pith. sign in

arxiv: 2605.30622 · v1 · pith:CDVCYQQZnew · submitted 2026-05-28 · 💻 cs.NI · cs.LG

Jamming-Resilient PRB Reservation for Latency-Critical O-RAN Network Slicing

Elahe Delavari , Junaid Farooq This is my paper

Pith reviewed 2026-06-29 00:05 UTC · model grok-4.3

classification 💻 cs.NI cs.LG
keywords O-RANnetwork slicingjamming resiliencePRB reservationDeep Q-NetworkURLLCxApplatency-critical
0
0 comments X

The pith

A masked Deep Q-Network learns when to activate reserved PRBs for hybrid mitigation of jamming in O-RAN slices.

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

The paper establishes a reserve-based resilience framework for PRB allocation in O-RAN network slicing that counters adversarial jamming. A near-RT RIC xApp manages a finite pool of reserved PRBs to clear backlog proactively and supply extra capacity reactively during jammer activity. Reserve activation is cast as a constrained sequential decision problem solved by a masked Deep Q-Network trained under non-stationary jamming. Simulations show fewer URLLC latency violations and higher reserve efficiency than reactive baselines. This matters for industrial 5G systems where jamming at cell edges can trigger persistent queue buildup and deadline misses.

Core claim

The central claim is that a hybrid proactive-reactive strategy for activating a pool of reserved PRBs, learned via a masked Deep Q-Network, delivers substantial reductions in URLLC latency violations and improved reserve efficiency under jamming compared with reactive baselines in sliced O-RAN deployments.

What carries the argument

The masked Deep Q-Network that learns reserve activation policies as a constrained sequential decision problem under non-stationary jamming.

If this is right

  • URLLC slices maintain lower latency violation rates even when effective PRB capacity drops abruptly.
  • Reserved capacity is allocated only during active jamming intervals, reducing waste outside attack periods.
  • Proactive backlog clearing builds latency margin before jamming begins.
  • The learned policy adapts to changing jamming patterns without explicit reprogramming.

Where Pith is reading between the lines

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

  • The same reserve-pool approach could extend to other sudden capacity threats such as strong interference or rapid user mobility.
  • Coordinating multiple xApps on the RIC could allow joint reserve decisions across slices with differing latency targets.
  • Online fine-tuning of the DQN after initial simulation training might be needed if real jamming statistics differ from the modeled process.

Load-bearing premise

The simulation environment and jamming model are representative enough that policies learned in simulation will transfer to real O-RAN deployments without additional real-world tuning or validation.

What would settle it

Running the trained masked DQN policy on a physical O-RAN testbed with actual jamming signals and measuring whether latency violation rates drop by the margins reported in simulation.

Figures

Figures reproduced from arXiv: 2605.30622 by Elahe Delavari, Junaid Farooq.

Figure 1
Figure 1. Figure 1: System overview of the resilient PRB allocator. The [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Training convergence of the DQN-based xApp. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Fixed-severity periodic on–off jamming sweep. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Reserve-efficiency under fixed-severity jamming. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Relative URLLC latency impact per-policy compared [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: URLLC latency during jammer-ON under time [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
read the original abstract

Open radio access network (O-RAN) architectures enable near real-time, software-driven control of network slicing through programmable xApps deployed on the near-real-time RAN Intelligent Controller (near-RT RIC). In industrial 5G downlink systems, adversarial jamming can abruptly reduce the effective physical resource block (PRB) capacity, triggering queue buildup and persistent latency violations, particularly in the presence of low spectral efficiency cell edge user equipments. This paper proposes a reserve-based resilience framework for PRB allocation in sliced O-RAN deployments. A finite pool of reserved PRBs is controlled by a near-RT RIC xApp that provides hybrid mitigation by proactively clearing backlog to build latency margin and reactively allocating reserve capacity during jammer active intervals. We formulate reserve activation as a constrained sequential decision problem and design a masked Deep Q-Network to learn effective control policies under non-stationary jamming. Simulation results show substantial reductions in URLLC latency violations and improved reserve efficiency compared to reactive baselines.

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

2 major / 0 minor

Summary. The manuscript proposes a reserve-based resilience framework for PRB allocation in O-RAN network slicing under adversarial jamming. It formulates reserve activation as a constrained sequential decision problem solved by a masked Deep Q-Network xApp on the near-RT RIC that combines proactive backlog clearing and reactive reserve allocation. The central claim, supported by simulation results, is that the learned policy yields substantial reductions in URLLC latency violations and improved reserve efficiency relative to reactive baselines.

