CoChat
Sign Up
Cited by 0Chaos-based Image/Signal EncryptionCryptographic Implementations and SecurityCryptography and Data Security

A hybrid encryption algorithm based on AES and Kyber for secure image protection in the post-quantum era

Open access
Singh, Preet, Hasija, Taniya, Ramkumar, K. R., Kanika·2026·Discover Computing
DOI 10.1007/s10791-026-10577-0
Open this paper in CoChat — free accountOpen PDF

Abstract

The exponential rise in the sharing of multimedia information is accompanied by an increase in vulnerability during data exchange. As quantum computing is threatening the fundamentals of classical cryptography, it is important that multimedia data be protected by quantum-secured encryption techniques. Hybrid encryption that combines symmetric cyphers with post-quantum key encapsulation is an established architectural pattern, but its use for image data has not been explored in depth. This paper presents Quantum Good Privacy (QGP), a hybrid image-encryption scheme that combines AES-256 in Cipher Block Chaining (CBC) mode with the Module-Lattice-based Key Encapsulation Mechanism (ML-KEM, originally Kyber) and evaluates this architecture on image data through statistical and timing tests. QGP-CBC achieves entropy consistently above 7.996 (mean: 7.9963), approaching the ideal 8 bits/pixel, a low Peak Signal-to-Noise Ratio (8.24 dB), a strong Number of Pixel Change Rate (99.62%), and a Unified Average Changing Intensity (31.67%) value, providing strong confusion and diffusion properties. It is an efficient method as it requires 2.1 ms for encrypting and 0.79 ms for decrypting 256 × 256 image data, respectively, while encapsulation and decapsulation of keys take only 0.068 ms and 0.051 ms, respectively. QGP-CBC passes the NIST SP800-22 randomness test, thus making it suitable for use in secure image transmission in the post-quantum world. QGP-CBC and QGP-ECB have similar Number of Pixel Change Rate and Unified Average Changing Intensity values, but QGP-CBC passes all 16 NIST SP800-22 randomness tests while QGP-ECB fails four, making it the more statistically robust choice for image encryption.

AI snapshot

Written from the paper’s full text.

TL;DR

The paper proposes QGP, a hybrid image-encryption scheme combining AES-256 in CBC or ECB mode with ML-KEM (Kyber) for post-quantum key encapsulation, and evaluates it on grayscale images. QGP-CBC achieves near-ideal entropy and strong differential metrics while passing all 16 NIST SP800-22 tests, with millisecond-level encryption/decryption and key encapsulation/decapsulation times.

Key findings

  • For QGP-CBC on a 256 × 256 grayscale image, mean entropy is 7.9963, PSNR is 8.24 dB, NPCR is 99.61968%, UACI is 31.67%, and MSE is 9741.12.
  • QGP-CBC takes 2.1 ms to encrypt and 0.79 ms to decrypt image data; key encapsulation and decapsulation take 0.068 ms and 0.051 ms, respectively. QGP-CBC image throughput is 22.69 MB/s encryption and 68.57 MB/s decryption, versus QGP-ECB at 27.48 MB/s and 125.84 MB/s.
  • QGP-CBC passes all 16 NIST SP800-22 tests, while QGP-ECB fails four: Discrete Fourier Transform (Spectral), Maurer’s Universal, Serial (both parts), and Approximate Entropy. The two modes have similar NPCR and UACI values.
  • Across 256 × 256, 512 × 512, and 1024 × 1024 images, QGP-CBC entropy rises from 7.9963 to 7.9982 to 7.9991, NPCR from 99.6197% to 99.6364% to 99.6651%, and UACI from 31.67% to 32.02% to 32.88%.

Methods

  • The design is a KEM-DEM hybrid: AES-256 in CBC or ECB mode encrypts image data, while ML-KEM-512 (Kyber) encapsulates the AES session key. It was implemented in C using liboqs version 0.10.1-dev on Ubuntu 64-bit, Intel Core i7 10th generation at 2.60 GHz, 16 GB DDR4 RAM, and a 512 GB SSD.
  • The main test image is a 256 × 256 grayscale Mona Lisa from a public-domain source; additional evaluations use 512 × 512 and 1024 × 1024 images. Metrics include entropy, MSE, PSNR, NPCR, UACI, throughput, encryption/decryption time, and the NIST SP800-22 suite, with timing values averaged over iterations 10–100.
  • NPCR and UACI are computed between cipher images from plain images differing by exactly one pixel under the same session key and, in CBC mode, the same IV; MSE and PSNR are computed between the plain image and its cipher image.

