Xuan Sun
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xuan Sun has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Xiaoying Qiu
College of Computer Science, Beijing Information Science and Technology University, Beijing, China
4 shared publications - Jun Li
School of Computer Science, Beijing Information Science and Technology University, Beijing, China
3 shared publications - Ben Qian
Beijing Information Science and Technology University
2 shared publications - Caixia Li
Beijing Information Science and Technology University
2 shared publications - Jinwei Yu
School of Information and Communication Engineering, Beijing University of Posts and Telecommunications, Beijing 100876, China
2 shared publications - Mengyan Qiao
Beijing Information Science and Technology University
2 shared publications
Latest Publications
- 21 March 2026
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 17 (5) , 215
Research on the construction method of industrial internet risk assessment indicator system based On LLMs
- 21 March 2026
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 17 (5) , 215
Research on the construction method of industrial internet risk assessment indicator system based On LLMs
- Open Access
- 8 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 13-18
Fuzz Testing for Polymorphic Network Firmware: A Collaborative Optimization Framework of LLMs and Neural Networks
- 8 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 13-18
Open AccessFuzz Testing for Polymorphic Network Firmware: A Collaborative Optimization Framework of LLMs and Neural Networks
- Open Access
- 25 October 2025
- Published by Springer Nature in Cybersecurity
- Vol. 8 (1) , 1-16
Classifying encrypted traffic using quad-directional convolution on pulse sequences
With the growing concerns for user privacy and communication security, the volume of encrypted traffic has surged, making accurate classification of encrypted traffic crucial. Current mainstream classification methods primarily rely on handcrafted features and machine learning techniques. However, these features often depend on expert knowledge and require substantial human resources. Additionally, using simple features can lead to inadequate feature extraction. To address these challenges, we have developed the Pulse-Driven Quad-Directional Temporal Convolutional Network. A core innovation of this model is transforming one-dimensional packet length sequences into two-dimensional pulse sequences, significantly enhancing the feature representation capability and classification accuracy of the model. Furthermore, the model employs forward, backward, middle-outward, and both-ends-inward temporal convolutions to effectively extract global features. Coupled with the self-attention mechanism, the model deeply explores the dependencies among features, greatly enhancing classification accuracy and robustness. On the public CESNET-TLS22 dataset, this method achieved an F1 score of 99.24% in the classification of five application services and an average F1 score of 98.74% in the 20-application classification test. Additionally, on the UNB ISCX dataset, tests on VPN and Tor traffic also demonstrated outstanding performance, with F1 scores exceeding 98%. These results highlight the precision and effectiveness of this method across various application scenarios.
- 25 October 2025
- Published by Springer Nature in Cybersecurity
- Vol. 8 (1) , 1-16
Open AccessClassifying encrypted traffic using quad-directional convolution on pulse sequences
With the growing concerns for user privacy and communication security, the volume of encrypted traffic has surged, making accurate classification of encrypted traffic crucial. Current mainstream classification methods primarily rely on handcrafted features and machine learning techniques. However, these features often depend on expert knowledge and require substantial human resources. Additionally, using simple features can lead to inadequate feature extraction. To address these challenges, we have developed the Pulse-Driven Quad-Directional Temporal Convolutional Network. A core innovation of this model is transforming one-dimensional packet length sequences into two-dimensional pulse sequences, significantly enhancing the feature representation capability and classification accuracy of the model. Furthermore, the model employs forward, backward, middle-outward, and both-ends-inward temporal convolutions to effectively extract global features. Coupled with the self-attention mechanism, the model deeply explores the dependencies among features, greatly enhancing classification accuracy and robustness. On the public CESNET-TLS22 dataset, this method achieved an F1 score of 99.24% in the classification of five application services and an average F1 score of 98.74% in the 20-application classification test. Additionally, on the UNB ISCX dataset, tests on VPN and Tor traffic also demonstrated outstanding performance, with F1 scores exceeding 98%. These results highlight the precision and effectiveness of this method across various application scenarios.