Significance. If the simulation outcomes prove robust and the policies transfer beyond the modeled environment, the framework would constitute a practical contribution to jamming-resilient industrial 5G slicing by exploiting O-RAN programmability. It targets a concrete operational vulnerability (abrupt PRB capacity drops causing persistent latency violations at cell-edge UEs) with a hybrid proactive-reactive mechanism.

major comments (2)
  1. [Abstract] Abstract: the claim of 'substantial reductions in URLLC latency violations and improved reserve efficiency' is presented without any quantitative description of simulation parameters, baseline implementations, statistical tests, sensitivity analyses, or the jamming process (duty cycle, power, frequency selectivity, activation intervals). This omission is load-bearing because the soundness of the central claim rests entirely on these simulation outcomes.
  2. [Simulation results (inferred from abstract and skeptic note)] The policy is obtained by training a masked DQN on an environment model whose discrete-time step size, PRB capacity drop dynamics, queue evolution, cell-edge spectral efficiency, and relation to O-RAN E2 latency are not specified or validated against real near-RT RIC control loops and 5G PHY behavior. Without such grounding, it is impossible to determine whether performance gains reflect robust features or simulator artifacts.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. We address each major comment below and will revise the manuscript to enhance quantitative support in the abstract and to provide fuller specification of the simulation environment.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim of 'substantial reductions in URLLC latency violations and improved reserve efficiency' is presented without any quantitative description of simulation parameters, baseline implementations, statistical tests, sensitivity analyses, or the jamming process (duty cycle, power, frequency selectivity, activation intervals). This omission is load-bearing because the soundness of the central claim rests entirely on these simulation outcomes.

    Authors: We agree that the abstract would be strengthened by quantitative anchors. In the revision we will incorporate specific simulation outcomes (e.g., percentage reductions in latency violations, reserve-utilization gains) together with concise references to the jamming duty cycle, power levels, and baseline definitions while remaining within length limits. The body already contains the full parameter tables and statistical details; the abstract update will simply surface the key numbers that support the central claim. revision: yes

  2. Referee: [Simulation results (inferred from abstract and skeptic note)] The policy is obtained by training a masked DQN on an environment model whose discrete-time step size, PRB capacity drop dynamics, queue evolution, cell-edge spectral efficiency, and relation to O-RAN E2 latency are not specified or validated against real near-RT RIC control loops and 5G PHY behavior. Without such grounding, it is impossible to determine whether performance gains reflect robust features or simulator artifacts.

    Authors: We will add an explicit subsection that lists the discrete-time step size, the exact PRB-capacity-drop model (including frequency-selective and abrupt-drop cases), the queue-evolution equations, the cell-edge spectral-efficiency values employed, and the mapping of control actions to O-RAN E2 latency budgets. The study remains simulation-based; we will therefore also include a short discussion of how the chosen parameters align with 3GPP NR and O-RAN timing specifications, thereby clarifying the intended scope and reducing the risk that readers interpret results as ungrounded artifacts. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper formulates PRB reserve activation as a constrained sequential decision problem and trains a masked DQN policy on a simulated environment under non-stationary jamming, then reports empirical latency and efficiency gains versus baselines. No algebraic derivation chain is claimed that reduces a result to its own inputs by construction; the performance claims are simulation outputs rather than predictions forced by fitted parameters or self-citations. No self-definitional steps, uniqueness theorems, or ansatz smuggling appear in the abstract or described approach. The work is self-contained as a simulation study with independent content from the learned policy.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no explicit free parameters, axioms, or invented entities are stated in the provided text.

pith-pipeline@v0.9.1-grok · 5701 in / 939 out tokens · 30907 ms · 2026-06-29T00:05:39.087717+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

18 extracted references

  1. [1]

    Field trial of network slicing in 5G and PON-enabled industrial networks,

    Y . Luo, M. Jiang, D. Zhang, and F. Effenberger, “Field trial of network slicing in 5G and PON-enabled industrial networks,”IEEE Wireless Communications, vol. 30, no. 1, pp. 78–85, 2023

    2023

  2. [2]

    Un- derstanding O-RAN: Architecture, interfaces, algorithms, security, and research challenges,

    M. Polese, L. Bonati, S. D’oro, S. Basagni, and T. Melodia, “Un- derstanding O-RAN: Architecture, interfaces, algorithms, security, and research challenges,”IEEE Communications Surveys & Tutorials, vol. 25, no. 2, pp. 1376–1411, 2023

    2023

  3. [3]

    RIC: A RAN intelligent controller platform for AI-enabled cellular networks,

    B. Balasubramanian, E. S. Daniels, M. Hiltunen, R. Jana, K. Joshi, R. Sivaraj, T. X. Tran, and C. Wang, “RIC: A RAN intelligent controller platform for AI-enabled cellular networks,”IEEE Internet Computing, vol. 25, no. 2, pp. 7–17, 2021

    2021

  4. [4]

    Demon- strating jamming mitigation in O-RAN via AI enabled intrusion detection and secure slicing xApps,

    J. Moore, A. S. Abdalla, C. Ueltschey, and V . Marojevic, “Demon- strating jamming mitigation in O-RAN via AI enabled intrusion detection and secure slicing xApps,” inIEEE Military Communications Conference (MILCOM), Los Angeles, CA, USA, Oct. 2025

    2025

  5. [5]

