Dong-Seong Kim

• Professor
• Managing Director at Kumoh National Institute of Technology

As the Metaverse progresses, its security measures must evolve to safeguard users from cyberattacks. To this end, Artificial Intelligence (AI)-powered Network Intrusion Detection Systems (NIDSs) have been implemented at the network layer to detect and respond to threats in real-time. Our survey -MetaWatch provides a comprehensive overview and analy...

A Brain tumor is considered one of the most aggressive diseases among humans of all ages, and one of the most crucial tasks of neurologists and radiologists is early brain tumor detection. Traditional manual prediction and detection from Magnetic Resonance Imaging (MRI) is challenging and error-prone. Our proposed model enhances brain tumor detecti...

This study presents early results of a web-based digital twin (DT) for battery management systems (BMS). The proposed DT explores a hybrid of model-based and data-driven approaches, enabling the exploitation of each approach’s distinctive merits and constraints. Experiments employing explainable artificial intelligence (XAI) techniques were underta...

This paper introduces PureTwin, a robust framework for battery management systems that develops Non-Fungible Digital Twins. The framework leverages PureChain, a lightweight blockchain alternative, and PureEdge, an offline-capable edge computing concept. By using real-time battery data, Physics-Informed Long-Short Term Memory Networks (PiLSTMs), and...

Vehicular networks require continuous connectivity for efficient communication, but security concerns necessitate isolating malicious vehicles. This paper proposes a secure framework that allows compromised vehicles to receive critical messages during isolation while ensuring controlled outbound communication. The system is evaluated for its effect...

This paper uses the Gumi industrial complex dataset to introduce a practical framework for explainable artificial intelligence (XAI)-driven energy optimization using digital twins (DT) in the smart manufacturing industry. The framework integrates real-time data from manufacturing systems with predictive AI models, is designed to be easily implement...

In the rapidly evolving Industrial Internet of Things (IIoT) landscape, ensuring robust security measures for detecting and mitigating cyber threats is paramount. This paper suggests a decentralized intrusion detection system (IDS) that uses Federated Learning (FL) integrated with a permissioned blockchain layer to secure IIoT networks. The system...

Intelligent transportation systems (ITS) are vital in improving road safety, efficiency, and user experience. However, vehicular networks face critical security and privacy challenges due to the constant exchange of sensitive data. This paper proposes a robust, privacy-preserving framework for vehicular networks using homomorphic encryption, which...

This study explores energy-efficient machine learning approaches for intrusion detection in SCADA systems, addressing the dual challenges of cybersecurity and sustainability. A comprehensive evaluation of models, including Decision Trees, Random Forests, and LightGBM, highlights their performance across SCADA, IIoT, and Edge IoT environments. Decis...

This study presents a digital twin framework for predicting the state of health (SoH) in battery management systems (BMS). This framework integrates a single particle model with electrolytes (SPME) and a long-short-term memory (LSTM) network to model battery behaviour based on a NASA battery dataset. To ensure the security of battery data, data is...

The rapid growth of Internet of Things (IoT) devices has brought about significant advancements in automation, data collection, and connectivity across various domains. However, this increased interconnectedness also poses substantial security challenges, making IoT networks attractive targets for malicious actors. Intrusion detection systems (IDSs...

This paper presents a novel fire detection system for smart homes, combining convolutional neural networks (CNN), custom blockchain, and a digital twin for enhanced safety and accuracy. The system features a pretrained CNN in an Internet of Things (IoT) camera for real-time detection using a custom blockchain network (CBN) for secure data managemen...

Efficient communication between emergency vehicles and regular traffic is crucial for saving lives and ensuring public safety, especially in high-density traffic scenarios. This paper presents a blockchain-enabled priority-based messaging system that dynamically prioritizes emergency messages and optimizes traffic light patterns to facilitate seaml...

This study introduces a Quantum-Enhanced Physical Internet framework that leverages Quantum Key Distribution (QKD) and hybrid blockchain technologies to address global logistics security, scalability, and efficiency challenges. Using the BB84 protocol, the framework ensures secure communication, robust data integrity, and streamlined operations, ev...

This paper applies NLP-inspired techniques to enhance signal detection in Rayleigh fading wireless systems. It introduces repetition coding (signal expansion) and channel estimation (contextual awareness) to improve robustness by adding redundancy and leveraging channel information. A blockchain framework ensures secure, decentralized data manageme...

TwinLink6G is a novel explainable autoencoder-based communication framework tailored for transmitting high-dimensional battery digital twin data over 6G Open Radio Access Networks (O-RAN). The framework integrates Explainable Artificial Intelligence (XAI) techniques, specifically Local Interpretable Model-agnostic Explanations (LIME), to ensure bot...

Proteomic data storage faces challenges in security, consistency, and integration with unconventional bioinformatics systems. This paper proposes a microservices-based blockchain architecture to address these issues, focusing on amino acid sequences and 3D protein structures. Using HyperLedger and AES-256 encryption ensures secure access and data c...

