Mihui Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Mihui Kim has a total of 33 co-authors. Here are the top co-authors by number of shared publications:
- Haesoo Kim
School of Computer Engineering & Applied Mathematics, Computer System Institute, Hankyong National University, Jungang-ro, Anseong-si 17579, Gyeonggi-do, Republic of Korea
3 shared publications - Junhyeok Yun
School of Computer Engineering & Applied Mathematics, Computer System Institute, Hankyong National University, Anseong 17579, Korea
3 shared publications - Yoomin Go
School of Computer Engineering & Applied Mathematics, Computer System Institute, Hankyong National University, 327 Jungang-ro, Anseong-si 17579, Gyeonggi-do, Republic of Korea
2 shared publications - Younghee Park
Department of Computer Engineering, San Jose State University, San Jose, CA 95192, USA
2 shared publications - An Liu
Zhejiang University
1 shared publication - Doo-Soon Park
Soonchunhyang University
1 shared publication
Latest Publications
- Open Access
- 21 October 2026
- Published by MDPI AG in Electronics
- Vol. 15 (20) , 4866
A Comparative Performance and Scalability Analysis of Blockchain Anchoring Designs for Shamir’s Secret Sharing-Based Private Key Recovery
Private-key loss can permanently prevent access to blockchain-based assets, motivating recovery mechanisms that avoid retaining an entire key at a single location. Shamir’s Secret Sharing (SSS) distributes a secret across multiple shares, but off-chain shares require a verifiable integrity reference. This paper compares two blockchain anchoring designs under controlled inputs: Per-Share anchoring, which records one hash per share, and Merkle Root Anchoring, which commits the complete share set through a single root. Seven share settings (n = 5, 10, 20, 50, 100, 500, and 1000) were evaluated with k = ⌊n/2⌋ + 1 and 30 paired datasets per setting in a Hardhat local Ethereum environment. The evaluation consisted of 420 main runs, 420 counterbalanced recovery runs, and 510 runs of a paper-guided Pedin baseline. At n = 1000, the mean cumulative anchoring gas decreased from 47,479,746.8 for Per-Share to 46,538.8 for Merkle Root Anchoring, a 99.90% reduction, while the transaction count decreased from 1000 to 1. Merkle Root Anchoring achieved lower mean recovery times across all settings; however, the paired 95% confidence interval excluded zero only at n = 20, 50, 100, and 500. At n = 1000, mean recovery times converged to 459.29 ms for Per-Share and 453.74 ms for Merkle Root Anchoring. These results show that Merkle Root Anchoring substantially improves on-chain scalability, but the resulting savings do not translate proportionally into end-to-end recovery latency because the dominant latency source shifts from repeated blockchain queries to Merkle-proof verification.
- 21 October 2026
- Published by MDPI AG in Electronics
- Vol. 15 (20) , 4866
Open AccessA Comparative Performance and Scalability Analysis of Blockchain Anchoring Designs for Shamir’s Secret Sharing-Based Private Key Recovery
Private-key loss can permanently prevent access to blockchain-based assets, motivating recovery mechanisms that avoid retaining an entire key at a single location. Shamir’s Secret Sharing (SSS) distributes a secret across multiple shares, but off-chain shares require a verifiable integrity reference. This paper compares two blockchain anchoring designs under controlled inputs: Per-Share anchoring, which records one hash per share, and Merkle Root Anchoring, which commits the complete share set through a single root. Seven share settings (n = 5, 10, 20, 50, 100, 500, and 1000) were evaluated with k = ⌊n/2⌋ + 1 and 30 paired datasets per setting in a Hardhat local Ethereum environment. The evaluation consisted of 420 main runs, 420 counterbalanced recovery runs, and 510 runs of a paper-guided Pedin baseline. At n = 1000, the mean cumulative anchoring gas decreased from 47,479,746.8 for Per-Share to 46,538.8 for Merkle Root Anchoring, a 99.90% reduction, while the transaction count decreased from 1000 to 1. Merkle Root Anchoring achieved lower mean recovery times across all settings; however, the paired 95% confidence interval excluded zero only at n = 20, 50, 100, and 500. At n = 1000, mean recovery times converged to 459.29 ms for Per-Share and 453.74 ms for Merkle Root Anchoring. These results show that Merkle Root Anchoring substantially improves on-chain scalability, but the resulting savings do not translate proportionally into end-to-end recovery latency because the dominant latency source shifts from repeated blockchain queries to Merkle-proof verification.
