries and can be provisioned using network slicing over already installed 5G public network infrastructure. In campus networks, the ultra-reliable low latency communication (URLLC) service category is of major interest for applications with strict latency and high-reliability requirements. One way to achieve high reliability in a shared infrastructure is through resource isolation, whereby network slicing can be optimized to adequately reserve computation and transmission capacity. This paper proposes an approach for vertical slicing the radio access network (RAN) to enable the deployment of multiple and isolated campus networks to accommodate URLLC services. To this end, we model RAN function placement as a mixed integer linear programming problem with URLLC-related constraints. We demonstrate that our approach can find optimal solutions in real-world scenarios. Furthermore, unlike existing solutions, our model considers the user traffic flow from a known source node on the network's edge to an unknown \textit{a priori} destination node. This flexibility could be explored in industrial campus networks by allowing dynamic placement of user plane functions (UPFs) to serve the URLLC.

Optimum Network Slicing for Ultra-reliable Low Latency Communication (URLLC) Services in Campus Networks

TL;DR

The paper addresses the challenge of delivering ultra-reliable low-latency communications (URLLC) in campus networks that reuse public 5G infrastructure. It proposes a MILP framework for vertical RAN slicing across a multi-tier edge-to-core topology, with isolated RU/DU/CU NF chains and end-to-end routing that originates at a known edge source but terminates at a CU placed dynamically along the path. A key novelty is jointly accounting for processing, queuing, and propagation delays, and allowing dynamic UPF placement to meet URLLC requirements while maintaining slice isolation. Results on a realistic 51-node, 70-link topology show the MILP can find optimal placements and that edge-focused deployment helps satisfy URLLC budgets, whereas a heuristic may falter at moderate to high slice counts; these insights guide practical campus-network designs and resource provisioning for URLLC. Future work points to stronger slice isolation, NF replication, and shared CU/DU NF scenarios to further enhance reliability and scalability.

Abstract

Within 3GPP, the campus network architecture has evolved as a deployment option for industries and can be provisioned using network slicing over already installed 5G public network infrastructure. In campus networks, the ultra-reliable low latency communication (URLLC) service category is of major interest for applications with strict latency and high-reliability requirements. One way to achieve high reliability in a shared infrastructure is through resource isolation, whereby network slicing can be optimized to adequately reserve computation and transmission capacity. This paper proposes an approach for vertical slicing the radio access network (RAN) to enable the deployment of multiple and isolated campus networks to accommodate URLLC services. To this end, we model RAN function placement as a mixed integer linear programming problem with URLLC-related constraints. We demonstrate that our approach can find optimal solutions in real-world scenarios. Furthermore, unlike existing solutions, our model considers the user traffic flow from a known source node on the network's edge to an unknown \textit{a priori} destination node. This flexibility could be explored in industrial campus networks by allowing dynamic placement of user plane functions (UPFs) to serve the URLLC.

Paper Structure

This paper contains 11 sections, 15 equations, 8 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Campus Network Slicing
  4. Reference Network Scenario
  5. System Model
  6. Objective function
  7. Constraints
  8. First-fit algorithm (heuristics)
  9. Evaluation Scenario
  10. Results and Verification
  11. Conclusions and Future Work

Figures (8)

  • Figure 1: Deployment scenario of a public network integrated campus (non-public) network in a factory plant composed of several production lines connected by multiple base stations. Adapted from 5g-acia-npn.
  • Figure 2: Three examples of virtual RAN allocation in a multi-tier network.
  • Figure 3: Network topology used in the evaluation.
  • Figure 4: Average server utilization by RAN NFs.
  • Figure 5: Link utilization.
  • ...and 3 more figures