Stochastic Modeling of Human-Machine Authentication Channels under Partial Information Leakage

Burak Kantarci; Mohammad Zulkernine; Nilesh Chakraborty

arxiv: 2605.02102 · v1 · submitted 2026-05-03 · 💻 cs.CR · cs.HC· cs.LG

Stochastic Modeling of Human-Machine Authentication Channels under Partial Information Leakage

Nilesh Chakraborty , Mohammad Zulkernine , Burak Kantarci This is my paper

Pith reviewed 2026-05-08 19:00 UTC · model grok-4.3

classification 💻 cs.CR cs.HCcs.LG

keywords PIN authenticationpartial information leakagestochastic modelingprobabilistic inferenceIoT securityhuman-machine communicationcredential predictionreliability degradation

0 comments

The pith

Treating PIN entry as a noisy human-IoT communication channel lets a probabilistic model recover up to 55 percent of single missing digits and 12 percent of triple missing digits from real user data.

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

The paper models PIN-based authentication as a stochastic communication process in which users transmit numeric credentials under practical constraints, yet conventional binary security views overlook the gradual reliability loss from partial symbol exposure in IoT settings. It introduces a context-conditioned probabilistic inference framework that treats missing digits as latent variables and estimates them through smoothed conditional probability distributions with fallback priors, avoiding explicit parameterization of hidden-state transitions. Evaluated on over one million real four-digit PIN samples, the approach produces position-dependent reliability metrics across single-, double-, and triple-digit leakage cases while outperforming sequence-model baselines and classical machine learning models on precision, recall, and F1-score. A sympathetic reader would care because the work shows authentication security erodes progressively with each leaked digit rather than failing in an all-or-nothing manner, enabling more realistic assessments of human-machine channels in cyber-physical systems.

Core claim

PIN entry can be formalized as a noisy human-IoT communication channel in which missing digits are treated as latent variables and recovered via smoothed conditional probability distributions with fallback priors that approximate dependencies across positions without explicit hidden-state models; on more than one million real four-digit PIN samples this yields prediction accuracies reaching 55.31 percent for one missing digit and 12.12 percent for three missing digits while delivering higher precision, recall, and F1 scores than standard sequence and classical machine-learning baselines.

What carries the argument

Smoothed conditional probability distributions with fallback priors that perform context-driven inference to estimate latent digit values across positions.

If this is right

Position-dependent reliability metrics can be computed for any number of exposed digits from one to three.
Quality-of-Service degradation in human-machine authentication can be quantified without building explicit hidden Markov models.
The framework demonstrates that partial leakage produces measurable, non-binary drops in channel reliability on real data.
The method consistently exceeds the performance of contiguous-sequence baselines and standard classifiers across leakage scenarios.

Where Pith is reading between the lines

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

IoT authentication policies could shift from binary accept/reject rules toward dynamic thresholds that incorporate expected leakage levels.
The same latent-variable approach might extend directly to other short numeric credentials such as one-time passwords or door codes.
Practical deployment would benefit from periodic retraining on region-specific PIN corpora to keep the conditional distributions current.

Load-bearing premise

That the smoothed conditional distributions fitted to one million real-world PIN samples accurately reflect typical user digit-choice dependencies and that the fallback priors handle unseen contexts without large bias.

What would settle it

Testing the same inference procedure on a fresh collection of several hundred thousand PINs drawn from a different user population and finding single-digit recovery accuracy below 40 percent would indicate the model does not generalize.

Figures

Figures reproduced from arXiv: 2605.02102 by Burak Kantarci, Mohammad Zulkernine, Nilesh Chakraborty.