- Open Access
- 11 June 2026
- Published by MDPI AG in Electronics
- Vol. 15 (12) , 2573
A Hyperledger Fabric-Based SBOM Management System for Secure Software Supply Chain Integrity
Recently, there has been an increasing prevalence of software supply chain attacks on software component suppliers. These attacks have targeted suppliers with relatively weak security or they have exploited vulnerabilities in open-source software. The software bill of materials (SBOM) has gained significant attention as a mechanism for improving software supply chain transparency and traceability. In this study, we propose an SBOM distribution architecture based on Hyperledger Fabric, which is a permissioned blockchain platform, to facilitate secure SBOM management. This approach utilizes Hyperledger Fabric private data collections (PDCs) to separate SBOM metadata from sensitive component information, thereby enabling confidential data sharing while reducing the blockchain storage overhead compared to a fully on-chain approach. The proposed PDC-based architecture achieves lower latency and higher throughput than the fully on-chain approach under the evaluated workload conditions, while supporting integrity verification and controlled sharing of sensitive component data.
- 11 June 2026
- Published by MDPI AG in Electronics
- Vol. 15 (12) , 2573
Open AccessA Hyperledger Fabric-Based SBOM Management System for Secure Software Supply Chain Integrity
Recently, there has been an increasing prevalence of software supply chain attacks on software component suppliers. These attacks have targeted suppliers with relatively weak security or they have exploited vulnerabilities in open-source software. The software bill of materials (SBOM) has gained significant attention as a mechanism for improving software supply chain transparency and traceability. In this study, we propose an SBOM distribution architecture based on Hyperledger Fabric, which is a permissioned blockchain platform, to facilitate secure SBOM management. This approach utilizes Hyperledger Fabric private data collections (PDCs) to separate SBOM metadata from sensitive component information, thereby enabling confidential data sharing while reducing the blockchain storage overhead compared to a fully on-chain approach. The proposed PDC-based architecture achieves lower latency and higher throughput than the fully on-chain approach under the evaluated workload conditions, while supporting integrity verification and controlled sharing of sensitive component data.
- Open Access
- 4 October 2024
- Published by MDPI AG in Electronics
- Vol. 13 (19) , 3923
Rare Data Image Classification System Using Few-Shot Learning
Advances in deep learning can address a variety of computer vision problems. In particular, deep learning has shown high performance in image processing. However, large datasets are required to train deep learning models. Previous studies have addressed the problem of data scarcity via the few-shot learning technique. However, a drawback of these studies is that large datasets are required when new tasks are performed. Hence, this study uses data augmentation techniques to address this shortcoming. Furthermore, we propose an image classification system with a few-shot learning technique that achieves high accuracy, even on rare datasets. Compared with traditional image classification models, the proposed system improves classification accuracy by approximately 18% using 100 data points.
- 4 October 2024
- Published by MDPI AG in Electronics
- Vol. 13 (19) , 3923
Open AccessRare Data Image Classification System Using Few-Shot Learning
Advances in deep learning can address a variety of computer vision problems. In particular, deep learning has shown high performance in image processing. However, large datasets are required to train deep learning models. Previous studies have addressed the problem of data scarcity via the few-shot learning technique. However, a drawback of these studies is that large datasets are required when new tasks are performed. Hence, this study uses data augmentation techniques to address this shortcoming. Furthermore, we propose an image classification system with a few-shot learning technique that achieves high accuracy, even on rare datasets. Compared with traditional image classification models, the proposed system improves classification accuracy by approximately 18% using 100 data points.
- 29 September 2024
- Published by Springer Nature
- p. 35-41
A Study on Improvement of EV Charging Occupancy Prediction Efficiency
- 29 September 2024
- Published by Springer Nature
- p. 35-41
A Study on Improvement of EV Charging Occupancy Prediction Efficiency
- 29 September 2024
- Published by Springer Nature
- p. 56-61
Malware Detection System Based on Static-Dynamic Preprocessing Techniques Combined in an Ensemble Model
- 29 September 2024
- Published by Springer Nature
- p. 56-61
Malware Detection System Based on Static-Dynamic Preprocessing Techniques Combined in an Ensemble Model
- Open Access
- 28 June 2024
- Published by MDPI AG in Electronics
- Vol. 13 (13) , 2539
Malware Detection and Classification System Based on CNN-BiLSTM
For malicious purposes, attackers hide malware in the software used by their victims. New malware is continuously being shared on the Internet, which differs both in terms of the type of malware and method of damage. When new malware is discovered, it is possible to check whether there has been similar malware in the past and to use the old malware to counteract the new malware; however, it is difficult to check the maliciousness and similarity of all software. Thus, deep learning technology can be used to efficiently detect and classify malware. This study improves this technology’s accuracy by converting static features, which are binary data, into images and by converting time-series data, such as API call sequences, which are dynamic data with different lengths for each datum, into data with fixed lengths. We propose a system that combines AI-based malware detection and classification systems trained on both static and dynamic features. The experimental results showed a detection accuracy of 99.34%, a classification accuracy of 95.1%, and a prediction speed of approximately 0.1 s.