Limitations

  • The paper states that evaluation covers only grayscale, static images; no colour images, video, or other multimedia were tested. It used a limited set of test images generated for the study rather than a standard benchmark dataset due to copyright restrictions, limiting direct pixel-level comparison with prior work. UACI (31.67%) remains below the theoretical optimum (33.4635%), CBC mode is inherently sequential, ML-KEM adds ciphertext/public-key overhead, and validation is software-only on a desktop platform.
  • The paper states that no formal proof of the hybrid KEM-DEM composition is provided; the scheme offers confidentiality only, without a MAC or AEAD, leaving potential vulnerability to ciphertext manipulation. It does not assess noise, bit errors, cropping, partial data loss, timing leakage, power-analysis exposure, or fault-injection vulnerabilities.
  • The paper states that NIST SP800-22 results come from one 524,288-bit ciphertext stream; formal suite validation requires 100 independent sequences and some tests need longer sequences, so the p-values are a like-for-like comparison rather than formal validation. It also notes that direct comparison with recent post-quantum image-encryption methods was not performed because of differing datasets, image sizes, implementation environments, security settings, and limited reproducible implementations.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1007/s10791-026-10577-0

Cites

71 of 71 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

9 of 29 cited sources checked · 3 supported · 0 flagged · accuracy not yet measured

  • Premisecitation

    Quantum computers could break current cryptographic systems.

    “Rapid developments have led to more precise calculations that are made by quantum computers to easily convert the ciphered message, which could break current cryptographic systems [17,18].”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [17] Bernstein DJ, Lange T. Post-quantum cryptography. Nature. 2017;549(7671):188–94. https://doi.org/10.1038/nature23461. source
    • cites [18] Nithila EE, R BL, Nithila EE, C SS, A R. A survey about post quantum cryptography methods. EAI Endorsed Trans IoT. 2024;10. https://doi.org/10.4108/eetiot.5099. source
  • Findingown data

    QGP-ECB fails four NIST SP800-22 randomness tests.

    “while QGP-ECB fails four”
    Quote in the paperSubstantial claim
  • Findingnone given

    QGP-CBC passes the NIST SP800-22 randomness test.

    “QGP-CBC passes the NIST SP800-22 randomness test”
    Quote in the paperSubstantial claim
  • Findingnone given

    QGP-CBC and QGP-ECB have similar Number of Pixel Change Rate and Unified Average Changing Intensity values.

    “QGP-CBC and QGP-ECB have similar Number of Pixel Change Rate and Unified Average Changing Intensity values”
    Quote in the paperSubstantial claim
  • Findingown data

    QGP-CBC achieves a Peak Signal-to-Noise Ratio of 8.24 dB.

    “a low Peak Signal-to-Noise Ratio (8.24 dB)”
    Quote in the paperSubstantial claim
  • May combine two claims. 33
  • Classified as probably not a substantial claim. 11

  • May combine two claims.

    Premisecitation

    ML-KEM security relies on the Module Learning-with-Errors problem, for which there is no known polynomial-time quantum algorithm.

    “its security relies on the Module Learningwith-Errors (M-LWE) problem, for which there is no known polynomial-time quantum algorithm [56,57].”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites [56] Kute S, Desai C, Jadhav M. Analysis of RSA and Shor’s algorithm for cryptography: a quantum perspective. AIP Conf Proc. 2024;3156(1: AIP Publishing LLC, p. 040004. source
    • cites [57] Gerjuoy E. Shor’s factoring algorithm and modern cryptography. An illustration of the capabilities inherent in quantum computers. Am J Phys. 2005;73(6):521–40. source

      The Introduction (p. 1) says the paper addresses RSA and Shor’s factoring algorithm, with other aspects of quantum cryptography and computation beyond its scope. Searches for ML-KEM, Module Learning-with-Errors, and a polynomial-time quantum algorithm for that problem found no passage addressing either part of the claim. (LLM, conf=0.78)

      Experimental — accuracy not yet measured

  • Classified as probably not a substantial claim.