**Figure 1.** Figure 1: Eavesdropping clear-text BFI transmitted to the view at source ↗

**Figure 2.** Figure 2: Performance of trained shoulder surfers with prob view at source ↗

**Figure 3.** Figure 3: Comparison of prediction accuracy across Decision view at source ↗

read the original abstract

Reliable and secure human-machine communication is fundamental to IoT and cyber-physical ecosystems, where smartphones and wearables commonly serve as authentication controllers. PIN-based authentication can be viewed as a low-bandwidth communication channel through which users transmit numeric credentials under practical constraints. However, conventional evaluations adopt a binary view of security-treating such channels as either fully secure or fully compromised-thereby overlooking the progressive reliability degradation caused by partial information leakage in real-world IoT settings. In this paper, we model the PIN entry process as a stochastic human-IoT communication system and propose a context-conditioned probabilistic inference framework to quantify reliability loss and Quality-of-Service degradation under partial symbol exposure. The proposed approach treats missing digits as latent variables and estimates them using smoothed conditional probability distributions with fallback priors. Unlike traditional sequential models that assume contiguous positional dependencies, the method does not explicitly parameterize hidden-state transitions or emissions; instead, it performs context-driven probabilistic inference to approximate latent dependencies across digit positions. Using over one million real-world four-digit PIN samples, we evaluate single-, double-, and triple-digit leakage scenarios and derive position-dependent reliability metrics. The proposed model achieves up to 55.31% prediction accuracy for one missing digit and 12.12% for three missing digits, while consistently outperforming a standard sequence-model baseline and classical machine learning models in terms of precision, recall, and F1-score. These results formalize PIN entry as a noisy human--IoT communication channel and demonstrate substantial reliability degradation under realistic partial exposure conditions.

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.

Desk Editor's Note private letter to a colleague

The paper frames partial PIN leakage as a stochastic channel and reports empirical guess accuracies on a large real dataset, but the smoothed conditional method looks too basic to reliably recover positional dependencies.

read the letter

The core contribution is treating missing digits in four-digit PINs as latent variables and estimating them via context-conditioned smoothed probabilities plus fallback priors, then measuring prediction accuracy across one-, two-, and three-digit leakage cases on over a million real samples. They also give position-dependent reliability numbers and claim consistent gains over a sequence baseline and standard ML models on precision, recall, and F1. That is new enough: most prior PIN work stays binary or focuses on full compromise, so quantifying gradual QoS drop under realistic partial exposure is a reasonable step for IoT device design. The dataset size is a clear strength and the numbers (55% for one missing digit, 12% for three) are concrete enough to be useful for rough risk estimates. The soft spots are in the method itself. Without explicit state transitions the smoothed conditionals plus priors can easily flatten higher-order correlations, especially once two or three digits are gone and inference falls back toward marginals. The abstract gives no ablation on smoothing strength, no error bars, no data-split details, and no check for whether common patterns like 1234 are driving the reported outperformance. If the dataset over-represents those patterns, the gains could be artifactual rather than evidence that the channel model works. The scope is also narrow—only four-digit PINs—so broader claims about human-machine channels need more support. This is mainly for researchers in IoT security or usable authentication who want empirical leakage numbers rather than pure theory. A serious editor should send it to peer review so the authors can supply the missing methodological checks and ablations; the data and framing are worth that effort even if the current claims need tightening.

Referee Report

3 major / 2 minor

Summary. The paper models PIN-based authentication as a stochastic human-IoT communication channel subject to partial symbol leakage. It proposes a context-conditioned probabilistic inference method that treats missing digits as latent variables and estimates them via smoothed conditional probability distributions augmented with fallback priors, without explicit hidden-state transitions or emissions. Using a dataset of over one million real-world four-digit PINs, the work evaluates single-, double-, and triple-digit leakage scenarios, reports position-dependent reliability metrics, and claims prediction accuracies reaching 55.31% for one missing digit and 12.12% for three missing digits while outperforming a sequence-model baseline and classical machine-learning models on precision, recall, and F1-score.

Significance. If the empirical claims hold under rigorous validation, the work supplies a concrete probabilistic lens for quantifying gradual QoS degradation in authentication channels rather than binary secure/compromised views, which could guide leakage-resilient IoT protocol design. The scale of the real-world PIN corpus is a positive attribute, yet the absence of statistical rigor and ablation studies limits the immediate generalizability of the reported gains.

major comments (3)

