pith. machine review for the scientific record. sign in

arxiv: 2603.28657 · v2 · submitted 2026-03-30 · 💻 cs.NI

Recognition: no theorem link

Performance Analysis of 5G RAN Slicing Deployment Options in Industry 4.0 Factories

Oscar Adamuz-Hinojosa , Abdelhilah Abdeselam , Pablo Mu\~noz , Pablo Ameigeiras , Juan M. Lopez-Soler
Authors on Pith no claims yet

Pith reviewed 2026-05-14 00:52 UTC · model grok-4.3

classification 💻 cs.NI
keywords 5G RAN slicingIndustry 4.0uRLLCstochastic network calculusdelay guaranteesradio resource utilizationproduction line isolationO-RAN
0
0 comments X

The pith

Only per-flow RAN slicing prevents delay violations for critical industrial flows when 5G radio resources are scarce.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper compares four RAN slicing options for 5G networks in Industry 4.0 factories that differ in how much they share slices and isolate traffic from different production lines or individual flows. It applies a stochastic network calculus model to compute delay bounds and resource needs for heterogeneous uRLLC traffic. Under tight radio resources, only the strictest per-flow isolation keeps all flows within their delay targets. Slice sharing and hybrid options free up capacity overall but leave the most sensitive flows exposed to violations. The slice planner runs quickly enough to fit inside non-real-time control loops.

Core claim

Within the considered RAN-level analytical model, the results show that, under resource scarcity, only per-flow slicing prevents delay violations, whereas slice-sharing and hybrid deployments improve aggregation efficiency at the cost of weaker protection for the most delay-critical flows.

What carries the argument

Four RAN slicing deployment options that vary the degree of slice sharing and per-line versus per-flow isolation, evaluated through stochastic network calculus arrival and service curves that bound worst-case delays and resource use.

If this is right

  • Per-flow slicing is the only option that guarantees no delay violations under resource scarcity.
  • Slice-sharing and hybrid approaches raise overall radio resource utilization.
  • Hybrid deployments trade some isolation for better efficiency but still risk violating the tightest deadlines.
  • The heuristic slice planner completes its work at non-real-time scales and can therefore run inside an O-RAN Non-RT RIC.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Factories with many delay-critical flows may have to accept lower overall efficiency to meet reliability targets.
  • The same isolation trade-offs likely appear in other 5G uRLLC settings that mix traffic from separate sites or tenants.
  • Dynamic adjustment of isolation level based on measured load could be tested as an extension of the planner.
  • Validation against real hardware traces would show how much the analytical bounds can be tightened.

Load-bearing premise

The stochastic network calculus arrival and service curves chosen for the heterogeneous industrial flows accurately capture worst-case delay behavior under the four isolation levels when multiple production lines share the same radio resources.

What would settle it

A measurement campaign in a 5G testbed with multiple production lines and deliberately scarce radio resources that records delay violations for critical flows even when per-flow slicing is applied.

Figures

Figures reproduced from arXiv: 2603.28657 by Abdelhilah Abdeselam, Juan M. Lopez-Soler, Oscar Adamuz-Hinojosa, Pablo Ameigeiras, Pablo Mu\~noz.

Figure 1
Figure 1. Figure 1: Network slicing deployment options in a 5G-based industrial network. We assume production lines in a factory connect wirelessly to the 5G system, [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Slice-level RB utilization and delay–utilization trade-off across [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Execution time of the network slice planner as a function of the number of flows |F| and production lines. |F|. As the same pool of N RB cell RBs is shared among more flows, the number of admissible RB reallocations or removals becomes more limited, causing the heuristic to converge in fewer iterations. Despite this faster convergence, the increase in per-iteration cost dominates, resulting in higher overa… view at source ↗
read the original abstract

This paper studies Radio Access Network (RAN) slicing strategies for 5G Industry~4.0 networks with ultra-reliable low-latency communication (uRLLC) requirements. We compare four RAN slicing deployment options that differ in slice sharing and in the degree of per-line or per-flow isolation. Unlike prior works that assume a fixed slicing structure, this work addresses how RAN slicing should be instantiated in the presence of multiple production lines and heterogeneous industrial flows. A Stochastic Network Calculus (SNC)-based analytical framework and a heuristic slice planner are used to evaluate per-flow delay guarantees and radio resource utilization. Within the considered RAN-level analytical model, the results show that, under resource scarcity, only per-flow slicing prevents delay violations, whereas slice-sharing and hybrid deployments improve aggregation efficiency at the cost of weaker protection for the most delay-critical flows. Execution-time results show that the proposed planner operates at non-real-time (Non-RT) time scales, supporting its implementation as an rApp within Open RAN (O-RAN) Non-RT RAN Intelligent Controller (RIC) control loops.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper analyzes four RAN slicing deployment options (per-flow, per-line, slice-sharing, and hybrid) for 5G networks supporting uRLLC traffic in Industry 4.0 factories with multiple production lines. It employs a Stochastic Network Calculus (SNC) analytical framework combined with a heuristic slice planner to derive per-flow delay guarantees and assess radio resource utilization. The central result is that, under resource scarcity, only per-flow slicing prevents delay violations, while sharing and hybrid options improve aggregation efficiency at the cost of weaker protection for the most delay-critical flows. The planner's execution time is shown to be compatible with Non-RT RIC control loops in O-RAN.

