Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

Rogue Cell: Adversarial Attack and Defense in Untrusted O-RAN Setup Exploiting the Traffic Steering xApp

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A malicious cell in a multi-operator O-RAN can trick traffic steering into over-allocating user equipment by 248.5 percent, and a contextual LSTM-autoencoder detector catches the tampering.

desk verdict New cell-as-attacker threat model and a useful closed-loop testbed, but the attack's feasibility rests on an underspecified substitute-model query step that the testbed silently grants. read the letter →

arxiv 2505.01816 v1 pith:PT6WNRH7 submitted 2025-05-03 cs.CR cs.LG

Eran Aizikovich , Dudu Mimran , Edita Grolman , Yuval Elovici , Asaf Shabtai This is my paper · ORCID
classification cs.CRcs.LG
keywords O-RANsecurityadversarialmachinelearningtrafficsteeringmulti-operatorRANKPImanipulationLSTMautoencoderevasionattackxApp
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that the shift to multi-operator open radio access networks (O-RAN) creates a new attack surface: an untrusted cell operator can manipulate the key performance indicators it reports and deceive the traffic steering loop into handing over far more user equipment than the cell should receive. The authors build a testbed combining a wireless network simulator with an official O-RAN RIC platform and demonstrate APATE, an evasion attack in which a malicious cell queries a substitute QoE predictor to craft adversarial KPI reports. In their experiments, the attacked cell serves 248.5% more UEs on average than in the benign baseline, while neighboring cells lose users. To counter this, they propose MARRS, a two-stage LSTM-autoencoder detector that learns each cell's normal telemetry together with network-wide context and flags tampered reports with 99.2% accuracy and an F1 score of 0.978. A sympathetic reader would take the paper as establishing that self-reported cell telemetry is a credible attack vector in multi-operator O-RAN, and that contextual reconstruction-based detection is a viable defense.

What carries the argument

The load-bearing object is the QoE predictor (QP) in the traffic steering loop, together with the attack's substitute-model pipeline and MARRS's two-layer contextual autoencoder. APATE's machinery is Eq. (1): the attacker adds an optimal perturbation $\delta^*$ to its KPI report $R$ so that the QP's prediction $Q(R+\delta^*)$ is pushed toward a higher QoE class, with $\delta^*$ found by querying a locally trained substitute QP using the HopSkipJump decision-based attack. MARRS's machinery is Eq. (3): each cell's latent embedding $\mathrm{emb}_{c_i}$ from a first LSTM autoencoder is concatenated with the average embedding of all other cells to form an enriched feature vector $X^{(2)}_{c_i}$, which a second autoencoder must decode back into the original features; the reconstruction loss $\ell(F(R),R)$ compared to a threshold $T$ decides trusted versus untrusted. The contextual second stage is what lets MARRS spot tampering that a per-cell autoencoder alone would tolerate.

What would settle it

Run the APATE recipe against a production-like QoE predictor where the attacker can only observe the traffic steering handover decisions (not the QP's internal scores) and must train its substitute from those observable outcomes; if the substitute's predictions do not transfer to the target model, the crafted perturbations will not raise the QoE forecast and the UE allocation gain will not materialize. A simpler check: measure the substitute model's agreement with the target QP on a held-out set of benign KPI reports; if agreement is near chance, the attack fails before it starts.

Watch

Extended reading notes

Core claim

The central discovery is a concrete attack-defense pair for the traffic steering (TS) flow of an O-RAN near-real-time RIC. Under the threat model where a cell is operated by an untrusted third-party operator with a financial incentive to serve more users, APATE works by training a substitute of the QoE predictor (QP) xApp, using a decision-based adversarial attack (HopSkipJump) to compute minimal perturbations to the cell's reported KPIs, and then reporting those perturbed KPIs to the RIC; the QP then forecasts an artificially high QoE for that cell, and the TS xApp allocates additional UEs to it. In the testbed, this yields a 248.5% average increase in UEs served by the malicious cell in the single-attacker scenario and similar gains (154.16% and 245.68%) for two simultaneous attackers, at the expense of neighboring cells. The companion result is MARRS, which trains a per-cell LSTM autoencoder, enriches each cell's latent embedding with the aggregated embeddings of all other cells, and trains a second autoencoder to reconstruct the original features; reconstruction loss above a threshold marks the report as untrusted. On the same testbed data, MARRS reaches 99.2% accuracy and 0.978 F1, beating isolation forest, one-class SVM, and a plain linear autoencoder, and a sequence-based variant eliminates false positives entirely.

Load-bearing premise

The attack depends on the premise that a malicious cell operator can train an accurate substitute of the QoE predictor using data and query access that a real operator would actually have in a multi-operator deployment; in the testbed this access is granted by construction, and the paper does not specify how the substitute is trained, how many queries are needed, or whether the attacker only observes handover decisions rather than QP outputs.