- 28 June 2024
- Published by MDPI AG in Electronics
- Vol. 13 (13) , 2539
Open AccessMalware Detection and Classification System Based on CNN-BiLSTM
For malicious purposes, attackers hide malware in the software used by their victims. New malware is continuously being shared on the Internet, which differs both in terms of the type of malware and method of damage. When new malware is discovered, it is possible to check whether there has been similar malware in the past and to use the old malware to counteract the new malware; however, it is difficult to check the maliciousness and similarity of all software. Thus, deep learning technology can be used to efficiently detect and classify malware. This study improves this technology’s accuracy by converting static features, which are binary data, into images and by converting time-series data, such as API call sequences, which are dynamic data with different lengths for each datum, into data with fixed lengths. We propose a system that combines AI-based malware detection and classification systems trained on both static and dynamic features. The experimental results showed a detection accuracy of 99.34%, a classification accuracy of 95.1%, and a prediction speed of approximately 0.1 s.
- Open Access
- 2 March 2024
- Published by MDPI AG in Electronics
- Vol. 13 (5) , 963
A Dynamic Analysis Data Preprocessing Technique for Malicious Code Detection with TF-IDF and Sliding Windows
When using dynamic analysis data to detect malware, time-series data such as API call sequences are used to determine malicious activity through deep learning models such as recurrent neural networks (RNN). However, in API call sequences, APIs are called differently when different programs are executed. To use these data as input for deep learning, preprocessing is performed to unify the size of the data by adding dummy zeros to the data using the zero-padding technique. However, when the standard deviation of the size is significant, the amount of dummy data added increases, making it difficult for the deep learning model to reflect the characteristics of the data. Therefore, this paper proposes a preprocessing technique using term frequency–inverse document frequency (TF-IDF) and a sliding window algorithm. We trained the long short-term memory (LSTM) model on the data with the proposed preprocessing, and the results, with an accuracy of 95.94%, a recall of 97.32%, a precision of 95.71%, and an F1-score of 96.5%, showed that the proposed preprocessing technique is effective.
- 2 March 2024
- Published by MDPI AG in Electronics
- Vol. 13 (5) , 963
Open AccessA Dynamic Analysis Data Preprocessing Technique for Malicious Code Detection with TF-IDF and Sliding Windows
When using dynamic analysis data to detect malware, time-series data such as API call sequences are used to determine malicious activity through deep learning models such as recurrent neural networks (RNN). However, in API call sequences, APIs are called differently when different programs are executed. To use these data as input for deep learning, preprocessing is performed to unify the size of the data by adding dummy zeros to the data using the zero-padding technique. However, when the standard deviation of the size is significant, the amount of dummy data added increases, making it difficult for the deep learning model to reflect the characteristics of the data. Therefore, this paper proposes a preprocessing technique using term frequency–inverse document frequency (TF-IDF) and a sliding window algorithm. We trained the long short-term memory (LSTM) model on the data with the proposed preprocessing, and the results, with an accuracy of 95.94%, a recall of 97.32%, a precision of 95.71%, and an F1-score of 96.5%, showed that the proposed preprocessing technique is effective.
- Open Access
- 17 December 2022
- Published by Wiley in Security and Communication Networks
- Vol. 2022, 1-11
AEGuard: Image Feature-Based Independent Adversarial Example Detection Model
With the rapid development of image processing technology, image recognition systems based on massive image data are being developed and deployed. The wrong decision regarding an image recognition system for security-sensitive systems can cause serious problems such as personal accidents and property damage. Furthermore, adversarial attacks, which are security attacks that cause malfunctions in image recognition systems by inserting adversarial noise, have emerged and evolved. Several studies have been conducted to prevent adversarial attacks. However, existing mechanisms have low classification accuracy and low detection accuracy for adversarial examples with small adversarial noise. This paper proposes an adversarial example detection mechanism based on image feature extraction and a deep neural network (DNN) model. The proposed system achieves versatility and independence by detecting adversarial examples based on image features, such as edge noise and discrete cosine transform (DCT) bias, which adversarial examples have in common. The proposed system shows relatively higher detection accuracy than existing mechanisms for various types and amounts of adversarial noise and different sharpness of adversarial examples because the proposed system detects them depending on the characteristics of each type of adversarial example.