- 30 August 2025
- Published by Springer Nature
- p. 854-867
Encrypted Traffic Classification Based on Hybrid Selective State Model
- 30 August 2025
- Published by Springer Nature
- p. 854-867
Encrypted Traffic Classification Based on Hybrid Selective State Model
- 30 August 2025
- Published by Springer Nature
- p. 794-805
Research on Lightweight Learning in the Field of Traffic Analysis
- 30 August 2025
- Published by Springer Nature
- p. 794-805
Research on Lightweight Learning in the Field of Traffic Analysis
- Open Access
- 13 May 2025
- Published by MDPI AG in Electronics
- Vol. 14 (10) , 1977
Diffusion–Attention Traffic Generation: Traffic Generation Based on the Fusion of a Diffusion Model and a Self-Attention Mechanism
Network traffic generation technology plays a critical role in network testing and security protection. It simulates various network traffic patterns, offering intrusion detection systems with diverse samples. These samples enhance the system’s adaptability to emerging traffic types and improve detection capabilities. Additionally, it provides high-quality data for traffic analysis algorithms. These data help optimize model performance by increasing robustness and applicability. However, traditional generative models, like GAN networks, have limited capacity for capturing temporal flow features. As a result, they produce suboptimal generation outcomes. To address this issue, we propose the diffusion–attention traffic generation (DATG) framework. This framework combines a diffusion model with a self-attention mechanism. This integration ensures more accurate simulation of temporal characteristics. The diffusion model’s progressive denoising process guarantees generation stability. Meanwhile, the self-attention mechanism captures global temporal dependencies in traffic sequences. Experimental results validate the superior performance of our DATG by reducing the JSD to 0.18 through the synergistic optimization of the diffusion model and the self-attention mechanism, which is 41.9% better than that of GAN. Additionally, it establishes cross-step-length dependencies through the time dimension self-attention mechanism, which reduces the CRPS value to 0.13, which is 31.6% lower than that of GAN.
- 13 May 2025
- Published by MDPI AG in Electronics
- Vol. 14 (10) , 1977
Open AccessDiffusion–Attention Traffic Generation: Traffic Generation Based on the Fusion of a Diffusion Model and a Self-Attention Mechanism
Network traffic generation technology plays a critical role in network testing and security protection. It simulates various network traffic patterns, offering intrusion detection systems with diverse samples. These samples enhance the system’s adaptability to emerging traffic types and improve detection capabilities. Additionally, it provides high-quality data for traffic analysis algorithms. These data help optimize model performance by increasing robustness and applicability. However, traditional generative models, like GAN networks, have limited capacity for capturing temporal flow features. As a result, they produce suboptimal generation outcomes. To address this issue, we propose the diffusion–attention traffic generation (DATG) framework. This framework combines a diffusion model with a self-attention mechanism. This integration ensures more accurate simulation of temporal characteristics. The diffusion model’s progressive denoising process guarantees generation stability. Meanwhile, the self-attention mechanism captures global temporal dependencies in traffic sequences. Experimental results validate the superior performance of our DATG by reducing the JSD to 0.18 through the synergistic optimization of the diffusion model and the self-attention mechanism, which is 41.9% better than that of GAN. Additionally, it establishes cross-step-length dependencies through the time dimension self-attention mechanism, which reduces the CRPS value to 0.13, which is 31.6% lower than that of GAN.
- Open Access
- 14 April 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (8) , 4317
SDN Anomalous Traffic Detection Based on Temporal Convolutional Network
The wide application of software-defined network (SDN) architecture, combined with its centralized control characteristics, have exacerbated the potential risk of network attacks, and the traditional anomaly traffic detection methods are facing the challenges of high false alarm rate and insufficient generalization ability due to the reliance on manual rule design and the difficulty in capturing dynamic temporal features. In response to these challenges, we propose a Temporal Convolutional Network (TCN)-based anomalous traffic detection method for SDN. Taking the packet length sequence as the core feature, the long-term temporal dependency in the traffic data is effectively captured by causal convolution and dilation convolution operations of the TCN model, combined with the residual connection mechanism to optimize the gradient propagation and improve the stability of the model training. The experiments validate the model performance based on the public InSDN dataset, and the results show that the method achieves high accuracy in the binary classification task of normal and malicious traffic and improves its detection accuracy by about 5% compared with traditional statistical methods and mainstream deep learning models.
- 14 April 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (8) , 4317
Open AccessSDN Anomalous Traffic Detection Based on Temporal Convolutional Network
The wide application of software-defined network (SDN) architecture, combined with its centralized control characteristics, have exacerbated the potential risk of network attacks, and the traditional anomaly traffic detection methods are facing the challenges of high false alarm rate and insufficient generalization ability due to the reliance on manual rule design and the difficulty in capturing dynamic temporal features. In response to these challenges, we propose a Temporal Convolutional Network (TCN)-based anomalous traffic detection method for SDN. Taking the packet length sequence as the core feature, the long-term temporal dependency in the traffic data is effectively captured by causal convolution and dilation convolution operations of the TCN model, combined with the residual connection mechanism to optimize the gradient propagation and improve the stability of the model training. The experiments validate the model performance based on the public InSDN dataset, and the results show that the method achieves high accuracy in the binary classification task of normal and malicious traffic and improves its detection accuracy by about 5% compared with traditional statistical methods and mainstream deep learning models.