[Abstract and Evaluation] Abstract and Evaluation section: the central performance claims (55.31% single-missing and 12.12% triple-missing accuracy, plus consistent outperformance) are presented without error bars, statistical significance tests, details on data splitting, smoothing-parameter selection, or prior-estimation procedure. These omissions directly affect the load-bearing assertion that the smoothed-conditionals approach reliably recovers positional dependencies.
[Proposed framework] Proposed framework (context-conditioned inference): the method relies on smoothed conditional distributions with fallback priors to approximate latent digit-position dependencies without hidden-state transitions. For the triple-missing case this inference largely collapses to marginals; without reported ablations on smoothing strength or prior choice, it remains unclear whether the measured precision/recall/F1 gains reflect genuine modeling of the channel or dataset-specific artifacts (e.g., over-represented 1234-style patterns).
[Evaluation] Evaluation section: the conditional distributions and fallback priors are estimated from the same one-million-PIN corpus used for testing. This introduces moderate data dependence that could inflate the reported metrics; explicit cross-validation or held-out evaluation protocols must be documented to substantiate the outperformance claims.

minor comments (2)

[Abstract] Abstract: replace the vague phrase 'over one million' with the exact sample count and a brief description of the data source or collection method to support reproducibility.
[Evaluation] Throughout: provide implementation details (hyper-parameters, training procedure, feature engineering) for the sequence-model baseline and the classical ML comparators so that the fairness of the reported precision/recall/F1 comparisons can be assessed.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments. We address each major comment below and indicate the revisions planned to improve statistical rigor, ablation analysis, and evaluation protocols.

read point-by-point responses

Referee: [Abstract and Evaluation] Abstract and Evaluation section: the central performance claims (55.31% single-missing and 12.12% triple-missing accuracy, plus consistent outperformance) are presented without error bars, statistical significance tests, details on data splitting, smoothing-parameter selection, or prior-estimation procedure. These omissions directly affect the load-bearing assertion that the smoothed-conditionals approach reliably recovers positional dependencies.

Authors: We agree that the reported accuracies lack error bars, significance tests, and full methodological details, which weakens the claims. In the revised manuscript we will add bootstrap confidence intervals for all metrics, statistical significance tests versus baselines, explicit documentation of the full-corpus usage for empirical probability estimation (as the model captures population statistics), and the procedures for smoothing-parameter selection and prior estimation. These additions will appear in the Evaluation section. revision: yes
Referee: [Proposed framework] Proposed framework (context-conditioned inference): the method relies on smoothed conditional distributions with fallback priors to approximate latent digit-position dependencies without hidden-state transitions. For the triple-missing case this inference largely collapses to marginals; without reported ablations on smoothing strength or prior choice, it remains unclear whether the measured precision/recall/F1 gains reflect genuine modeling of the channel or dataset-specific artifacts (e.g., over-represented 1234-style patterns).

Authors: The referee correctly observes that triple-missing inference reduces largely to marginals plus priors. To demonstrate that performance gains reflect genuine positional modeling rather than artifacts, we will add ablation experiments varying smoothing strength and prior choices, plus discussion of common PIN patterns. These will be included in the Proposed framework section. revision: yes
Referee: [Evaluation] Evaluation section: the conditional distributions and fallback priors are estimated from the same one-million-PIN corpus used for testing. This introduces moderate data dependence that could inflate the reported metrics; explicit cross-validation or held-out evaluation protocols must be documented to substantiate the outperformance claims.

Authors: We acknowledge the data-dependence concern. Although the model is an empirical distribution fitted to the corpus, we will revise the Evaluation section to include results on a held-out test subset and k-fold cross-validation. This will strengthen evidence for the reported outperformance over sequence-model and classical ML baselines. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical accuracies measured against external real-world PIN data

full rationale

The paper defines a context-conditioned probabilistic inference method that estimates missing digits via smoothed conditional distributions and fallback priors computed from a large external dataset of one million real PIN samples. It then applies this method to evaluate single/double/triple leakage scenarios and reports measured accuracies (e.g., 55.31% for one missing digit) by direct comparison to the held-out actual digits in the data. No step in the abstract or described framework reduces a claimed result to an input by construction, self-definition, or self-citation chain; the performance numbers are standard empirical metrics on independent benchmarks rather than tautological outputs of fitted parameters. The derivation chain remains self-contained.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The central claim rests on empirical probability estimates derived from the PIN dataset and the assumption that context-driven conditional inference suffices to capture digit dependencies.

free parameters (2)

smoothed conditional probability distributions
Estimated from the one million PIN samples to perform inference on missing digits.
fallback priors
General priors invoked when context is insufficient, derived from overall data statistics.