Editorial extensions

If this is right

  • In a single-attacker scenario, the malicious cell's average UE count rises from 4.27 to 10.61, a 248.5% gain, while the neighboring cell BS6 drops to 57.27% of its benign allocation.
  • With two simultaneous attackers, both malicious cells gain (154.16% and 245.68%) and the neighboring cell BS6 loses about half its users, showing the attack redistributes load network-wide, not just to the attacker.
  • The same threat model covers a supply-chain-compromised cell in a single-operator network, so the defense is relevant beyond multi-operator settings.
  • Deploying MARRS as an xApp on the near-RT RIC detects the tampered telemetry in real time, with accuracy improving as more training data accumulates over time.
  • The sequence-based S-MARRS variant removes false positives entirely (precision 1.0) while preserving detection, at the cost of requiring several consecutive windows before flagging.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper's testbed grants the attacker the ability to train a substitute QP by construction; a real multi-operator attacker may only observe handover decisions, not QP scores, so the practical gain depends on whether a substitute trained on observable outcomes transfers to the target model.
  • If the attack transfers, the same mechanism should generalize to any RIC use case that acts on self-reported cell KPIs, such as slicing or load balancing; MARRS's context-enriched reconstruction is a template for defending those flows too.
  • An adaptive attacker aware of MARRS could try to craft perturbations that stay within the benign reconstruction-loss manifold; testing MARRS against such an adversary would be a natural next experiment.
  • The financial incentive structure (operators paid per served UE) means the attack's real-world impact scales with how directly UE counts map to revenue; quantifying that mapping in actual roaming agreements would sharpen the risk assessment.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper claims to be the first to study security challenges introduced by multi-operator O-RAN deployments, in which a cell operator is treated as an untrusted actor. The authors propose APATE, an evasion attack in which a malicious cell perturbs its reported KPIs to fool the QoE predictor (QP) in the O-RAN traffic-steering flow, thereby obtaining an unfair share of UEs. They also propose MARRS, a context-aware LSTM-autoencoder detector that classifies cell telemetry as trusted or untrusted, with an extension S-MARRS for sequence-level detection. The evaluation is performed in a closed-loop testbed that couples a wireless network simulator with an OSC near-RT RIC cluster. The headline results are a 248.5% increase in the malicious cell's average UE allocation in the single-attack scenario and a MARRS accuracy of 99.2% with F1 of 0.978 on the test set.

Significance. If the central claims hold, the paper makes a useful contribution: it identifies a new threat actor (the cell itself) in multi-operator O-RAN, demonstrates a concrete attack path through the QP model, and proposes a contextual detection method that outperforms standard baselines in the testbed. The open-access testbed integrating a simulator with the official OSC RIC cluster is a practical resource for the community. The novelty relative to prior work on O-RAN attacks is real. However, the strength of the empirical evidence is currently limited by an underspecified attacker model, a single simulation run per scenario, and post-hoc threshold selection for the detector; these gaps must be closed before the quantitative claims can be accepted at face value.

major comments (4)
  1. [Section 5, Section 4] The substitute QP training procedure is not specified. Section 5 states that 'the attacker begins by training a substitute QP model,' but the paper never describes the dataset, feature set, label source, number of queries, or whether the cell can observe the QP's QoE outputs in a real multi-operator deployment. The adversarial capabilities in Section 4 list only KPI manipulation and knowledge of the TS flow. In the testbed, the attacker is part of the closed loop controlled by the authors, so substitute training could silently rely on access that a real cell operator would not have. Because HopSkipJump (Section 8.1.1) is a decision-based attack requiring many hard-label queries, it is also unclear whether the attacker queries the target QP or only the substitute. This is a load-bearing gap: if the attacker cannot obtain accurate QP predictions or labels in practice, the 248.5% UE gain in Table 3 does not materialize. Please specify the substitute training protocol and demonstrate the attack under realistic observability assumptions, e.g., where the cell observes only its own KPI reports and the handover decisions addressed to it, with a bounded query budget.
  2. [Section 8.2.1, Table 3] The headline result is based on a single simulation run per scenario with no variance reporting. Table 3 reports mean, min, and max UE counts for one benign and one malicious run in SAS and MAS, but there are no confidence intervals, standard deviations across seeds, or statistical tests. It is therefore impossible to assess whether the observed differences are stable or could be artifacts of one random trajectory. Please provide multiple independent runs with different random seeds for both benign and attack scenarios, and report the distribution of the attack's impact (e.g., mean and 95% CI of the UE-allocation ratio).
  3. [Section 8.1.2, Section 8.2.2, Table 5] The MARRS detection threshold T is selected post-hoc on the test set. Section 8.1.2 states 'we set the threshold policy T to maximize the F1 score in the classification processes,' and Table 5 reports metrics on the test set. This is a form of test-set tuning: the threshold is chosen to optimize the very metric later reported, which inflates the accuracy and F1 figures. The threshold should instead be selected on a validation set or by cross-validation, and only then evaluated on a held-out test set. This is a central methodological point for the detection claim, since the entire benefit of MARRS over the baselines is quantified with these numbers.
  4. [Abstract, Section 8.2.1, Table 3] The reported percentage difference is mathematically overstated. In Table 3, BS5's mean UE count in the benign SAS is 4.27 and in the malicious SAS is 10.61; the ratio is 10.61/4.27 = 2.485, i.e., the malicious cell receives 248.5% of the benign allocation, which is a 148.5% increase, not a 248.5% increase. The same mislabeling appears in the MAS rows (e.g., 11.21/4.56 = 2.458, i.e., 145.8% greater, not 245.68%). The abstract's claim of '248.5% greater UE allocation' is therefore incorrect. Please correct the wording throughout, or change the calculation to (malicious - benign) / benign.
minor comments (5)
  1. [Section 6, Eq. (2)] Equation (2) contains stray commas and the surrounding text is inconsistent: the sentence before the equation says a loss higher than T is classified as trusted, while the subsequent text and the equation itself say the opposite. Please unify the description and the equation.
  2. [Section 9, Discussion] The acronym MARRS is misspelled as 'MARSS' in the sentence 'Deploying MARSS on the near-RT RIC'.
  3. [Section 8.2.2, Table 4] The training-set subsets x1, x2, x3, x4 are not defined in the text. Please specify what portion of the collected time series each subset corresponds to, so the reader can interpret the 'more data over time' claim.
  4. [Section 6.1.1, Table 2] The feature table uses the abbreviation 'RSSNIR' for what appears to be signal-to-noise ratio, while the text and simulator description mention RSRQ and SNIR separately. Please align the terminology and units.
  5. [Section 8.1.1] The sentence 'To accurately model real-world attack progression, we initialize both attack scenarios using identical conditions to their benign baseline scenarios, while the benign scenarios initialized randomly' is ambiguous. Please clarify whether the benign runs use a single random seed or multiple seeds, and whether the attack runs reuse the same trajectories as the benign runs.