    Premisecitation

    Post-quantum encryption algorithms must be based on math problems that would be difficult for both conventional and quantum computers to solve.

    “post-quantum encryption algorithms must be based on math problems that would be difficult for both conventional and quantum computers to solve [20,21].”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites [20] Opiłka F, Niemiec M, Gagliardi M, Kourtis MA. Performance analysis of post-quantum cryptography algorithms for digital signature. Appl Sci. 2024;14(12):4994. https://doi.org/10.3390/app14124994. source
    • cites [21] Hasija T, Ramkumar K, Singh B, Kaur A, Mittal SK. Applying polynomials for developing Post quantum cryptography algorithms to secure online information-an initial hypothesis. RACSC. 2025;18(5). https://doi.org/10.2174/0126662558331360240924004351. source

  • May combine two claims.

    Premisecitation

    In CBC mode, chaining propagates local changes of plaintext pixels in each block to wipe out spatial correlation, and stronger diffusion is measured as NPCR/UACI values.

    “in this case, the propagation of the chaining of local changes of plaintext pixels in each block will wipe out the spatial correlation, and the stronger diffusion will be measured in this case as NPCR/UACI values [58,59].”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites [58] Vatsavayi VK, Bommireddy DR. Secure and efficient block cipher mode design for parallel processing and reliable security. Cryptography. 2026;10(1):13. https://doi.org/10.3390/cryptography10010013. source

      Section 2.2, Traditional Modes (page unavailable), says CBC chains blocks through the previous ciphertext block. Section 5.1, Empirical Randomness and Diffusion Analysis (page unavailable), reports that CBC-encrypted images hide visible patterns. Our copy does not establish that local plaintext-pixel changes propagate to wipe out spatial correlation, or that CBC diffusion is measured by NPCR/UACI values. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites [59] Lata K, Gupta C, Cenkeramaddi LR. A cryptographic framework for secure medical imaging in smart healthcare environments. Results Eng. 2025;27:106780. https://doi.org/10.1016/j.rineng.2025.106780. source

  • May combine two claims.

    Premisecitation

    Classical hybrid cryptosystems combine a symmetric image cipher with asymmetric key transport such as ElGamal, ECC, Hill cipher variants or RSA, all of which are broken in polynomial time by Shor's algorithm.

    “Second, classical hybrid cryptosystems [32][33][34][35][36][37][38] consist of a combination of a symmetric image cypher and asymmetric key transport, which is either ElGamal, ECC, Hill cypher variants or RSA; all of which are the weakest links broken in polynomial time by Shor's algorithm.”
    Quote in the paperSubstantial claim4 of 7 sources checked
    • cites [32] Gao X. Image encryption algorithm based on 2D hyperchaotic map. Optics Laser Technol. 2021;142:107252. https://doi.org/10.1016/j.optlastec.2021.107252. source
    • cites [33] Li T, Yang M, Wu J, Jing X. A novel image encryption algorithm based on a fractional-order Hyperchaotic system and DNA computing. Complexity. 2017;2017(1):9010251. https://doi.org/10.1155/2017/9010251. source

      The Introduction mentions a symmetric image-encryption scheme, while §4.2 discusses the proposed algorithm’s secret keys. Neither passage addresses hybrid cryptosystems, asymmetric key transport, or Shor’s algorithm; searches for those terms and the named algorithms found no supporting passage in our copy. (LLM, conf=0.76)

      Experimental — accuracy not yet measured
    • cites [34] Wu T-Y, Fan X, Wang K-H, Lai C-F, Xiong N, Wu JM-T. A dna computation-based image encryption scheme for cloud cctv systems. IEEE Access. 2019;7:181434–43. https://doi.org/10.1109/ACCESS.2019.2946890. source

      The Introduction and Conclusion describe an image-encryption scheme based on hyperchaotic maps and DNA coding, not a classical hybrid cryptosystem. Searches for asymmetric key transport, ElGamal, ECC, Hill cipher variants, RSA, and Shor’s algorithm found no passage in our copy addressing the attributed claim. (LLM, conf=0.76)

      Experimental — accuracy not yet measured
    • cites [35] Hardi S, Tarigan J, Safrina N. Hybrid cryptosystem for image file using elgamal and double Playfair cipher algorithm. J Phys Conf Ser. 2018;978(1: IOP Publishing, p. 012068. https://doi.org/10.1088/1742-6596/978/1/012068. source

