Selected work

Representative Papers

Beyond the QBER Threshold: A Temporal QBER Based Machine Learning Framework for Multi Attack Detection in BB84 QKD

Aug 04, 2026

This work addresses the vulnerability of the conventional BB84 quantum key distribution protocol, which relies on a fixed 11% quantum bit error rate (QBER) threshold and fails to detect covert eavesdropping attacks occurring below this limit. To overcome this limitation, the authors propose a machine learning framework leveraging temporal QBER dynamics, introducing for the first time a 63-dimensional feature set derived from physical-layer time-series characteristics alongside interpretability analysis. The approach transcends static threshold constraints and enables highly sensitive detection of multiple attack types. Evaluated using XGBoost, Random Forest, and SVM-RBF classifiers, the framework achieves superior performance, with XGBoost attaining an accuracy of 88.01% and a macro F1-score of 0.8803 in multi-attack scenarios, while reducing the false negative rate from 0.8477 to 0.0198.

0 citationsRead paper

A Survey on Sensor-based Planning and Control for Unmanned Underwater Vehicles

Apr 05, 2026

This study addresses the navigation and control challenges faced by unmanned underwater vehicles (UUVs) in complex underwater environments characterized by GNSS denial, sensor noise, and limited communication. It provides a systematic review and classification of local reactive planning and control methods that leverage real-time sensing data from sonar, inertial measurement units (IMUs), and other onboard sensors. The work innovatively proposes an architectural taxonomy that distinguishes between decoupled and coupled planning–control paradigms, emphasizing adaptive local replanning mechanisms. By integrating strategies such as PID control, model predictive control (MPC), and invariant-set-based control, the paper offers a thorough analysis of the trade-offs among path optimality, computational cost, safety, and maneuverability, thereby establishing a theoretical foundation and practical design guidance for enhancing UUV autonomy.

0 citationsRead paper

On the generalization of $g$-circulant MDS matrices

Feb 10, 2026

This study addresses the enhancement of data diffusion in cryptography and coding theory by investigating the construction and verification of generalized $g$-circulant matrices with maximum distance separable (MDS) properties. The work introduces a novel structure termed consta-$g$-circulant matrices and establishes criteria for their invertibility and MDS property through a connection with polynomial factorization over finite fields. A complete characterization is provided for the cases of order 3 and 4, and inspired by skew polynomial rings, new variants are constructed. The paper derives an exact counting formula for invertible consta-$g$-circulant matrices, substantially reducing the enumeration complexity required for MDS verification. Theoretical findings are corroborated through concrete examples.

0 citationsRead paper

Self Distillation via Iterative Constructive Perturbations

May 20, 2025

Balancing training performance and generalization remains challenging in deep neural network optimization. To address this, we propose a cyclic optimization framework that jointly updates model parameters and input data. Our core innovation is the Iterative Constructive Perturbation (ICP) mechanism: it generates input perturbations guided by model loss, while integrating self-distillation and intermediate-layer feature alignment to establish a bidirectional model–data adaptation paradigm. This approach unifies loss-driven input reconstruction with progressive knowledge transfer, effectively mitigating overfitting and training stagnation. Extensive experiments across diverse training regimes—including standard supervised learning, label-noise robustness, and few-shot learning—demonstrate consistent improvements in both accuracy and generalization. The framework exhibits strong robustness and broad applicability, validating its effectiveness beyond specific task assumptions.

0 citationsRead paper

Complete Key Recovery of a DNA-based Encryption and Developing a Novel Stream Cipher for Color Image Encryption: Bio-SNOW

Mar 10, 2025

This paper addresses critical security vulnerabilities in existing DNA-based encryption algorithms—specifically, susceptibility to key-recovery attacks and low avalanche effect—by proposing the first complete key-recovery attack requiring only two plaintext–ciphertext pairs with constant time complexity O(1). Building on this cryptanalysis, we design Bio-SNOW, a novel biologically inspired stream cipher that integrates DNA encoding/decoding, chaotic mapping, and the SNOW-3G framework, augmented by a biologically inspired S-box to enhance nonlinearity and resistance to cryptanalysis. Experimental evaluation demonstrates that Bio-SNOW passes all NIST statistical randomness tests, achieves twice the encryption throughput of SNOW-3G, and exhibits significantly improved avalanche effect and key sensitivity. The scheme thus offers high security, strong robustness, and lightweight computational overhead, making it particularly suitable for resource-constrained IoT applications and image encryption.