Circularity Check

1 steps flagged · score 6.0 of 10

MARRS detection metrics are partially constructed: the threshold T is tuned to maximize F1 on the same evaluation, while APATE remains an empirical testbed demonstration.

  1. fitted input called prediction [Section 6.4 (Eq. 2); Section 8.1.2 (Model Training); Section 8.2.2, Table 5]
    ""To classify these reports, a threshold T needs to be defined based on a certain policy provided by the operator." ... "In these experiments, we set the threshold policy T to maximize the F1 score in the classification processes." ... "MARRS 0.992 0.958 1 0.978" (Table 5)."

    The classifier in Eq. (2) is defined entirely by the threshold T; T is set to maximize F1 'in the classification processes,' and the same process is then reported as the detector's F1 (0.978) and accuracy (0.992). No validation split or held-out threshold-selection step is described. Thus the headline detection numbers are not independent predictions: T is chosen to optimize the exact metric reported on the same data, so the reported F1 is by construction the best threshold-dependent value on that test set. This also biases the comparison against IF/OCSVM/LAE, whose thresholds are not described as tuned the same way.

full rationale

APATE's central result is not circular: the 248.5% UE allocation increase is an observed comparison between attack and benign simulator runs, and the substitute-QP training step is under-specified (a correctness/feasibility risk) rather than an equation that reduces to its input. The self-citation [29] in Section 4 supports only the plausible capability that the malicious cell knows the TS flow; it is not load-bearing for the attack derivation. The concrete circularity is in MARRS: Section 6.4 defines the classifier via threshold T, and Section 8.1.2 sets T to maximize F1 'in the classification processes.' The same processes yield the F1=0.978 and accuracy=0.992 reported in Table 5, so these metrics are partly fitted rather than predicted. Because the attack claim is independent and only the detector evaluation is affected, I score this as partial circularity rather than wholesale input-equivalence.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The empirical claims rest on several domain assumptions: KPI manipulation is possible, the OSC traffic steering flow is representative, operators are paid per UE, and the simulator mirrors reality. The free parameters are dominated by threshold and hyperparameter choices that are tuned rather than derived. No new physical or architectural entities are introduced, so the invented-entities ledger is empty.

free parameters (4)
  • MARRS detection threshold T = chosen to maximize F1 score on the test set; numerical value not reported
    Section 8.1.2 states 'we set the threshold policy T to maximize the F1 score in the classification processes.' This makes the reported 99.2% accuracy optimistic and is a test-set-tuned parameter.
  • LSTM autoencoder hyperparameters (hidden size, learning rate, layers) = tuned via Optuna; final values not reported
    Section 8.1.2 says hyperparameters are tuned using Optuna, but the chosen values are not listed, so replication requires re-tuning on the same data.
  • S-MARRS sequence length and classification rule = e.g., 5A, 5M, etc. in Fig. 9; no principled criterion
    Section 6.5 and Fig. 9 evaluate several sequence configurations; the selection affects FPR and F1, and the choice is experimental rather than derived.
  • HopSkipJump attack parameters = not specified; no perturbation budget or query count reported
    Section 8.1.1 states HopSkipJump from ART was used, but no bound on perturbation magnitude or number of queries is given, leaving the attack strength unconstrained.
assumptions (5)
  • domain assumption The malicious cell can modify the KPI reports it sends to the RIC without being prevented by existing integrity mechanisms.
    Section 4, 'Adversarial Capabilities': 'The malicious cell can manipulate the KPIs it reports to the RIC.' This is the core vulnerability the paper assumes; real deployments with attestation or integrity controls could invalidate it.
  • domain assumption The traffic steering flow implemented in the OSC RIC cluster is representative of production O-RAN traffic steering behavior.
    Sections 2.3 and 7 use the OSC QP, AD, KPIMON, and TS xApps. The paper's claims about 'the O-RAN traffic steering' depend on this representativeness.
  • domain assumption Operators are compensated per served UE, giving the malicious operator financial incentive to inflate its allocation.
    Section 4, 'Financial Model', cites Farhat et al. [22,23]. If the economic model is not per-UE payment, the attack motivation weakens.
  • domain assumption The wireless network simulator produces KPI distributions representative of real cellular networks.
    Section 7.1 describes the simulator as the source of all KPI data; no calibration against real RAN telemetry is presented, so the 248.5% figure is simulator-specific.
  • ad hoc to paper The attacker has sufficient knowledge and data to train an accurate substitute model of the target QP.
    Section 4 asserts 'The malicious cell has knowledge of the targeted TS task flow.' The training data and query access for the substitute model are never specified; this is asserted, not derived.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Rogue Cell: Adversarial Attack and Defense in Untrusted O-RAN Setup Exploiting the Traffic Steering xApp." pith.science (2026). https://pith.science/paper/PT6WNRH7

@misc{pith2026250501816,
  author       = {Pith},
  title        = {Pith review of: Rogue Cell: Adversarial Attack and Defense in Untrusted O-RAN Setup Exploiting the Traffic Steering xApp},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PT6WNRH7}},
  note         = {Machine review of arXiv:2505.01816}
}
read the original abstract

The Open Radio Access Network (O-RAN) architecture is revolutionizing cellular networks with its open, multi-vendor design and AI-driven management, aiming to enhance flexibility and reduce costs. Although it has many advantages, O-RAN is not threat-free. While previous studies have mainly examined vulnerabilities arising from O-RAN's intelligent components, this paper is the first to focus on the security challenges and vulnerabilities introduced by transitioning from single-operator to multi-operator RAN architectures. This shift increases the risk of untrusted third-party operators managing different parts of the network. To explore these vulnerabilities and their potential mitigation, we developed an open-access testbed environment that integrates a wireless network simulator with the official O-RAN Software Community (OSC) RAN intelligent component (RIC) cluster. This environment enables realistic, live data collection and serves as a platform for demonstrating APATE (adversarial perturbation against traffic efficiency), an evasion attack in which a malicious cell manipulates its reported key performance indicators (KPIs) and deceives the O-RAN traffic steering to gain unfair allocations of user equipment (UE). To ensure that O-RAN's legitimate activity continues, we introduce MARRS (monitoring adversarial RAN reports), a detection framework based on a long-short term memory (LSTM) autoencoder (AE) that learns contextual features across the network to monitor malicious telemetry (also demonstrated in our testbed). Our evaluation showed that by executing APATE, an attacker can obtain a 248.5% greater UE allocation than it was supposed to in a benign scenario. In addition, the MARRS detection method was also shown to successfully classify malicious cell activity, achieving accuracy of 99.2% and an F1 score of 0.978.