axioms (1)

domain assumption Missing digits act as latent variables whose dependencies can be approximated via context-conditioned conditional probabilities without explicit parameterization of hidden-state transitions or emissions.
Stated in the description of the proposed inference framework.

pith-pipeline@v0.9.0 · 5583 in / 1480 out tokens · 109522 ms · 2026-05-08T19:00:16.511476+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

Cost.FunctionalEquation / Foundation.LogicAsFunctionalEquation washburn_uniqueness_aczel (no overlap) unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

treats missing digits as latent variables and estimates them using smoothed conditional probability distributions with fallback priors ... performs context-driven probabilistic inference to approximate latent dependencies across digit positions
Foundation.ArithmeticFromLogic / Foundation.AlexanderDuality n/a — empirical Laplace smoothing on a 10-symbol alphabet, no ratio-symmetric cost or dimensional forcing unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

P(x | C) = (N(x,C) + α)/(N(C) + α·|D|) with α = 1.0 ... PIN d = [d1,d2,d3,d4], di ∈ {0,...,9}

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

23 extracted references · 23 canonical work pages

[1]

Reliability on the Internet of Things with Designing Approach for Exploratory Analysis,

K. Singh, M. Yadav, Y. Singh, D. Barak, A. Saini, and F. Mor- eira, “Reliability on the Internet of Things with Designing Approach for Exploratory Analysis,” Frontiers in Computer Science, vol. 6, p. 1382347, 2024

work page 2024
[2]

Minding the source: Toward an Integrative Theory of Human-Machine Communication,

E.-J. Lee, “Minding the source: Toward an Integrative Theory of Human-Machine Communication,” Human Communication Research, vol. 50, no. 2, pp. 184–193, 2024

work page 2024
[3]

The human factor: A challenge for network reliability design,

M. Mushi, E. Murphy-Hill, and R. Dutta, “The human factor: A challenge for network reliability design,” in Intl Conf. on the Design of Reliable Communication Networks, 2015, pp. 115–118

work page 2015
[4]

See You Next Time: A Model for Modern Shoulder Surfers,

O. Wiese and V. Roth, “See You Next Time: A Model for Modern Shoulder Surfers,” in Proceedings of the 18th Interna- tional Conference on Human-Computer Interaction with Mobile Devices and Services, ser. MobileHCI ’16. New York, NY, USA: Association for Computing Machinery, 2016, p. 453–464

work page 2016
[5]

PrivacyScout: Assessing Vulnerability to Shoulder Surfing on Mobile Devices,

M. Bâce, A. Saad, M. Khamis, S. Schneegass, and A. Bulling, “PrivacyScout: Assessing Vulnerability to Shoulder Surfing on Mobile Devices,” Proceedings on Privacy Enhancing Technolo- gies, 2022

work page 2022
[6]

Hollow-Pass: A Dual-View Pattern Password Against Shoulder-Surfing Attacks,

J. Tan and D. K. Sarmah, “Hollow-Pass: A Dual-View Pattern Password Against Shoulder-Surfing Attacks,” in Intl Symp. on Cyber Security, Cryptology, and Machine Learning. Springer, 2023, pp. 251–272

work page 2023
[7]

Password-stealing without Hacking: Wi-Fi Enabled Practical Keystroke Eavesdropping,

J. Hu, H. Wang, T. Zheng, J. Hu, Z. Chen, H. Jiang, and J. Luo, “Password-stealing without Hacking: Wi-Fi Enabled Practical Keystroke Eavesdropping,” in Proceedings of the 2023 ACM SIGSAC conference on computer and communications security, 2023, pp. 239–252

work page 2023
[8]

Covert Attentional Shoulder Surf- ing: Human Adversaries Are More Powerful Than Expected,

T. Kwon, S. Shin, and S. Na, “Covert Attentional Shoulder Surf- ing: Human Adversaries Are More Powerful Than Expected,” IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 44, no. 6, pp. 716–727, 2014

work page 2014
[9]

Is Your PIN Safe Against Advanced Human-Centric Shoulder Surfing?