      The Introduction and Implementation sections describe a hybrid image cryptosystem that encrypts the image with symmetric Double Playfair and its keys with asymmetric ElGamal; Related Works also gives RSA as a key-encryption example. The paper does not describe Hill cipher variants as asymmetric key transport or say that these schemes are broken in polynomial time by Shor’s algorithm. Page numbers were not available in our copy. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites [36] Dawahdeh ZE, Yaakob SN, Othman RR. A new image encryption technique combining elliptic curve cryptosystem with Hill Cipher. J King Saud Univ - Comput Inf Sci. 2018;30(3):349–55. https://doi.org/10.1016/j.jksuci.2017.06.004. source

      The Introduction and Conclusions describe an image-encryption scheme combining the symmetric Hill cipher with ECC-based key generation. They say the parties produce a secret key without transmitting it, rather than describing asymmetric key transport. The paper does not address Shor’s algorithm or claim polynomial-time attacks on the listed cryptosystems. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites [37] Wang X, Li Y. Chaotic image encryption algorithm based on hybrid multi-objective particle swarm optimization and DNA sequence. Optics Lasers Eng. 2021;137:106393. https://doi.org/10.1016/j.optlaseng.2020.106393. source
    • cites [38] Nahta R, Jain T, Narwaria A, Kumar H, Gupta RK. Image encryption decryption using simple and modified version of AES. In: Smart systems and IoT: innovations in computing: proceeding of SSIC 2019. Springer; 2019. p. 641–53. source

  • May combine two claims.

    Findingown data

    QGP-CBC achieves entropy consistently above 7.996, with a mean of 7.9963, approaching the ideal 8 bits/pixel.

    “QGP-CBC achieves entropy consistently above 7.996 (mean: 7.9963), approaching the ideal 8 bits/pixel”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    For image encryption, average time is less for QGP-ECB at 0.001746 seconds, whereas QGP-CBC averages 0.0021146 seconds, indicating QGP-ECB is faster.

    “For image encryption, the average time is less for QGP-ECB at 0.001746 seconds, whereas QGP-CBC is appended with an average of 0.0021146 seconds, indicating that in the process of content encryption, QGP-ECB is faster”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    High NPCR and UACI values for QGP-ECB, with mean values of 99.59% and 31.4%-31.7%, make the attack-resistance attribute of the algorithm stronger.

    “High NPCR and UACI values, with mean values of 99.59% and 31.4%-31.7%,makes the attack-resistance attribute of the algorithm stronger.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    UACI values for QGP-CBC range between 31.46% and 31.83%, aligning with ideal UACI benchmarks (around 33%), further reinforcing the algorithm's ability to introduce significant intensity variations.

    “UACI values range between 31.46% and 31.83%,which aligns with ideal UACI benchmarks (around 33%), further reinforcing the algorithm's ability to introduce significant intensity variations.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    NPCR (99.61968%) is near-optimal, but UACI (31.67%) stays slightly below the theoretical optimum (33.4635%), so there is room to strengthen pixel-intensity diffusion.

    “NPCR (99.61968%) is near-optimal, but UACI (31.67%) stays slightly below the theoretical optimum (33.4635%), so there's room to strengthen pixel-intensity diffusion.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Entropy values for QGP-CBC consistently stay around 7.995-7.997, close to the ideal value of 8, signifying high randomness and unpredictability in the encrypted images, making statistical attacks ineffective.

    “Entropy values consistently stay around 7.995-7.997,which is very close to the ideal value of 8, signifying a high level of randomness and unpredictability in the encrypted images, making statistical attacks ineffective.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    NPCR values for QGP-CBC are consistently high, ranging from 99.58007% to 99.68356%, suggesting that a tiny change in the original image results in a large number of pixel changes in the encrypted output, highlighting strong resistance against differential attacks.

    “NPCR values are consistently high, ranging from 99.58007% to 99.68356%, suggesting that a tiny change in the original image results in a large number of pixel changes in the encrypted output, highlighting strong resistance against differential attacks.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    QGP-ECB passes some NIST SP800-22 tests but fails four important ones: Discrete Fourier Transform (Spectral), Maurer's Universal, Serial (both parts), and Approximate Entropy.