- Open Access
- 29 September 2024
- Published by MDPI AG in Electronics
- Vol. 13 (19) , 3870
A Stealthy Communication Model for Protecting Aggregated Results Integrity in Federated Learning
Given how quickly artificial intelligence technology is developing, federated learning (FL) has emerged to enable effective model training while protecting data privacy. However, when using homomorphic encryption (HE) techniques for privacy protection, FL faces challenges related to the integrity of HE ciphertexts. In the HE-based privacy-preserving FL framework, the public disclosure of the public key and the homomorphic additive property of the HE algorithm pose serious threats to the integrity of the ciphertext of FL’s aggregated results. For the first time, this paper employs covert communication by embedding the hash value of the aggregated result ciphertext received by the client into the ciphertext of local model parameters using the lossless homomorphic additive property of the Paillier algorithm. When the server receives the ciphertext of the local model parameters, it can extract and verify the hash value to determine whether the ciphertext of the FL’s aggregated results has been tampered with. We also used chaotic sequences to select the embedding positions, further enhancing the concealment of the scheme. The experimental findings demonstrate that the suggested plan passed the Welch’s t-test, the K–L divergence test, and the K–S test. These findings confirm that ciphertexts containing covert information are statistically indistinguishable from normal ciphertexts, thereby affirming the proposed scheme’s effectiveness in safeguarding the integrity of the FL’s aggregated ciphertext results. The channel capacity of this scheme can reach up to 512 bits per round, which is higher compared to other FL-based covert channels.
- 29 September 2024
- Published by MDPI AG in Electronics
- Vol. 13 (19) , 3870
Open AccessA Stealthy Communication Model for Protecting Aggregated Results Integrity in Federated Learning
Given how quickly artificial intelligence technology is developing, federated learning (FL) has emerged to enable effective model training while protecting data privacy. However, when using homomorphic encryption (HE) techniques for privacy protection, FL faces challenges related to the integrity of HE ciphertexts. In the HE-based privacy-preserving FL framework, the public disclosure of the public key and the homomorphic additive property of the HE algorithm pose serious threats to the integrity of the ciphertext of FL’s aggregated results. For the first time, this paper employs covert communication by embedding the hash value of the aggregated result ciphertext received by the client into the ciphertext of local model parameters using the lossless homomorphic additive property of the Paillier algorithm. When the server receives the ciphertext of the local model parameters, it can extract and verify the hash value to determine whether the ciphertext of the FL’s aggregated results has been tampered with. We also used chaotic sequences to select the embedding positions, further enhancing the concealment of the scheme. The experimental findings demonstrate that the suggested plan passed the Welch’s t-test, the K–L divergence test, and the K–S test. These findings confirm that ciphertexts containing covert information are statistically indistinguishable from normal ciphertexts, thereby affirming the proposed scheme’s effectiveness in safeguarding the integrity of the FL’s aggregated ciphertext results. The channel capacity of this scheme can reach up to 512 bits per round, which is higher compared to other FL-based covert channels.
- Open Access
- 20 April 2024
- Published by MDPI AG in Electronics
- Vol. 13 (8) , 1577
Intelligent Security Authentication for Connected and Autonomous Vehicles: Attacks and Defenses
The emergence of integrated positioning, communication, and sensing technologies has paved the way for a surge in connected and autonomous vehicles. The control system has been successful in reliable and fast transmission. However, practical applications face security risks, especially data tampering and spoofing attacks. To improve the resilience of the system against potential attacks, we attempt to leverage a generative adversarial network learning-assisted authentication framework (GAF). In addition to proposing a new method for validating vehicles, we also introduce a new architectural innovation in the generator–discriminator pair to achieve improved results. The generator sub-network is constructed using an advanced convolutional neural network, whereas the discriminator is designed to leverage global and local information to determine whether a signal is real or fake. On this basis, we propose a signal enhancement-based authentication method, a deep convolutional generative adversarial network (DCGAN). Experimental results using the National Institute of Standards and Technology (NIST) dataset show that the proposed method is effective in denoising and improving the detection performance.
- 20 April 2024
- Published by MDPI AG in Electronics
- Vol. 13 (8) , 1577
Open AccessIntelligent Security Authentication for Connected and Autonomous Vehicles: Attacks and Defenses
The emergence of integrated positioning, communication, and sensing technologies has paved the way for a surge in connected and autonomous vehicles. The control system has been successful in reliable and fast transmission. However, practical applications face security risks, especially data tampering and spoofing attacks. To improve the resilience of the system against potential attacks, we attempt to leverage a generative adversarial network learning-assisted authentication framework (GAF). In addition to proposing a new method for validating vehicles, we also introduce a new architectural innovation in the generator–discriminator pair to achieve improved results. The generator sub-network is constructed using an advanced convolutional neural network, whereas the discriminator is designed to leverage global and local information to determine whether a signal is real or fake. On this basis, we propose a signal enhancement-based authentication method, a deep convolutional generative adversarial network (DCGAN). Experimental results using the National Institute of Standards and Technology (NIST) dataset show that the proposed method is effective in denoising and improving the detection performance.