- 17 December 2022
- Published by Wiley in Security and Communication Networks
- Vol. 2022, 1-11
Open AccessAEGuard: Image Feature-Based Independent Adversarial Example Detection Model
With the rapid development of image processing technology, image recognition systems based on massive image data are being developed and deployed. The wrong decision regarding an image recognition system for security-sensitive systems can cause serious problems such as personal accidents and property damage. Furthermore, adversarial attacks, which are security attacks that cause malfunctions in image recognition systems by inserting adversarial noise, have emerged and evolved. Several studies have been conducted to prevent adversarial attacks. However, existing mechanisms have low classification accuracy and low detection accuracy for adversarial examples with small adversarial noise. This paper proposes an adversarial example detection mechanism based on image feature extraction and a deep neural network (DNN) model. The proposed system achieves versatility and independence by detecting adversarial examples based on image features, such as edge noise and discrete cosine transform (DCT) bias, which adversarial examples have in common. The proposed system shows relatively higher detection accuracy than existing mechanisms for various types and amounts of adversarial noise and different sharpness of adversarial examples because the proposed system detects them depending on the characteristics of each type of adversarial example.
- Open Access
Deepfake Detection Using the Rate of Change between Frames Based on Computer Vision
Recently, artificial intelligence has been successfully used in fields, such as computer vision, voice, and big data analysis. However, various problems, such as security, privacy, and ethics, also occur owing to the development of artificial intelligence. One such problem are deepfakes. Deepfake is a compound word for deep learning and fake. It refers to a fake video created using artificial intelligence technology or the production process itself. Deepfakes can be exploited for political abuse, pornography, and fake information. This paper proposes a method to determine integrity by analyzing the computer vision features of digital content. The proposed method extracts the rate of change in the computer vision features of adjacent frames and then checks whether the video is manipulated. The test demonstrated the highest detection rate of 97% compared to the existing method or machine learning method. It also maintained the highest detection rate of 96%, even for the test that manipulates the matrix of the image to avoid the convolutional neural network detection method.
Open AccessDeepfake Detection Using the Rate of Change between Frames Based on Computer Vision
Recently, artificial intelligence has been successfully used in fields, such as computer vision, voice, and big data analysis. However, various problems, such as security, privacy, and ethics, also occur owing to the development of artificial intelligence. One such problem are deepfakes. Deepfake is a compound word for deep learning and fake. It refers to a fake video created using artificial intelligence technology or the production process itself. Deepfakes can be exploited for political abuse, pornography, and fake information. This paper proposes a method to determine integrity by analyzing the computer vision features of digital content. The proposed method extracts the rate of change in the computer vision features of adjacent frames and then checks whether the video is manipulated. The test demonstrated the highest detection rate of 97% compared to the existing method or machine learning method. It also maintained the highest detection rate of 96%, even for the test that manipulates the matrix of the image to avoid the convolutional neural network detection method.
- Open Access
JLVEA: Lightweight Real-Time Video Stream Encryption Algorithm for Internet of Things
Along with the recent growth of Internet of Things (IoT) security camera market, there have been a number of personal information leakage incidents from security attacks targeting such cameras. Therefore, a permutation-based video encryption algorithm was proposed to secure video streams in low-performance processors such as IoT security cameras. However, existing permutation-based video encryption algorithms are vulnerable to known-plaintext attacks since they use the same permutation list for every frame. Moreover, situation deduction based on the color composition is possible. In this paper, we propose a new permutation-based video encryption algorithm that updates the permutation list for every frame using a crypto secure pseudo-random number generator without significantly increasing memory usage. By doing so, the algorithm becomes robust to known-plaintext attacks, which has been a common problem with existing permutation-based video encryption algorithms. In addition, color channel separation can prevent attackers from deducing situations through color composition. Pre-compression encryption is applied to make the algorithm robust to data loss because of packet loss. We implement the proposed algorithm and conduct an experiment to show its performance in terms of probability of data loss because of packet loss, encryption speed, and memory usage.
Open AccessJLVEA: Lightweight Real-Time Video Stream Encryption Algorithm for Internet of Things
Along with the recent growth of Internet of Things (IoT) security camera market, there have been a number of personal information leakage incidents from security attacks targeting such cameras. Therefore, a permutation-based video encryption algorithm was proposed to secure video streams in low-performance processors such as IoT security cameras. However, existing permutation-based video encryption algorithms are vulnerable to known-plaintext attacks since they use the same permutation list for every frame. Moreover, situation deduction based on the color composition is possible. In this paper, we propose a new permutation-based video encryption algorithm that updates the permutation list for every frame using a crypto secure pseudo-random number generator without significantly increasing memory usage. By doing so, the algorithm becomes robust to known-plaintext attacks, which has been a common problem with existing permutation-based video encryption algorithms. In addition, color channel separation can prevent attackers from deducing situations through color composition. Pre-compression encryption is applied to make the algorithm robust to data loss because of packet loss. We implement the proposed algorithm and conduct an experiment to show its performance in terms of probability of data loss because of packet loss, encryption speed, and memory usage.