    Anti-Jamming RIS Communications Using DQN-Based Algorithm,

    P. D. Thanh, H. T. H. Giang, and I.-P. Hong, “Anti-Jamming RIS Communications Using DQN-Based Algorithm,”IEEE Access, vol. 10, pp. 28422–28433, 2022

    2022

  6. [6]

    Mitigating jamming attack in 5g heterogeneous networks: A federated deep reinforce- ment learning approach,

    H. Sharma, N. Kumar, and R. Tekchandani, “Mitigating jamming attack in 5g heterogeneous networks: A federated deep reinforce- ment learning approach,”IEEE Transactions on Vehicular Technology, vol. 72, no. 2, pp. 2439–2452, 2022

    2022

  7. [7]

    Anti-Jamming Task Schedul- ing in MEC-O-RAN with Hierarchical DRL and Transformer-Based Control,

    G. Asemian, M. Amini, and B. Kantarci, “Anti-Jamming Task Schedul- ing in MEC-O-RAN with Hierarchical DRL and Transformer-Based Control,”IEEE Internet of Things Journal, pp. 1–1, 2025

    2025

  8. [8]

    Federated Deep Reinforce- ment Learning for Efficient Jamming Attack Mitigation in O-RAN,

    Z. A. E. Houda, H. Moudoud, and B. Brik, “Federated Deep Reinforce- ment Learning for Efficient Jamming Attack Mitigation in O-RAN,” IEEE Transactions on Vehicular Technology, vol. 73, pp. 9334–9343, July 2024

    2024

  9. [9]

    Dynamic spectrum reservation for CR networks in the presence of channel failures: Channel allocation and reliability analysis,

    I. A. Balapuwaduge, F. Li, and V . Pla, “Dynamic spectrum reservation for CR networks in the presence of channel failures: Channel allocation and reliability analysis,”IEEE Transactions on Wireless Communica- tions, vol. 17, no. 2, pp. 882–898, 2018

    2018

  10. [10]

    Centralized dynamic channel reservation mechanism via SDN for CR networks spectrum allocation,

    N. M. El Azaly, E. F. Badran, H. N. Kheirallah, and H. H. Farag, “Centralized dynamic channel reservation mechanism via SDN for CR networks spectrum allocation,”IEEE Access, vol. 8, pp. 192493– 192505, 2020

    2020

  11. [11]

    Spectrum efficiency in CRNs using hybrid dynamic channel reservation and enhanced dynamic spectrum access,

    G. Abbas, Z. H. Abbas, T. Baker, M. Waqas,et al., “Spectrum efficiency in CRNs using hybrid dynamic channel reservation and enhanced dynamic spectrum access,”Ad Hoc Networks, vol. 107, p. 102246, 2020

    2020

  12. [12]

    An enhanced spectrum reservation framework for heterogeneous users in CR-Enabled IoT networks,

    A. U. Khan, M. Tanveer, W. U. Khan, J. Nebhen, X. Li, M. Zeng, and O. A. Dobre, “An enhanced spectrum reservation framework for heterogeneous users in CR-Enabled IoT networks,”IEEE Wireless Communications Letters, 2021

    2021

  13. [13]

    A novel reservation- based MAC scheme for distributed cognitive radio networks,

    M. Lu ´ıs, R. Oliveira, R. Dinis, and L. Bernardo, “A novel reservation- based MAC scheme for distributed cognitive radio networks,”IEEE Transactions on Vehicular Technology, vol. 66, no. 5, pp. 4327–4340, 2016

    2016

  14. [14]

    5G; study on channel model for frequencies from 0.5 to 100 GHz (3GPP TR 38.901 version 16.1.0 release 16),

    ETSI, “5G; study on channel model for frequencies from 0.5 to 100 GHz (3GPP TR 38.901 version 16.1.0 release 16),” Tech. Rep. TR 138 901 V16.1.0, European Telecommunications Standards Institute, Nov. 2020

    2020

  15. [15]

    NR; Physical layer procedures for data (Release 15),

    3GPP, “NR; Physical layer procedures for data (Release 15),” Technical Specification TS 38.214 V15.3.0, 3rd Generation Partnership Project (3GPP), 2018

    2018

  16. [16]

    Action space shaping in deep reinforcement learning,

    A. Kanervisto, C. Scheller, and V . Hautam ¨aki, “Action space shaping in deep reinforcement learning,” inIEEE conference on games (CoG), Osaka, Japan, Aug. 2020

    2020

  17. [17]

    AI-RAN simulator

    “AI-RAN simulator.” https://github.com/ntutangyun/ai-ran-sim. Ac- cessed: 2025-11-24

    2025

  18. [18]

    Jamming Attacks and Anti-Jamming Strategies in Wireless Networks: A Comprehensive Survey,

    H. Pirayesh and H. Zeng, “Jamming Attacks and Anti-Jamming Strategies in Wireless Networks: A Comprehensive Survey,”IEEE Communications Surveys & Tutorials, vol. 24, no. 2, pp. 767–809, 2022

    2022