N. Chakraborty and M. Zulkernine, “Is Your PIN Safe Against Advanced Human-Centric Shoulder Surfing?” in IEEE 49th Annual Computers, Software, and Applications Conf. IEEE, 2025, pp. 2307–2312

work page 2025
[10]

On Designing a Lesser Ob- trusive Authentication Protocol to Prevent Machine-Learning- Based Threats in Internet of Things,

N. Chakraborty, J.-Q. Li, S. Mondal, C. Luo, H. Wang, M. Alazab, F. Chen, and Y. Pan, “On Designing a Lesser Ob- trusive Authentication Protocol to Prevent Machine-Learning- Based Threats in Internet of Things,” IEEE Internet of Things J., vol. 8, no. 5, pp. 3255–3267, 2021

work page 2021
[11]

Study to Improve Security for IoT Smart Device Controller: Drawbacks and Countermeasures,

X. Su, Z. Wang, X. Liu, C. Choi, and D. Choi, “Study to Improve Security for IoT Smart Device Controller: Drawbacks and Countermeasures,” Security and Communication Networks, vol. 2018, no. 1, p. 4296934, 2018

work page 2018
[12]

An Improved Methodology Towards Providing Immunity Against Weak Shoulder Surfing Attack,

N. Chakraborty and S. Mondal, “An Improved Methodology Towards Providing Immunity Against Weak Shoulder Surfing Attack,” in Information Systems Security: 10th International Conference, ICISS 2014, Hyderabad, India, December 16-20, 2014, Proceedings 10. Springer, 2014, pp. 298–317

work page 2014
[13]

A PIN-entry Method Resilient Against Shoulder Surfing,

V. Roth, K. Richter, and R. Freidinger, “A PIN-entry Method Resilient Against Shoulder Surfing,” in Proceedings of the 11th ACM Conference on Computer and Communications Security, 2004, pp. 236–245

work page 2004
[14]

SkullSecurity Password Datasets,

R. Bowes, “SkullSecurity Password Datasets,” https://www. skullsecurity.org/wiki/Passwords, 2010, accessed: 2025-10-30

work page 2010
[15]

Authentication, access control and scalability models in internet of things security–a review,

M. Kokila and S. Reddy, “Authentication, access control and scalability models in internet of things security–a review,” Cyber Security and Applications, vol. 3, 2025

work page 2025
[16]

An Investigation of Shoulder Surfing Attacks on Touch-Based Unlock Events,

S. Schneegass, A. Saad, R. Heger, S. Delgado Rodriguez, R. Poguntke, and F. Alt, “An Investigation of Shoulder Surfing Attacks on Touch-Based Unlock Events,” Proceedings of the ACM on Human-Computer Interaction, vol. 6, no. MHCI, pp. 1–14, 2022

work page 2022
[17]

Stealing Passwords by Observing Hands Movement,

D. Shukla and V. V. Phoha, “Stealing Passwords by Observing Hands Movement,” IEEE Transactions on Information Forensics and Security, vol. 14, no. 12, pp. 3086–3101, 2019

work page 2019
[18]

Thermal Imaging Attacks on Keypad Security Systems,

W. Wodo and L. Hanzlik, “Thermal Imaging Attacks on Keypad Security Systems,” in SECRYPT, 2016, pp. 458–464

work page 2016
[19]

VibroAuth: Authentication with Haptics based Non-Visual, Rearranged Keypads to Mitigate Shoulder Surfing Attacks,

M. Varma, S. Watson, L. Chan, and R. Peiris, “VibroAuth: Authentication with Haptics based Non-Visual, Rearranged Keypads to Mitigate Shoulder Surfing Attacks,” in International Conference on Human-Computer Interaction. Springer, 2022, pp. 280–303

work page 2022
[20]

A Study of Probabilistic Password Models,

J. Ma, W. Yang, M. Luo, and N. Li, “A Study of Probabilistic Password Models,” in 2014 IEEE Symposium on Security and Privacy. IEEE, 2014, pp. 689–704

work page 2014
[21]

A Study on Markov-Based Password Strength Meters,

B. L. T. Thai and H. Tanaka, “A Study on Markov-Based Password Strength Meters,” IEEE Access, vol. 12, pp. 69 066– 69 075, 2024

work page 2024
[22]

Pass- word Cracking Using Probabilistic Context-Free Grammars,

M. Weir, S. Aggarwal, B. De Medeiros, and B. Glodek, “Pass- word Cracking Using Probabilistic Context-Free Grammars,” in 2009 30th IEEE symposium on security and privacy. IEEE, 2009, pp. 391–405

work page 2009
[23]