Figures

Figures reproduced from arXiv: 2505.01816 by the authors.

Figure 1
Figure 1. O-RAN architecture high-level overview. and radio unit (RU), enabling greater flexibility and modu￾larity. These units connect to RICs via open interfaces, en￾abling the streaming of RAN telemetry and the deployment of control actions and policies. The O-RAN architecture’s components and their open interfaces are illustrated in [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. High-level overview of the relevant components in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Threat analysis based on NIST ontology [ [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Attack flow steps: (1) the attacker trains a substitute QP model, replicating the behavior of the target model; (2) [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: LSTM-autoencoder framework architecture. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Testbed Environment; left - OSC near-RT RIC Ku [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Amount of UE for each cell (BS1-BS6) during each of the network iterations (x-axis) in the examined scenarios: the [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Network topology during the benign scenario [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: Sequence-based detection approach results. [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Distilling Unitary Operations: A No-Go Theorem and Minimal Realization

    quant-ph 2026-04 unverdicted novelty 6.0 of 10

    Under depolarizing noise, universal single-qubit unitary purification is impossible with any nontrivial 2-slot indefinite-causal-order process, but a parallel 3-slot architecture is minimal and achieves a strictly bet...

  2. Jamming Smarter, Not Harder: Exploiting O-RAN Y1 RAN Analytics for Efficient Interference

    cs.CR 2025-09 conditional novelty 6.0 of 10

    Authorized Y1 RAN analytics can be misused to guide a selective jammer, replicating most of an always-on jammer's disruption with roughly 27% less transmission time.

Reference graph

Works this paper leans on

76 extracted references · 67 canonical work pages · cited by 2 Pith papers

  1. [1]

    TR 22.852: Study on RAN Sharing Enhancements (Release 12 & 13)

    3rd Generation Partnership Project. TR 22.852: Study on RAN Sharing Enhancements (Release 12 & 13). Technical report, Technical Specification Group Radio Access Networks, 2013

    work page 2013

  2. [2]

    Overview of 5g security challenges and solutions

    Ijaz Ahmad, Tanesh Kumar, Madhusanka Liyanage, Jude Okwuibe, Mika Ylianttila, and Andrei Gurtov. Overview of 5g security challenges and solutions. IEEE Communications Standards Magazine, 2(1):36– 43, 2018

    work page 2018

  3. [3]

    Optuna: A next- generation hyperparameter optimization framework

    Takuya Akiba, Shotaro Sano, Toshihiko Yanase, Takeru Ohta, and Masanori Koyama. Optuna: A next- generation hyperparameter optimization framework. In Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, pages 2623–2631, 2019

    work page 2019

  4. [4]

    Kpi poison- ing: An attack in open ran near real-time control loop

    Hamed Alimohammadi, Sotiris Chatzimiltis, Samara Mayhoub, Mohammad Shojafar, Seyed Ahmad Soley- mani, Ayhan Akbas, and Chuan Heng Foh. Kpi poison- ing: An attack in open ran near real-time control loop

    work page

  5. [5]

    WG11: O-RAN Work Group 11 (Security Work Group) Security Requirements and Controls Specifications, 2024

    O-RAN ALLIANCE. WG11: O-RAN Work Group 11 (Security Work Group) Security Requirements and Controls Specifications, 2024

    work page 2024

  6. [6]

    A survey on open radio access networks: Challenges, research directions, and open source approaches.Sen- sors, 24(3):1038, 2024

    Wilfrid Azariah, Fransiscus Asisi Bimo, Chih-Wei Lin, Ray-Guang Cheng, Navid Nikaein, and Rittwik Jana. A survey on open radio access networks: Challenges, research directions, and open source approaches.Sen- sors, 24(3):1038, 2024

    work page 2024

  7. [7]

    Attacking o-ran inter- faces: Threat modeling, analysis and practical experi- mentation.IEEE Open Journal of the Communications Society, 2024

    Pau Baguer, Girma M Yilma, Esteban Municio, Gines Garcia-Aviles, Andres Garcia-Saavedra, Marco Lieb- sch, and Xavier Costa-Pérez. Attacking o-ran inter- faces: Threat modeling, analysis and practical experi- mentation.IEEE Open Journal of the Communications Society, 2024

    work page 2024

  8. [8]

    En- hancing o-ran security: Evasion attacks and robust de- fenses for graph reinforcement learning-based connec- tion management.arXiv preprint arXiv:2405.03891, 2024

    Ravikumar Balakrishnan, Marius Arvinte, Nageen Hi- mayat, Hosein Nikopour, and Hassnaa Moustafa. En- hancing o-ran security: Evasion attacks and robust de- fenses for graph reinforcement learning-based connec- tion management.arXiv preprint arXiv:2405.03891, 2024

    arXiv 2024

Show all 76 references
  1. [9]

    Ric: A ran intelligent controller platform for ai-enabled cellu- lar networks.IEEE Internet Computing, 25(2):7–17, 2021

    Bharath Balasubramanian, E Scott Daniels, Matti Hiltunen, Rittwik Jana, Kaustubh Joshi, Rajarajan Sivaraj, Tuyen X Tran, and Chengwei Wang. Ric: A ran intelligent controller platform for ai-enabled cellu- lar networks.IEEE Internet Computing, 25(2):7–17, 2021

    2021

  2. [10]

    Evasion attacks against machine learning at test time

    Battista Biggio, Igino Corona, Davide Maiorca, Blaine Nelson, Nedim Šrndi ´c, Pavel Laskov, Giorgio Giac- into, and Fabio Roli. Evasion attacks against machine learning at test time. InMachine Learning and Knowl- edge Discovery in Databases: European Conference, ECML PKDD 2013...