    “The tests confirm that the QGP-ECB pass some tests but fails in four important ones: the Discrete Fourier Transform (Spectral) test, Maurer's Universal test, Serial test (both parts) and the Approximate Entropy test.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    QGP-CBC encryption yields entropy consistently above 7.996, with a mean of 7.9963.

    “Experimental analyses yield entropy consistently above 7.996 (mean: 7.9963), MSE of 9741.12, and PSNR of 8.24 dB for post-encryption has become a severe level of image distortion.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    PSNR values for QGP-ECB rest within a narrow margin of around 8.2-8.3 dB, asserting high distortion caused by strong encryption mechanisms.

    “The PSNR values rest within a narrow margin of around 8.2-8.3 dB, further asserting high distortion caused by strong encryption mechanisms.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Entropy for QGP-ECB sustains its value at around 8 through all instances, meaning the encrypted images are random.

    “Entropy sustains its value at around 8 through all the instances, which is to say that the encrypted images are random.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisenone given

    Chaos-and DNA-based schemes provide high statistical security, but their security depends on the classical channel secrecy of chaotic keys used as keys, which can be broken by quantum adversaries.

    “These chaos-and DNA-based schemes provide a high order of statistical security, but their security depends on the classical channel secrecy of chaotic keys used as keys, which can be broken by quantum adversaries.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisereasoning

    In ECB mode, identical plaintext blocks are mapped to identical ciphertext blocks, so large homogeneous areas of natural images can leak structural patterns.

    “identical plaintext blocks are mapped to identical ciphertext blocks, which means large homogeneous areas of natural images can leak structural patterns.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Key security relies on the hardness of the Learning with Errors (LWE) problem, which remains computationally infeasible even on a quantum computer.

    “Key security relies on the hardness of the Learning with Errors (LWE) problem, and it will remain computationally infeasible even on a quantum computer”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [60] Regev O. On lattices, learning with errors, random linear codes, and cryptography. J Acm. 2009;56(6):1–40. https://doi.org/10.1145/1568318.1568324. source
    • cites [61] Reazaei P, Rashid ASK, Mand AA, Saberikamarposhti M. Post-quantum Key encapsulation mechanisms: a comparative review of ML-KEM, BIKE, and HQC. 2026 22nd IEEE International Colloquium on Signal Processing & Its Applications (CSPA). IEEE; 2026, pp. 342–47. source

  • May combine two claims.

    Interpretationreasoning

    The proposed system's key exchange remains resistant to quantum attacks, and the AES key itself is never transmitted.

    “This way, the key exchange stays resistant to quantum attacks, and the AES key itself is never transmitted.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The recipient's public key is authentic and not substituted.

  • Classified as probably not a substantial claim. May combine two claims.

    Interpretationreasoning

    QGP-CBC is the preferred mode, adding a small additional time cost on encryption of image data in exchange for semantic security.

    “QGP-CBC is the preferred mode, which puts a small additional time cost on encryption of image data in exchange for semantic security.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    CRYSTALS-Kyber has been recently renamed Module-Lattice Based Key Encapsulation Mechanism.

    “One of the algorithms implemented in this research is the CRYSTALS-Kyber, recently renamed Module-Lattice Based Key Encapsulation Mechanism [22].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [22] Sharma S, Ramkumar K, Kaur A, Hasija T, Mittal S, Singh B. Post-quantum cryptography: a solution to the challenges of classical encryption algorithms. Modern electronics devices and communication systems: select proceedings of MEDCOM 2021. 2023, pp. 23–38. source

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    The use of hybrid encryption for image data has not been explored in depth.

    “its use for image data has not been explored in depth.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    The QGP framework validates cipher images and benchmarks performance to the millisecond level against the NIST SP800-22 specification.

    “validate cipher images and benchmark performance to the millisecond level against the NIST SP800-22 specification”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    No prior work has reported this type of evaluation for AES+ML-KEM hybrid schemes.

    “No prior work has reported this type of evaluation for AES+ML-KEM hybrid schemes.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Recovering the secret s from the public key requires solving the M-LWE problem, which stays hard even for quantum adversaries.

    “Recovering s from the public key requires solving the M-LWE problem, which stays hard even for quantum adversaries [61].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [61] Reazaei P, Rashid ASK, Mand AA, Saberikamarposhti M. Post-quantum Key encapsulation mechanisms: a comparative review of ML-KEM, BIKE, and HQC. 2026 22nd IEEE International Colloquium on Signal Processing & Its Applications (CSPA). IEEE; 2026, pp. 342–47. source

  • May combine two claims.