0 citationsRead paper

Recent publications

Latest Papers

Beyond the QBER Threshold: A Temporal QBER Based Machine Learning Framework for Multi Attack Detection in BB84 QKD

Aug 04, 2026

This work addresses the vulnerability of the conventional BB84 quantum key distribution protocol, which relies on a fixed 11% quantum bit error rate (QBER) threshold and fails to detect covert eavesdropping attacks occurring below this limit. To overcome this limitation, the authors propose a machine learning framework leveraging temporal QBER dynamics, introducing for the first time a 63-dimensional feature set derived from physical-layer time-series characteristics alongside interpretability analysis. The approach transcends static threshold constraints and enables highly sensitive detection of multiple attack types. Evaluated using XGBoost, Random Forest, and SVM-RBF classifiers, the framework achieves superior performance, with XGBoost attaining an accuracy of 88.01% and a macro F1-score of 0.8803 in multi-attack scenarios, while reducing the false negative rate from 0.8477 to 0.0198.

0 citationsRead paper

A Survey on Sensor-based Planning and Control for Unmanned Underwater Vehicles

Apr 05, 2026

This study addresses the navigation and control challenges faced by unmanned underwater vehicles (UUVs) in complex underwater environments characterized by GNSS denial, sensor noise, and limited communication. It provides a systematic review and classification of local reactive planning and control methods that leverage real-time sensing data from sonar, inertial measurement units (IMUs), and other onboard sensors. The work innovatively proposes an architectural taxonomy that distinguishes between decoupled and coupled planning–control paradigms, emphasizing adaptive local replanning mechanisms. By integrating strategies such as PID control, model predictive control (MPC), and invariant-set-based control, the paper offers a thorough analysis of the trade-offs among path optimality, computational cost, safety, and maneuverability, thereby establishing a theoretical foundation and practical design guidance for enhancing UUV autonomy.

0 citationsRead paper

On the generalization of $g$-circulant MDS matrices

Feb 10, 2026

This study addresses the enhancement of data diffusion in cryptography and coding theory by investigating the construction and verification of generalized $g$-circulant matrices with maximum distance separable (MDS) properties. The work introduces a novel structure termed consta-$g$-circulant matrices and establishes criteria for their invertibility and MDS property through a connection with polynomial factorization over finite fields. A complete characterization is provided for the cases of order 3 and 4, and inspired by skew polynomial rings, new variants are constructed. The paper derives an exact counting formula for invertible consta-$g$-circulant matrices, substantially reducing the enumeration complexity required for MDS verification. Theoretical findings are corroborated through concrete examples.

0 citationsRead paper

Self Distillation via Iterative Constructive Perturbations

May 20, 2025

Balancing training performance and generalization remains challenging in deep neural network optimization. To address this, we propose a cyclic optimization framework that jointly updates model parameters and input data. Our core innovation is the Iterative Constructive Perturbation (ICP) mechanism: it generates input perturbations guided by model loss, while integrating self-distillation and intermediate-layer feature alignment to establish a bidirectional model–data adaptation paradigm. This approach unifies loss-driven input reconstruction with progressive knowledge transfer, effectively mitigating overfitting and training stagnation. Extensive experiments across diverse training regimes—including standard supervised learning, label-noise robustness, and few-shot learning—demonstrate consistent improvements in both accuracy and generalization. The framework exhibits strong robustness and broad applicability, validating its effectiveness beyond specific task assumptions.

0 citationsRead paper

Complete Key Recovery of a DNA-based Encryption and Developing a Novel Stream Cipher for Color Image Encryption: Bio-SNOW

Mar 10, 2025

This paper addresses critical security vulnerabilities in existing DNA-based encryption algorithms—specifically, susceptibility to key-recovery attacks and low avalanche effect—by proposing the first complete key-recovery attack requiring only two plaintext–ciphertext pairs with constant time complexity O(1). Building on this cryptanalysis, we design Bio-SNOW, a novel biologically inspired stream cipher that integrates DNA encoding/decoding, chaotic mapping, and the SNOW-3G framework, augmented by a biologically inspired S-box to enhance nonlinearity and resistance to cryptanalysis. Experimental evaluation demonstrates that Bio-SNOW passes all NIST statistical randomness tests, achieves twice the encryption throughput of SNOW-3G, and exhibits significantly improved avalanche effect and key sensitivity. The scheme thus offers high security, strong robustness, and lightweight computational overhead, making it particularly suitable for resource-constrained IoT applications and image encryption.

0 citationsRead paper