Understanding Human-Chosen PINs: Characteristics, Distribution and Secu- rity,

D. Wang, Q. Gu, X. Huang, and P. Wang, “Understanding Human-Chosen PINs: Characteristics, Distribution and Secu- rity,” in Proc. of the ACM on Asia Conf. on Computer and Communications Security, 2017, pp. 372–385

work page 2017

[1] [1]

Reliability on the Internet of Things with Designing Approach for Exploratory Analysis,

K. Singh, M. Yadav, Y. Singh, D. Barak, A. Saini, and F. Mor- eira, “Reliability on the Internet of Things with Designing Approach for Exploratory Analysis,” Frontiers in Computer Science, vol. 6, p. 1382347, 2024

work page 2024

[2] [2]

Minding the source: Toward an Integrative Theory of Human-Machine Communication,

E.-J. Lee, “Minding the source: Toward an Integrative Theory of Human-Machine Communication,” Human Communication Research, vol. 50, no. 2, pp. 184–193, 2024

work page 2024

[3] [3]

The human factor: A challenge for network reliability design,

M. Mushi, E. Murphy-Hill, and R. Dutta, “The human factor: A challenge for network reliability design,” in Intl Conf. on the Design of Reliable Communication Networks, 2015, pp. 115–118

work page 2015

[4] [4]

See You Next Time: A Model for Modern Shoulder Surfers,

O. Wiese and V. Roth, “See You Next Time: A Model for Modern Shoulder Surfers,” in Proceedings of the 18th Interna- tional Conference on Human-Computer Interaction with Mobile Devices and Services, ser. MobileHCI ’16. New York, NY, USA: Association for Computing Machinery, 2016, p. 453–464

work page 2016

[5] [5]

PrivacyScout: Assessing Vulnerability to Shoulder Surfing on Mobile Devices,

M. Bâce, A. Saad, M. Khamis, S. Schneegass, and A. Bulling, “PrivacyScout: Assessing Vulnerability to Shoulder Surfing on Mobile Devices,” Proceedings on Privacy Enhancing Technolo- gies, 2022

work page 2022

[6] [6]

Hollow-Pass: A Dual-View Pattern Password Against Shoulder-Surfing Attacks,

J. Tan and D. K. Sarmah, “Hollow-Pass: A Dual-View Pattern Password Against Shoulder-Surfing Attacks,” in Intl Symp. on Cyber Security, Cryptology, and Machine Learning. Springer, 2023, pp. 251–272

work page 2023

[7] [7]

Password-stealing without Hacking: Wi-Fi Enabled Practical Keystroke Eavesdropping,

J. Hu, H. Wang, T. Zheng, J. Hu, Z. Chen, H. Jiang, and J. Luo, “Password-stealing without Hacking: Wi-Fi Enabled Practical Keystroke Eavesdropping,” in Proceedings of the 2023 ACM SIGSAC conference on computer and communications security, 2023, pp. 239–252

work page 2023

[8] [8]

Covert Attentional Shoulder Surf- ing: Human Adversaries Are More Powerful Than Expected,

T. Kwon, S. Shin, and S. Na, “Covert Attentional Shoulder Surf- ing: Human Adversaries Are More Powerful Than Expected,” IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 44, no. 6, pp. 716–727, 2014

work page 2014

[9] [9]

Is Your PIN Safe Against Advanced Human-Centric Shoulder Surfing?

N. Chakraborty and M. Zulkernine, “Is Your PIN Safe Against Advanced Human-Centric Shoulder Surfing?” in IEEE 49th Annual Computers, Software, and Applications Conf. IEEE, 2025, pp. 2307–2312

work page 2025

[10] [10]

On Designing a Lesser Ob- trusive Authentication Protocol to Prevent Machine-Learning- Based Threats in Internet of Things,

N. Chakraborty, J.-Q. Li, S. Mondal, C. Luo, H. Wang, M. Alazab, F. Chen, and Y. Pan, “On Designing a Lesser Ob- trusive Authentication Protocol to Prevent Machine-Learning- Based Threats in Internet of Things,” IEEE Internet of Things J., vol. 8, no. 5, pp. 3255–3267, 2021

work page 2021

[11] [11]