    Interpretationreasoning

    The cost of using larger ML-KEM parameter sets is limited to the overhead of transporting the key and ciphertext and a small increase in encapsulation time, which do not increase with image size.

    “The cost is limited to the overhead of transporting the key and the ciphertext and to a small increase in encapsulation time, which do not increase with image size.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductnone given

    The current implementation provides confidentiality only; it lacks an integrity or authentication mechanism such as a MAC or AEAD construction, which leaves it potentially vulnerable to ciphertext manipulation if deployed without extra safeguards.

    “The current implementation also provides confidentiality only, it lacks an integrity or authentication mechanism such as a MAC or AEAD construction, which leaves it potentially vulnerable to ciphertext manipulation if deployed without extra safeguards.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    The present work is software-only validation on a desktop platform.

    “Although these results are promising, the present work has some limitations, including its restriction to grayscale, static images; a UACI value (31.67%) slightly below the theoretical optimum (33.4635%); the inherently sequential nature of CBC mode; software-only validation on a desktop platform; and the lack of a formal security proof and side-channel testing.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Study design and conductnone given

    CBC mode is inherently sequential.

    “Although these results are promising, the present work has some limitations, including its restriction to grayscale, static images; a UACI value (31.67%) slightly below the theoretical optimum (33.4635%); the inherently sequential nature of CBC mode; software-only validation on a desktop platform; and the lack of a formal security proof and side-channel testing.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    QGP-CBC removes block-level structural patterns that remain in QGP-ECB, making it the better choice for applications that need strong statistical security.

    “This shows that QGP-CBC removes block-level structural patterns that remain in QGP-ECB, making it the better choice for applications that need strong statistical security.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    QGP-CBC forms a secure, efficient, and forward-compatible encryption framework suited for multimedia protection in the post-quantum era.

    “In summary, QGP-CBC forms a secure, efficient, and forward-compatible encryption framework suited for multimedia protection in the post-quantum era.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    The contribution of QGP is the replacement of the quantum-vulnerable key-transport layer with a lattice-based KEM standardised in FIPS 203, together with an image-specific experimental methodology that verifies the security and efficiency of the complete pipeline.

    “the contribution of QGP is not merely the comparison of a symmetric and an asymmetric primitive, as in earlier AES-RSA or ECC-based hybrids, but the replacement of the quantum-vulnerable key-transport layer with a lattice-based KEM standardised in Federal Information Processing Standards (FIPS 203), together with an image-specific experimental methodology that verifies the security and efficiency”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The Learning-With-Errors (LWE) problem, introduced by Regev, is the hardness assumption behind lattice-based cryptography.

    “The Learning-With-Errors (LWE) problem, introduced by Regev [60], is the hardness assumption behind lattice-based cryptography.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [60] Regev O. On lattices, learning with errors, random linear codes, and cryptography. J Acm. 2009;56(6):1–40. https://doi.org/10.1145/1568318.1568324. source

  • May combine two claims.

    Study design and conductcitation

    ML-KEM-512 was selected because it provides NIST security category 1 with the smallest key and ciphertext sizes (800-byte public key, 768-byte ciphertext) and the lowest latency, matching the real-time objective.

    “ML-KEM-512 was selected as it provides NIST security category 1 with the smallest key and ciphertext sizes (800-byte public key, 768-byte ciphertext) [21] and the lowest latency, matching the real-time objective of this work”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [21] Hasija T, Ramkumar K, Singh B, Kaur A, Mittal SK. Applying polynomials for developing Post quantum cryptography algorithms to secure online information-an initial hypothesis. RACSC. 2025;18(5). https://doi.org/10.2174/0126662558331360240924004351. source

  • May combine two claims.

    Study design and conductnone given

    The present work lacks a formal security proof and side-channel testing.

    “Although these results are promising, the present work has some limitations, including its restriction to grayscale, static images; a UACI value (31.67%) slightly below the theoretical optimum (33.4635%); the inherently sequential nature of CBC mode; software-only validation on a desktop platform; and the lack of a formal security proof and side-channel testing.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    QGP-CBC is cryptographically strong and random.