    2013

  3. [11]

    Osc community lab: The integration test bed for o-ran soft- ware community

    Fransiscus Asisi Bimo, Ferlinda Feliana, Shu-Hua Liao, Chih-Wei Lin, David F Kinsey, James Li, Rit- twik Jana, Richard Wright, and Ray-Guang Cheng. Osc community lab: The integration test bed for o-ran soft- ware community. In2022 IEEE Future Networks World Forum (FNWF), pages...

    2022

  4. [12]

    Eval- uating the cybersecurity risk of real-world, machine learning production systems.ACM Computing Surveys, 55(9):1–36, 2023

    Ron Bitton, Nadav Maman, Inderjeet Singh, Satoru Momiyama, Yuval Elovici, and Asaf Shabtai. Eval- uating the cybersecurity risk of real-world, machine learning production systems.ACM Computing Surveys, 55(9):1–36, 2023

    2023

  5. [13]

    Open, pro- grammable, and virtualized 5g networks: State-of- the-art and the road ahead.Computer Networks, 182:107516, 2020

    Leonardo Bonati, Michele Polese, Salvatore D’Oro, Stefano Basagni, and Tommaso Melodia. Open, pro- grammable, and virtualized 5g networks: State-of- the-art and the road ahead.Computer Networks, 182:107516, 2020

    2020

  6. [14]

    Evaluation of the latency of machine learning random access ddos detection in open ran

    Paulo Ricardo Branco da Silva, João Paulo Henriques Sales de Lima, Erika Costa Alves, William Sanchez Farfan, Victor Aguiar Coutinho, Thomas William do Prado Paiva, Daniel Lazkani Feferman, and Fran- cisco Hugo Costa Neto. Evaluation of the latency of machine learning random a...

    2024

  7. [15]

    Flexslice: Flexible and real-time programmable ran slicing framework

    Chieh-Chun Chen, Chia-Yu Chang, and Navid Nikaein. Flexslice: Flexible and real-time programmable ran slicing framework. InGLOBECOM 2023-2023 IEEE Global Communications Conference, pages 3807–

    2023

  8. [16]

    Hopskipjumpattack: A query-efficient 14 decision-based attack

    Jianbo Chen, Michael I Jordan, and Martin J Wainwright. Hopskipjumpattack: A query-efficient 14 decision-based attack. In2020 ieee symposium on se- curity and privacy (sp), pages 1277–1294. IEEE, 2020

    2020

  9. [17]

    Satellite integration into 5g: deep re- inforcement learning for network selection.Machine Intelligence Research, 19(2):127–137, 2022

    Emanuele De Santis, Alessandro Giuseppi, Anto- nio Pietrabissa, Michael Capponi, and Francesco Delli Priscoli. Satellite integration into 5g: deep re- inforcement learning for network selection.Machine Intelligence Research, 19(2):127–137, 2022

    2022

  10. [18]

    5g on the horizon: Key challenges for the radio-access network.IEEE vehicular technology magazine, 8(3):47–53, 2013

    Panagiotis Demestichas, Andreas Georgakopou- los, Dimitrios Karvounas, Kostas Tsagkaris, Vera Stavroulaki, Jianmin Lu, Chunshan Xiong, and Jing Yao. 5g on the horizon: Key challenges for the radio-access network.IEEE vehicular technology magazine, 8(3):47–53, 2013

    2013

  11. [19]

    Toward modular and flexible open ran implementations in 6g networks: Traffic steering use case and o-ran xapps.Sensors, 21(24):8173, 2021

    Marcin Dryja ´nski, Łukasz Kułacz, and Adrian Kliks. Toward modular and flexible open ran implementations in 6g networks: Traffic steering use case and o-ran xapps.Sensors, 21(24):8173, 2021

    2021

  12. [20]

    Freshwave says four way operator sharing on same indoor small cells a world first, 2024

    Keith Dyer. Freshwave says four way operator sharing on same indoor small cells a world first, 2024

    2024

  13. [21]

    Spacetime and O-RAN Interfaces 5G/6G NTN,

    ESA. Spacetime and O-RAN Interfaces 5G/6G NTN,

  14. [22]

    Access selec- tion and joint pricing in multi-operator wireless net- works: A stackelberg game

    Soha Farhat, Zahraa Chahine, Abed Ellatif Samhat, Samer Lahoud, and Bernard Cousin. Access selec- tion and joint pricing in multi-operator wireless net- works: A stackelberg game. In2015 Fifth International Conference on Digital Information and Communica- tion Technology and i...