Study to Improve Security for IoT Smart Device Controller: Drawbacks and Countermeasures,

X. Su, Z. Wang, X. Liu, C. Choi, and D. Choi, “Study to Improve Security for IoT Smart Device Controller: Drawbacks and Countermeasures,” Security and Communication Networks, vol. 2018, no. 1, p. 4296934, 2018

work page 2018

[12] [12]

An Improved Methodology Towards Providing Immunity Against Weak Shoulder Surfing Attack,

N. Chakraborty and S. Mondal, “An Improved Methodology Towards Providing Immunity Against Weak Shoulder Surfing Attack,” in Information Systems Security: 10th International Conference, ICISS 2014, Hyderabad, India, December 16-20, 2014, Proceedings 10. Springer, 2014, pp. 298–317

work page 2014

[13] [13]

A PIN-entry Method Resilient Against Shoulder Surfing,

V. Roth, K. Richter, and R. Freidinger, “A PIN-entry Method Resilient Against Shoulder Surfing,” in Proceedings of the 11th ACM Conference on Computer and Communications Security, 2004, pp. 236–245

work page 2004

[14] [14]

SkullSecurity Password Datasets,

R. Bowes, “SkullSecurity Password Datasets,” https://www. skullsecurity.org/wiki/Passwords, 2010, accessed: 2025-10-30

work page 2010

[15] [15]

Authentication, access control and scalability models in internet of things security–a review,

M. Kokila and S. Reddy, “Authentication, access control and scalability models in internet of things security–a review,” Cyber Security and Applications, vol. 3, 2025

work page 2025

[16] [16]

An Investigation of Shoulder Surfing Attacks on Touch-Based Unlock Events,

S. Schneegass, A. Saad, R. Heger, S. Delgado Rodriguez, R. Poguntke, and F. Alt, “An Investigation of Shoulder Surfing Attacks on Touch-Based Unlock Events,” Proceedings of the ACM on Human-Computer Interaction, vol. 6, no. MHCI, pp. 1–14, 2022

work page 2022

[17] [17]

Stealing Passwords by Observing Hands Movement,

D. Shukla and V. V. Phoha, “Stealing Passwords by Observing Hands Movement,” IEEE Transactions on Information Forensics and Security, vol. 14, no. 12, pp. 3086–3101, 2019

work page 2019

[18] [18]

Thermal Imaging Attacks on Keypad Security Systems,

W. Wodo and L. Hanzlik, “Thermal Imaging Attacks on Keypad Security Systems,” in SECRYPT, 2016, pp. 458–464

work page 2016

[19] [19]

VibroAuth: Authentication with Haptics based Non-Visual, Rearranged Keypads to Mitigate Shoulder Surfing Attacks,

M. Varma, S. Watson, L. Chan, and R. Peiris, “VibroAuth: Authentication with Haptics based Non-Visual, Rearranged Keypads to Mitigate Shoulder Surfing Attacks,” in International Conference on Human-Computer Interaction. Springer, 2022, pp. 280–303

work page 2022

[20] [20]

A Study of Probabilistic Password Models,

J. Ma, W. Yang, M. Luo, and N. Li, “A Study of Probabilistic Password Models,” in 2014 IEEE Symposium on Security and Privacy. IEEE, 2014, pp. 689–704

work page 2014

[21] [21]

A Study on Markov-Based Password Strength Meters,

B. L. T. Thai and H. Tanaka, “A Study on Markov-Based Password Strength Meters,” IEEE Access, vol. 12, pp. 69 066– 69 075, 2024

work page 2024

[22] [22]

Pass- word Cracking Using Probabilistic Context-Free Grammars,

M. Weir, S. Aggarwal, B. De Medeiros, and B. Glodek, “Pass- word Cracking Using Probabilistic Context-Free Grammars,” in 2009 30th IEEE symposium on security and privacy. IEEE, 2009, pp. 391–405

work page 2009

[23] [23]

Understanding Human-Chosen PINs: Characteristics, Distribution and Secu- rity,

D. Wang, Q. Gu, X. Huang, and P. Wang, “Understanding Human-Chosen PINs: Characteristics, Distribution and Secu- rity,” in Proc. of the ACM on Asia Conf. on Computer and Communications Security, 2017, pp. 372–385

work page 2017