    “Also, passing all the NIST SP800-22 statistical test suites makes QGP-CBC cryptographically strong and, hence, random in contrast to its classical counterparts.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Passing NIST SP800-22 statistical tests indicates cryptographic strength and randomness.

  • May combine two claims.

    Interpretationcitation

    The construction's post-quantum security guarantee comes from the hardness of M-LWE, which classical KEMs based on factoring or discrete logarithms (e.g., RSA, ECC) cannot offer, since Shor's algorithm solves those problems in polynomial time but has no known effect on lattice-based hardness assumptions.

    “This is what gives the construction its post-quantum security guarantee, something classical KEMs based on factoring or discrete logarithms (e.g., RSA, ECC) cannot offer, since Shor's algorithm solves those problems in polynomial time but has no known effect on lattice-based hardness assumptions [62].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [62] Alagic G, Dang Q, Moody D, Robinson A, Silberg H, Smith-Tone D. Module-lattice-based key-encapsulation mechanism standard,” 2024. source

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given

    The framework is agnostic to the KEM instantiation.

    “The framework is agnostic to the KEM instantiation.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    The current evaluation only covers grayscale, static images.

    “The current evaluation only covers grayscale, static images.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    Successful decryption confirms data confidentiality, integrity, and security against both classical and quantum attacks.

    “The successful decryption confirms data confidentiality, integrity, and security against both classical and quantum attacks.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    ML-KEM-512 has the smallest public key and smallest ciphertext (800 and 768 bytes, respectively) and the same latency as real-time.

    “ML-KEM-512 (NIST security category 1) because it has the smallest public key and smallest ciphertext (800 and 768 bytes, respectively) and the same latency as real-time”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [21] Hasija T, Ramkumar K, Singh B, Kaur A, Mittal SK. Applying polynomials for developing Post quantum cryptography algorithms to secure online information-an initial hypothesis. RACSC. 2025;18(5). https://doi.org/10.2174/0126662558331360240924004351. source

  • May combine two claims.

    Interpretationreasoning

    ML-KEM-768 or ML-KEM-1024 can be used as an alternative by setting only the liboqs algorithm identifier, because both parameter sets generate a 256-bit shared secret, and the AES-CBC data-encapsulation stage, KDF interface and image-processing pipeline remain the same.

    “ML-KEM-768 (category 3) or ML-KEM-1024 (category 5) can be used as an alternative, setting only the liboqs algorithm identifier, because both parameter sets generate a 256-bit shared secret, and the AES-CBC data-encapsulation stage, KDF interface and image-processing pipeline remain the same.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    The present work is restricted to grayscale, static images.

    “Although these results are promising, the present work has some limitations, including its restriction to grayscale, static images; a UACI value (31.67%) slightly below the theoretical optimum (33.4635%); the inherently sequential nature of CBC mode; software-only validation on a desktop platform; and the lack of a formal security proof and side-channel testing.”
    Quote in the paperNot a substantial claim
43 of 43

  • The paper relies on this without support.

    AES-256's 256-bit key is post-quantum secure against exhaustive search with Grover's algorithm.

    “its 256-bit key is post-quantum secure against exhaustive search with Grover's algorithm.”

  • The paper relies on this without support.

    AES-256 is the NIST-recommended symmetric primitive for the post-quantum transition.

    “it is the NIST-recommended symmetric primitive for the post-quantum transition”

  • The paper relies on this without support.

    NIST has finalized three post-quantum safe cryptographic algorithms.

    “the National Institute of Standards and Technology (NIST) has finalized the 3 post-Quantum safe cryptographic algorithms.”

  • The paper relies on this without support.

    Passing the NIST SP800-22 randomness test is sufficient for suitability for secure image transmission in the post-quantum world.

    “QGP-CBC passes the NIST SP800-22 randomness test, thus making it suitable for use in secure image transmission in the post-quantum world.”

  • The paper relies on this without support.

    The KDF used to derive the AES session key from the shared secret is secure.

    “The AES session key is then derived from this shared secret using KDF.”

  • The paper relies on this without support.

    The Open Quantum Safe Library implementation of Kyber is correct and secure.

    “The Kyber key pair (public and private keys) is generated using the C language Open Quantum Safe Library.”

  • The paper relies on this without support.

    The recipient's public key is authentic and not substituted.