    2015

  15. [23]

    Best operator policy in a heteroge- neous wireless network

    Soha Farhat, Abed Ellatif Samhat, Samer Lahoud, and Bernard Cousin. Best operator policy in a heteroge- neous wireless network. InThe Third International Conference on e-Technologies and Networks for Devel- opment (ICeND2014), pages 53–57. IEEE, 2014

    2014

  16. [24]

    Radio access network sharing in 5g: strategies and benefits.Wireless Personal Communica- tions, 96:2715–2740, 2017

    Soha Farhat, Abed Ellatif Samhat, Samer Lahoud, and Bernard Cousin. Radio access network sharing in 5g: strategies and benefits.Wireless Personal Communica- tions, 96:2715–2740, 2017

    2017

  17. [25]

    Explaining and harnessing adversar- ial examples.arXiv preprint arXiv:1412.6572, 2014

    Ian J Goodfellow. Explaining and harnessing adversar- ial examples.arXiv preprint arXiv:1412.6572, 2014

    2014 arXiv

  18. [26]

    Implementing and evaluating se- curity in o-ran: Interfaces, intelligence, and platforms

    Joshua Groen, Salvatore D’Oro, Utku Demir, Leonardo Bonati, Michele Polese, Tommaso Melodia, and Kaushik Chowdhury. Implementing and evaluating se- curity in o-ran: Interfaces, intelligence, and platforms. IEEE Network, 2024

    2024

  19. [27]

    V odafone announces expanded net- work sharing agreement with orange in spain, 2019

    V odafone Group. V odafone announces expanded net- work sharing agreement with orange in spain, 2019

    2019

  20. [28]

    Mobile infrastructure sharing

    GSMA. Mobile infrastructure sharing

  21. [29]

    Adversarial ma- chine learning threat analysis and remediation in open radio access network (o-ran).arXiv preprint arXiv:2201.06093, 2022

    Edan Habler, Ron Bitton, Dan Avraham, Dudu Mim- ran, Eitan Klevansky, Oleg Brodt, Heiko Lehmann, Yuval Elovici, and Asaf Shabtai. Adversarial ma- chine learning threat analysis and remediation in open radio access network (o-ran).arXiv preprint arXiv:2201.06093, 2022

    2022 arXiv

  22. [30]

    From centralized ran to open ran: A survey on the evolu- tion of distributed antenna systems.arXiv preprint arXiv:2411.12166, 2024

    Mahmoud A Hasabelnaby, Mohanad Obeed, Mo- hammed Saif, Anas Chaaban, and MJ Hossain. From centralized ran to open ran: A survey on the evolu- tion of distributed antenna systems.arXiv preprint arXiv:2411.12166, 2024

    2024 arXiv

  23. [31]

    Security threats to xapps access control and e2 interface in o-ran.IEEE Open Journal of the Communications Society, 2024

    Cheng-Feng Hung, You-Run Chen, CHI-Heng Tseng, and Shin-Ming Cheng. Security threats to xapps access control and e2 interface in o-ran.IEEE Open Journal of the Communications Society, 2024

    2024

  24. [32]

    Ric: The next phase of open ran, 2024

    Eugina Jordan. Ric: The next phase of open ran, 2024

    2024

  25. [33]

    How does open ran add value in multi- operator sharing?, 2021

    Shin Yuan Kee. How does open ran add value in multi- operator sharing?, 2021

    2021

  26. [34]

    Lstm-autoencoder based anomaly detection using vibration data of wind turbines.Sen- sors, 24(9):2833, 2024

    Younjeong Lee, Chanho Park, Namji Kim, Jisu Ahn, and Jongpil Jeong. Lstm-autoencoder based anomaly detection using vibration data of wind turbines.Sen- sors, 24(9):2833, 2024

    2024

  27. [35]

    Improving one-class svm for anomaly detec- tion

    Kun-Lun Li, Hou-Kuan Huang, Sheng-Feng Tian, and Wei Xu. Improving one-class svm for anomaly detec- tion. InProceedings of the 2003 international confer- ence on machine learning and cybernetics (IEEE Cat. No. 03EX693), volume 5, pages 3077–3081. IEEE, 2003

    2003

  28. [36]

    Iso- lation forest

    Fei Tony Liu, Kai Ming Ting, and Zhi-Hua Zhou. Iso- lation forest. In2008 eighth ieee international confer- ence on data mining, pages 413–422. IEEE, 2008

    2008

  29. [37]

    Democratizing{Direct-to-Cell}low earth orbit satellite networks

    Lixin Liu, Yuanjie Li, Hewu Li, Jiabo Yang, Wei Liu, Jingyi Lan, Yufeng Wang, Jiarui Li, Jianping Wu, Qian Wu, et al. Democratizing{Direct-to-Cell}low earth orbit satellite networks. In21st USENIX Symposium on Networked Systems Design and Implementation (NSDI 24), pages 791–808, 2024

    2024

  30. [38]

    Open ran security: Challenges and opportunities.Journal of Network and Computer Applications, 214:103621, 2023

    Madhusanka Liyanage, An Braeken, Shahriar Sha- habuddin, and Pasika Ranaweera. Open ran security: Challenges and opportunities.Journal of Network and Computer Applications, 214:103621, 2023

    2023

  31. [39]

    Deutsche telekom and telefónica share network infrastructure to enhance network cover- age, 2021

    Johannes Maisack. Deutsche telekom and telefónica share network infrastructure to enhance network cover- age, 2021. 15

    2021

  32. [40]

    Intelli- gent o-ran beyond 5g: Architecture, use cases, chal- lenges, and opportunities.IEEE Access, 12:27088– 27114, 2024

    Simona Marinova and Alberto Leon-Garcia. Intelli- gent o-ran beyond 5g: Architecture, use cases, chal- lenges, and opportunities.IEEE Access, 12:27088– 27114, 2024

    2024

  33. [41]

    Jan Markendahl and Amirhossein Ghanbari. Shared smallcell networks multi-operator or third party solutions-or both? In2013 11th International sympo- sium and workshops on modeling and optimization in mobile, Ad Hoc and wireless networks (WiOpt), pages 41–48. IEEE, 2013

    2013

  34. [42]

    Network cooperation between mobile operators-why and how competitors cooperate? InIMP conf, Atlanta, 2013

    Jan Markendahl, Amirhossein Ghanbari, and Bengt G Mölleryd. Network cooperation between mobile operators-why and how competitors cooperate? InIMP conf, Atlanta, 2013