Significance. If the SNC-derived bounds are accurate, the work offers concrete guidance on isolation-vs-efficiency trade-offs for industrial 5G slicing, a topic of growing practical importance. The explicit comparison of four deployment options and the O-RAN integration angle are useful contributions; the heuristic planner's reported runtime also supports deployability claims.

major comments (3)
  1. [§3] §3 (SNC Framework): The paper invokes an external SNC framework but does not supply the explicit arrival curves for heterogeneous industrial flows or the service curves adapted to each isolation level (per-flow vs. per-line vs. shared). Without these definitions or the resulting delay-bound expressions, it is impossible to verify whether the reported ordering of slicing options follows from the model or from unstated assumptions about burst alignment and contention.
  2. [§5] §5 (Numerical Results): The claim that only per-flow slicing prevents violations rests on the analytical bounds, yet no simulation cross-validation, tightness plots, or sensitivity analysis to traffic parameters (e.g., burst sizes, arrival rates) is presented. This is load-bearing because the skeptic concern about omitted 5G effects (channel variation, scheduling jitter) could reverse the comparative ordering if the curves are not worst-case tight.
  3. [§4] §4 (Heuristic Planner): The planner is described as operating at Non-RT timescales, but the tuning parameters listed as free in the axiom ledger are not quantified, nor is any convergence or optimality gap analysis provided relative to an exact solver. This weakens the practical-implementation claim.
minor comments (2)
  1. Notation for the four slicing options is introduced inconsistently between the abstract and the body; a single table summarizing isolation degree, sharing level, and expected delay behavior would improve readability.
  2. Figure captions for delay-bound plots should explicitly state the traffic mix and resource-scarcity level used, rather than referring only to 'the considered scenario'.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments, which highlight opportunities to improve the clarity and completeness of our analysis. We address each major comment below and outline the revisions we will incorporate.

read point-by-point responses

  1. Referee: [§3] §3 (SNC Framework): The paper invokes an external SNC framework but does not supply the explicit arrival curves for heterogeneous industrial flows or the service curves adapted to each isolation level (per-flow vs. per-line vs. shared). Without these definitions or the resulting delay-bound expressions, it is impossible to verify whether the reported ordering of slicing options follows from the model or from unstated assumptions about burst alignment and contention.

    Authors: We agree that explicit definitions are required for verification. In the revised manuscript we will include the arrival curves for the heterogeneous industrial flows (parameterized directly from the burst sizes and rates of the Industry 4.0 use cases) together with the service curves for each isolation level and the resulting closed-form delay-bound expressions. The ordering of the options follows from the model: per-flow slicing supplies a dedicated service curve to every flow, eliminating intra-slice contention, while per-line, shared, and hybrid options aggregate flows and therefore produce strictly looser bounds for the most critical flows. All assumptions on burst alignment are stated in the model section and will be reiterated. revision: yes

  2. Referee: [§5] §5 (Numerical Results): The claim that only per-flow slicing prevents violations rests on the analytical bounds, yet no simulation cross-validation, tightness plots, or sensitivity analysis to traffic parameters (e.g., burst sizes, arrival rates) is presented. This is load-bearing because the skeptic concern about omitted 5G effects (channel variation, scheduling jitter) could reverse the comparative ordering if the curves are not worst-case tight.

    Authors: SNC supplies worst-case bounds by construction, which is the appropriate metric for uRLLC guarantees. We will add sensitivity plots varying burst sizes and arrival rates, as well as tightness plots that compare the analytical bounds against Monte-Carlo simulations of the same traffic model. Channel variation and scheduling jitter are not modeled in the current RAN-level abstraction; they can be folded into the service curve as additional variability, but they do not alter the relative ordering driven by isolation level. Full end-to-end 5G simulator validation lies outside the paper’s scope and will be noted as future work. revision: partial

  3. Referee: [§4] §4 (Heuristic Planner): The planner is described as operating at Non-RT timescales, but the tuning parameters listed as free in the axiom ledger are not quantified, nor is any convergence or optimality gap analysis provided relative to an exact solver. This weakens the practical-implementation claim.