    “The sender uses the recipient's public key to encapsulate a shared secret with the Kyber algorithm”

  • The paper relies on this without support.

    The AES-256 implementation is correct and secure.

    “Each block undergoes the AES-256 encryption pipeline after that”

  • The paper relies on this without support.

    The random number generator used to generate the AES key and IV is cryptographically secure.

    “creating a random 256-bit AES key K_AES from a perfectly uniform distribution”

  • The paper relies on this without support.

    Conventional symmetric schemes are inadequate in post-quantum cases.

    “With the growing inadequacy of conventional symmetric schemes in the post-quantum cases, this solution has been considered to warrant its practical utility and security.”

  • The paper relies on this without support.

    Quantum computing threats to image encryption are real or current.

    “This work proposes QGP-CBC, a hybrid cryptography solution meant to secure image encryption for quantum computing threats.”

  • The paper relies on this without support.

    AES in CBC mode provides classical efficiency and ML-KEM (Kyber) provides quantum resistance.

    “The model offers the best of both quantum and modern era by combining classical efficiency and quantum resistance, i.e., AES in CBC mode and ML-KEM (Kyber).”

  • The paper relies on this without support.

    The theoretical optimum UACI is 33.4635%.

    “a UACI value (31.67%) slightly below the theoretical optimum (33.4635%)”

  • The paper relies on this without support.

    Passing NIST SP800-22 statistical tests indicates cryptographic strength and randomness.

    “Also, passing all the NIST SP800-22 statistical test suites makes QGP-CBC cryptographically strong and, hence, random in contrast to its classical counterparts.”
14 of 14

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Integrity claim conflicts with the acknowledged lack of authentication

    Major

    The paper says successful decryption confirms integrity, but later states that the implementation provides confidentiality only and has no integrity mechanism.

  • Noise and loss robustness is claimed despite no such assessment

    Major

    Table 15 claims quantified tolerance to pixel noise and pixel loss. The limitations section explicitly says that tolerance to noise and partial data loss was not assessed.

  • AES session-key generation is described in two incompatible ways

    Major

    One account derives the AES key from the secret produced by Kyber encapsulation. The algorithm description instead generates an independent random AES key and says Kyber encapsulates that key, leaving the implemented key-establishment procedure unclear.

  • The stated image size for all experiments conflicts with later size tests

    Minor

    The implementation section says the test image for all experiments is 256 × 256, whereas the paper also reports experiments on 512 × 512 and 1024 × 1024 images.

  • Key encapsulation average is tenfold too large in milliseconds

    Minor

    The reported CBC average of 0.0000678 seconds is 0.0678 ms, consistent with the other stated 0.068 ms value, not 0.68 ms.

    Reported 0.68 · recomputed 0.0678

  • CBC entropy is not consistently above 7.996

    Minor

    Table 6 reports a CBC entropy of 7.995914 at iteration 50, below the claimed lower bound.

6 of 6

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. Schneier B, Kelsey J, Whiting D, Wagner D, Hall C, Ferguson N. Twofish: a 128-bit block cipher. NIST AES Proposal. 1998;15(1):23–91.

Authors
Singh, P., Hasija, T., Ramkumar, K. R., & Kanika
Affiliations
  • Chitkara University (IN)
  • Manipal University Jaipur
Corresponding author
Kanika
ORCID
Hasija, Taniya
Venue
Discover Computing
Publisher
Springer Science and Business Media LLC
Published
2026-10-07
Volume
29
Issue
1
ISSN
2948-2992 (ISSN-L)
DOI
10.1007/s10791-026-10577-0
OpenAlex
W7220854476
Semantic Scholar
09ab343edc08c6f26d9ea06c85a0fb2ff4189093
Cited by
0
References
69
FWCI
0.00
Type
journal article
Language
en
Open access
Licence
Locations
  • Discover Computing (publishedVersion, cc-by-nc-nd)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
Topics
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Chaos-based Image/Signal Encryption
  • Physical Sciences › Computer Science › Artificial Intelligence › Cryptographic Implementations and Security
  • Physical Sciences › Computer Science › Artificial Intelligence › Cryptography and Data Security
Keywords
AES-256, Kyber, post-quantum cryptography, CBC mode, ECB mode, image security, quantum computing threats to cryptography, confusion and diffusion, secure image transmission

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-23.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-23.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review