    2013

  35. [43]

    Ablation stud- ies in artificial neural networks.arXiv preprint arXiv:1901.08644, 2019

    Richard Meyes, Melanie Lu, Constantin Waubert de Puiseau, and Tobias Meisen. Ablation stud- ies in artificial neural networks.arXiv preprint arXiv:1901.08644, 2019

    1901 arXiv

  36. [44]

    Evaluating the security of open ra- dio access networks.arXiv preprint arXiv:2201.06080, 2022

    Dudu Mimran, Ron Bitton, Yehonatan Kfir, Eitan Kl- evansky, Oleg Brodt, Heiko Lehmann, Yuval Elovici, and Asaf Shabtai. Evaluating the security of open ra- dio access networks.arXiv preprint arXiv:2201.06080, 2022

    2022 arXiv

  37. [45]

    Designing a performant ablation study framework for pytorch, 2020

    Alessio Molinari. Designing a performant ablation study framework for pytorch, 2020

    2020

  38. [46]

    Ran sharing: Nec’s approach to- wards active radio access network sharing

    NEC Corporation. Ran sharing: Nec’s approach to- wards active radio access network sharing. Techreport, NEC Corporation, 2013

    2013

  39. [47]

    Adversarial robustness toolbox v1

    Maria-Irina Nicolae, Mathieu Sinn, Minh Ngoc Tran, Beat Buesser, Ambrish Rawat, Martin Wistuba, Valentina Zantedeschi, Nathalie Baracaldo, Bryant Chen, Heiko Ludwig, et al. Adversarial robustness toolbox v1. 0.0.arXiv preprint arXiv:1807.01069, 2018

    2018 arXiv

  40. [48]

    Intelligent o-ran for beyond 5g and 6g wireless networks

    Solmaz Niknam, Abhishek Roy, Harpreet S Dhillon, Sukhdeep Singh, Rahul Banerji, Jeffery H Reed, Navrati Saxena, and Seungil Yoon. Intelligent o-ran for beyond 5g and 6g wireless networks. In2022 IEEE Globecom Workshops (GC Wkshps), pages 215–220. IEEE, 2022

    2022

  41. [49]

    WG2: Non-real-time RAN Intel- ligent Controller and A1 Interface Workgroup, 2022

    O-RAN ALLIANCE. WG2: Non-real-time RAN Intel- ligent Controller and A1 Interface Workgroup, 2022

    2022

  42. [50]

    WG3: Near-real-time RIC and E2 Interface Workgroup, 2022

    O-RAN ALLIANCE. WG3: Near-real-time RIC and E2 Interface Workgroup, 2022

    2022

  43. [51]

    Energy-efficient virtual radio access networks for multi-operators cooperative cellular net- works.IEEE Transactions on Green Communications and Networking, 3(3):603–614, 2019

    Johnson Opadere, Qiang Liu, Tao Han, and Nir- wan Ansari. Energy-efficient virtual radio access networks for multi-operators cooperative cellular net- works.IEEE Transactions on Green Communications and Networking, 3(3):603–614, 2019

    2019

  44. [52]

    Connection management xapp for o-ran ric: A graph neural network and reinforcement learning ap- proach

    Oner Orhan, Vasuki Narasimha Swamy, Thomas Tet- zlaff, Marcel Nassar, Hosein Nikopour, and Shilpa Tal- war. Connection management xapp for o-ran ric: A graph neural network and reinforcement learning ap- proach. In2021 20th IEEE International Confer- ence on Machine Learning a...

    2021

  45. [53]

    An investigation on open-ran specifications: Use cases, security threats, requirements, discussions

    Heejae Park, Tri-Hai Nguyen, and Laihyuk Park. An investigation on open-ran specifications: Use cases, security threats, requirements, discussions. CMES-Computer Modeling in Engineering & Sciences, 141(1), 2024

    2024

  46. [54]

    A survey on low latency towards 5g: Ran, core network and caching solu- tions.IEEE Communications Surveys & Tutorials, 20(4):3098–3130, 2018

    Imtiaz Parvez, Ali Rahmati, Ismail Guvenc, Arif I Sarwat, and Huaiyu Dai. A survey on low latency towards 5g: Ran, core network and caching solu- tions.IEEE Communications Surveys & Tutorials, 20(4):3098–3130, 2018

    2018

  47. [55]

    Au- tomatic differentiation in pytorch

    Adam Paszke, Sam Gross, Soumith Chintala, Gregory Chanan, Edward Yang, Zachary DeVito, Zeming Lin, Alban Desmaison, Luca Antiga, and Adam Lerer. Au- tomatic differentiation in pytorch. 2017

    2017

  48. [56]

    Understand- ing o-ran: Architecture, interfaces, algorithms, secu- rity, and research challenges.IEEE Communications Surveys & Tutorials, 25(2):1376–1411, 2023

    Michele Polese, Leonardo Bonati, Salvatore D’oro, Stefano Basagni, and Tommaso Melodia. Understand- ing o-ran: Architecture, interfaces, algorithms, secu- rity, and research challenges.IEEE Communications Surveys & Tutorials, 25(2):1376–1411, 2023

    2023

  49. [57]

    Traffic steering and network selection in 5g networks based on reinforcement learning

    Francesco Delli Priscoli, Alessandro Giuseppi, Francesco Liberati, and Antonio Pietrabissa. Traffic steering and network selection in 5g networks based on reinforcement learning. In2020 European Control Conference (ECC), pages 595–601. IEEE, 2020

    2020

  50. [58]

    Unsupervised pre- training of a deep lstm-based stacked autoencoder for multivariate time series forecasting problems.Scientific reports, 9(1):19038, 2019

    Alaa Sagheer and Mostafa Kotb. Unsupervised pre- training of a deep lstm-based stacked autoencoder for multivariate time series forecasting problems.Scientific reports, 9(1):19038, 2019