    Authors: We will explicitly list the numerical values of all tuning parameters used by the heuristic. In addition, we will report the optimality gap obtained by comparing the heuristic against an exact ILP solver on small instances (up to 10 lines) and will include runtime scaling curves that confirm execution remains compatible with Non-RT RIC loops. These additions will be placed in a new subsection of §4. revision: yes

Circularity Check

0 steps flagged

No circularity: SNC model and heuristic planner derive claims independently from external framework

full rationale

The paper's derivation chain rests on an external Stochastic Network Calculus (SNC) framework for arrival/service curves and a new heuristic slice planner. Delay bounds and comparative results for the four isolation levels are computed directly from these model assumptions without any fitted parameters from the target dataset being renamed as predictions, without self-definitional loops, and without load-bearing self-citations that reduce the central claim to unverified prior work by the same authors. The abstract and results explicitly qualify the findings as holding 'within the considered RAN-level analytical model,' confirming the analysis is self-contained against the stated SNC assumptions rather than circular.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Only the abstract is available, so the ledger is necessarily incomplete. The SNC framework is treated as a standard tool whose arrival and service curves are chosen to match industrial traffic; the heuristic planner introduces at least one tunable parameter whose value is not reported.

free parameters (1)
  • heuristic planner tuning parameters
    The abstract mentions a heuristic slice planner but gives no explicit values or fitting procedure; these parameters control resource allocation decisions and therefore affect the reported efficiency-delay trade-offs.
axioms (1)
  • domain assumption Stochastic network calculus arrival and service curves can be defined for heterogeneous industrial flows that accurately bound worst-case delays under different isolation levels
    Invoked when the SNC framework is used to evaluate per-flow delay guarantees for the four slicing options.

pith-pipeline@v0.9.0 · 5516 in / 1447 out tokens · 41907 ms · 2026-05-14T00:52:22.424351+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

19 extracted references · 19 canonical work pages

  1. [1]

    The Future of Indus- trial Communication: Automation Networks in the Era of the Internet of Things and Industry 4.0,

    M. Wollschlaeger, T. Sauter, and J. Jasperneite, “The Future of Indus- trial Communication: Automation Networks in the Era of the Internet of Things and Industry 4.0,”IEEE Ind. Electron. Mag., vol. 11, no. 1, pp. 17–27, 2017

    work page 2017

  2. [2]

    An Overview of Physical Layer Design for Ultra-Reliable Low-Latency Communications in 3GPP Releases 15, 16, and 17,

    T.-K. Le, U. Salim, and F. Kaltenberger, “An Overview of Physical Layer Design for Ultra-Reliable Low-Latency Communications in 3GPP Releases 15, 16, and 17,”IEEE Access, vol. 9, pp. 433–444, 2021

    work page 2021

  3. [3]

    Potential-Game-Based 5G RAN Slice Planning for GBR Services,

    O. Adamuz-Hinojosa, P. Mu ˜noz, P. Ameigeiras, and J. M. Lopez-Soler, “Potential-Game-Based 5G RAN Slice Planning for GBR Services,” IEEE Access, vol. 11, pp. 4763–4780, 2023

    work page 2023

  4. [4]

    5G QoS for Industrial Automation,

    5G-ACIA, “5G QoS for Industrial Automation,” white paper, 5G Alliance for Connected Industries and Automation (5G-ACIA), Nov. 2021

    work page 2021

  5. [5]

    5G RAN Slicing to Support Reliability in Industrial Applications,

    M. M. Rashid, M. Carmen Lucas-Esta ˜n, M. Sepulcre, and J. Gozalvez, “5G RAN Slicing to Support Reliability in Industrial Applications,” in IEEE ETFA, pp. 1–4, 2022

    work page 2022

  6. [6]

    Automating 5G net- work slice management for industrial applications,

    A. Perdig ˜ao, J. Quevedo, and R. L. Aguiar, “Automating 5G net- work slice management for industrial applications,”Comput. Commun., vol. 229, p. 107991, 2025

    work page 2025

  7. [7]

    Dynamic RAN Slicing with Effective Isolation under Imperfect CSI,

    J. Zhang, Y . Zu, Y . Zhang, and B. Hou, “Dynamic RAN Slicing with Effective Isolation under Imperfect CSI,” inIEEE CCET, pp. 238–242, 2022

    work page 2022

  8. [8]