    2019

  51. [59]

    How will ric leverage ai/ml to improve user experience?, December 2023

    Nisar Sanadi. How will ric leverage ai/ml to improve user experience?, December 2023

    2023

  52. [60]

    Experimental study of ad- versarial attacks on ml-based xapps in o-ran.arXiv preprint arXiv:2309.03844, 2023

    Naveen Naik Sapavath, Brian Kim, Kaushik Chowd- hury, and Vijay K Shah. Experimental study of ad- versarial attacks on ml-based xapps in o-ran.arXiv preprint arXiv:2309.03844, 2023

    2023 arXiv

  53. [61]

    Security threat analysis and treatment strat- egy for oran

    Chih-Ting Shen, Yu-Yi Xiao, Yi-Wei Ma, Jiann-Liang Chen, Cheng-Mou Chiang, Shiang-Jiun Chen, and Yu- Chuan Pan. Security threat analysis and treatment strat- egy for oran. In2022 24th International Conference on 16 Advanced Communication Technology (ICACT), pages 417–422. IEEE, 2022

    2022

  54. [62]

    Ddos detection and mitigation using d/xapp in o-ran

    Seyed Ahmad Soleymani, Mohsen Eslamnejad, Hamed Alimohammadi, Ayhan Akbas, Chuan Heng Foh, and Mohammad Shojafar. Ddos detection and mitigation using d/xapp in o-ran

  55. [63]

    Intelligent control in 6g open ran: Security risk or opportunity?arXiv preprint arXiv:2405.08577, 2024

    Sanaz Soltani, Mohammad Shojafar, Ali Amanlou, and Rahim Tafazolli. Intelligent control in 6g open ran: Security risk or opportunity?arXiv preprint arXiv:2405.08577, 2024

    2024 arXiv

  56. [64]

    Neutral host: how open ran and neutral host paves the way for 5g, 2021

    Otto T. Neutral host: how open ran and neutral host paves the way for 5g, 2021

    2021

  57. [65]

    A taxonomy and terminology of adversarial machine learning.NIST IR, 2019:1–29, 2019

    Elham Tabassi, Kevin J Burns, Michael Hadjimichael, Andres D Molina-Markham, and Julian T Sexton. A taxonomy and terminology of adversarial machine learning.NIST IR, 2019:1–29, 2019

    2019

  58. [66]

    A survey on handover management: From lte to nr

    Muhammad Tayyab, Xavier Gelabert, and Riku Jäntti. A survey on handover management: From lte to nr. IEEE Access, 7:118907–118930, 2019

    2019

  59. [67]

    Cell tower co-location

    Tower Genius LLC. Cell tower co-location. https://www.cell-phone-towers.com/ Cell-Tower-Colocation.html, 2018. Accessed: Jul. 30, 2018

    2018

  60. [68]

    Ai-driven network intrusion detection and resource allocation in real-world o-ran 5g networks

    Theodoros Tsourdinis, Nikos Makris, Thanasis Ko- rakis, and Serge Fdida. Ai-driven network intrusion detection and resource allocation in real-world o-ran 5g networks. InProceedings of the 30th Annual Interna- tional Conference on Mobile Computing and Network- ing, pages 1842–...

    2024

  61. [69]

    Technology and innovation open ran, 2024

    V odafone. Technology and innovation open ran, 2024

    2024

  62. [70]

    Lstm- autoencoder-based anomaly detection for indoor air quality time-series data.IEEE Sensors Journal, 23(4):3787–3800, 2023

    Yuanyuan Wei, Julian Jang-Jaccard, Wen Xu, Fariza Sabrina, Seyit Camtepe, and Mikael Boulic. Lstm- autoencoder-based anomaly detection for indoor air quality time-series data.IEEE Sensors Journal, 23(4):3787–3800, 2023

    2023

  63. [71]

    Freshwave pumps out 4g from all four uk operators in one unit, 2024

    Andrew Wooden. Freshwave pumps out 4g from all four uk operators in one unit, 2024

    2024

  64. [72]

    Machine learning-based early attack detection using open ran intelligent controller.arXiv preprint arXiv:2302.01864, 2023

    Bruno Missi Xavier, Merim Dzaferagic, Diarmuid Collins, Giovanni Comarela, Magnos Martinello, and Marco Ruffini. Machine learning-based early attack detection using open ran intelligent controller.arXiv preprint arXiv:2302.01864, 2023

    2023 arXiv

  65. [73]

    Cross-domain ai for early attack detection and defense against mali- cious flows in o-ran.arXiv preprint arXiv:2401.09204, 2024

    Bruno Missi Xavier, Merim Dzaferagic, Irene Vilà, Magnos Martinello, and Marco Ruffini. Cross-domain ai for early attack detection and defense against mali- cious flows in o-ran.arXiv preprint arXiv:2401.09204, 2024

    2024 arXiv

  66. [74]

    Ro- bust deep reinforcement learning against adversarial perturbations on state observations.Advances in Neu- ral Information Processing Systems, 33:21024–21037, 2020

    Huan Zhang, Hongge Chen, Chaowei Xiao, Bo Li, Mingyan Liu, Duane Boning, and Cho-Jui Hsieh. Ro- bust deep reinforcement learning against adversarial perturbations on state observations.Advances in Neu- ral Information Processing Systems, 33:21024–21037, 2020

    2020

  67. [75]

    Integrated satellite-terrestrial networks toward 6g: Architectures, applications, and challenges.IEEE Internet of Things Journal, 9(1):437–461, 2021

    Xiangming Zhu and Chunxiao Jiang. Integrated satellite-terrestrial networks toward 6g: Architectures, applications, and challenges.IEEE Internet of Things Journal, 9(1):437–461, 2021. 17

    2021

  68. [2024]

    https://connectivity.esa.int/projects/spacetime- and-oran-interfaces-5g6g-ntns

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.