    Orchestrating 5G Network Slices to Support Industrial Internet and to Shape Next-Generation Smart Factories,

    T. Taleb, I. Afolabi, and M. Bagaa, “Orchestrating 5G Network Slices to Support Industrial Internet and to Shape Next-Generation Smart Factories,”IEEE Netw., vol. 33, no. 4, pp. 146–154, 2019

    work page 2019

  9. [9]

    Slicing enabled 5G experimentation platform for the Robotics vertical industry,

    S. S. Mhatre, K. Ramantas, R. Qiu, and C. Verikoukis, “Slicing enabled 5G experimentation platform for the Robotics vertical industry,” in IEEE MeditCom, pp. 123–126, 2021

    work page 2021

  10. [10]

    Latency- Sensitive 5G RAN Slicing for Deterministic Aperiodic Traffic in Smart Manufacturing,

    M. C. Lucas-Esta ˜n, J. Garc ´ıa-Morales, and J. Gozalvez, “Latency- Sensitive 5G RAN Slicing for Deterministic Aperiodic Traffic in Smart Manufacturing,” inIEEE FNWF, pp. 1–5, 2023

    work page 2023

  11. [11]

    Radio Access Network Slicing Strategies at Spec- trum Planning Level in 5G and Beyond,

    P. Mu ˜noz, O. Adamuz-Hinojosa, J. Navarro-Ortiz, O. Sallent, and J. P ´erez-Romero, “Radio Access Network Slicing Strategies at Spec- trum Planning Level in 5G and Beyond,”IEEE Access, vol. 8, pp. 79604–79618, 2020

    work page 2020

  12. [12]

    5G-CLARITY: 5G-Advanced Private Networks Integrating 5GNR, WiFi, and LiFi,

    T. Cogalanet al., “5G-CLARITY: 5G-Advanced Private Networks Integrating 5GNR, WiFi, and LiFi,”IEEE Commun. Mag., vol. 60, no. 2, pp. 73–79, 2022

    work page 2022

  13. [13]

    Integration of 5G with Time-Sensitive Networking for Industrial Communications

    5G-ACIA, “Integration of 5G with Time-Sensitive Networking for Industrial Communications.” White Paper, Feb. 2021

    work page 2021

  14. [14]

    Unleashing 5G Seamless Integration With TSN for Industry 5.0: Frame Forwarding and QoS Treatment,

    O. Adamuz-Hinojosa, F. Delgado-Ferro, J. Navarro-Ortiz, P. Mu ˜noz, and P. Ameigeiras, “Unleashing 5G Seamless Integration With TSN for Industry 5.0: Frame Forwarding and QoS Treatment,”IEEE Open J. Commun. Soc., vol. 6, pp. 4874–4884, 2025

    work page 2025

  15. [15]

    Survey of deterministic and stochastic service curve models in the network calculus,

    M. Fidler, “Survey of deterministic and stochastic service curve models in the network calculus,”IEEE Commun. Surv. Tutor., vol. 12, no. 1, pp. 59–86, 2010

    work page 2010

  16. [16]

    A Guide to the Stochastic Network Calculus,

    M. Fidler and A. Rizk, “A Guide to the Stochastic Network Calculus,” IEEE Commun. Surveys Tuts., vol. 17, no. 1, pp. 92–105, 2015

    work page 2015

  17. [17]

    A Stochastic Network Calculus (SNC)- Based Model for Planning B5G uRLLC RAN Slices,

    O. Adamuz-Hinojosa, V . Sciancalepore, P. Ameigeiras, J. M. Lopez- Soler, and X. Costa-P ´erez, “A Stochastic Network Calculus (SNC)- Based Model for Planning B5G uRLLC RAN Slices,”IEEE Trans. Wireless Commun., vol. 2, no. 2, pp. 1250 – 1265, 2022

    work page 2022

  18. [18]

    ORANUS: Latency-tailored Orchestration via Stochastic Network Calculus in 6G O-RAN,

    O. Adamuz-Hinojosa, L. Zanzi, V . Sciancalepore, A. Garcia-Saavedra, and X. Costa-P ´erez, “ORANUS: Latency-tailored Orchestration via Stochastic Network Calculus in 6G O-RAN,”IEEE INFOCOM, 2024

    work page 2024

  19. [19]

    Under- standing O-RAN: Architecture, Interfaces, Algorithms, Security, and Research Challenges,

    M. Polese, L. Bonati, S. D’Oro, S. Basagni, and T. Melodia, “Under- standing O-RAN: Architecture, Interfaces, Algorithms, Security, and Research Challenges,”IEEE Commun. Surv. Tutor., pp. 1–1, 2023

    work page 2023