Deep learning-based intrusion detection systems (DL-IDS) have proven effective in detecting cyber threats. However, their vulnerability to adversarial attacks and environmental noise, particularly in industrial settings, limits practical application. Current IDS models often assume ideal conditions, overlooking noise and adversarial manipulations,...

The metaverse is a next-generation internet that is empowered by emerging technologies such as the Internet of Things (IoT), digital twins (DT), artificial intelligence (AI), blockchain, Web 3.0, and aug-mented/ virtual reality (VR/ AR) to facilitate a dynamic interaction between the physical and virtual worlds [1]. The metaverse presents a virtual...

The increasing complexity of battery management systems (BMS) has led to challenges processing the vast amounts of data required for accurate real-time monitoring and control. Existing BMS frameworks, which rely heavily on artificial intelligence (AI), often struggle with data limitations that impact the precision of state estimates, ultimately aff...

This paper proposes a PureChain-enhanced smart factory IoT data management system focusing on temperature data. The system is designed to support lightweight IoT devices by leveraging the PureChain smart auto-miner features that start mining upon transaction initiation and stop immediately after the last transaction is mined. The architecture ensur...

Using blockchain technology to oversee clinical trials has the potential to drive innovation and significantly improve traditional data management practices. This paper proposes a blockchain-oriented approach to tackle the prevailing issues within clinical trial management frameworks and promote in-teroperability among clinical trial stakeholders....

The proposed framework leverages Non-Fungible Tokens (NFTs) to revolutionize supply chain management by ensuring product authenticity. Each product document stored in a repository is minted as an NFT. The NFT is a digital twin of the product's certification, containing metadata for tracking its origin, authenticity, and key details. When a buyer ve...

Blockchain networks must ensure secure and efficient consensus algorithms in consumer electronics and Internet of Things (IoT) devices. This paper proposes a novel consensus algorithm, Proof-of-Authority-and-Association (PoA$^2$), designed specifically for IoT blockchain networks in consumer applications. PoA$^2$ leverages redundancy-based mechanis...

The advancement of intelligent transport systems and the rise of autonomous vehicles offer significant potential for reducing road accidents. However, mountainous regions, such as South Korea, are particularly susceptible to the formation of black ice, which poses a serious risk to both human and autonomous drivers due to its near-invisibility and...

Operational naval ship systems rely on advanced sensors and equipment to detect targets and manage weapon control and engagement. However, their growing complexity, driven by technological advancements and the increasing integration of clients and servers, necessitates changes in system architecture and network structure. To address these challenge...

The legitimacy, appropriation, and authorization of data during end-to-end communication between participating nodes in a network determines the intelligence and integrity of a real-time cyber-physical system. Given the persistent and increasing violations of the airspace by drones worldwide, the fundamental intelligence and security architectures...

Vulnerabilities in drone networks stem from the reliance on GPS and wireless communication technologies, combined with the lack of robust security mechanisms. This study proposes DroneGuard, a comprehensive cybersecurity framework leveraging supervised machine learning (ML) and explainable artificial intelligence (XAI) to detect intrusions and prov...

Network disconnections between edge devices and central servers can lead to significant disruptions in data-driven applications, highlighting a possible need for predictive systems to anticipate these disconnections. This preliminary study evaluates multiple machine learning models to predict network disconnections k instants ahead using key metric...

A survey of Gumi industrial complex companies highlights the need to optimize energy efficiency. This paper prunes the 12-layer BERT model, optimized with distillation loss, to improve energy usage efficiency in Supervisory Control and Data Acquisition (SCADA) systems. Window tokenization, embedding, and self-attention capture complex long-range de...

This paper presents a novel web-based federated learning framework to predict energy contract types based on decentralized client data. By incorporating a central validation dataset post-aggregation, the model improves its accuracy while ensuring data privacy. A web application manages clients, allowing real-time monitoring, adding/removing partici...

Anomalous energy consumption is a critical issue in the industrial sector that can signal operational inefficiencies, equipment malfunctions, or cyber-attacks. This work presents a method for predicting anomalous energy consumption using Long Short-Term Memory (LSTM) networks, a recurrent neural network well-known for its efficiency in time-series...

This paper investigates and compares the performance of various predictive models for energy consumption forecasting in the Gumi Industrial Complex. A 2024 survey highlighted energy efficiency initiatives such as equipment upgrades (19.5\%) and energy monitoring systems (4.3\%), emphasizing the need for accurate and interpretable predictive technol...

Security and privacy are critical in vehicular networks due to their sensitivity and high connectivity. This paper proposes a novel framework combining blockchain and federated learning to enhance security and intrusion detection. Through smart contracts, blockchain is implemented to manage vehicle participation, ensuring data integrity and immutab...

This paper propose human-centric AI-driven power grid distribution system (PGD) that dynamically allocates power to different clusters of buildings in a multiple complex environment using explainable cognition. The PGD system adopts Elastic Net regularization technique and SHAP method to achieve a cost-effective and lightweight interpretation of it...

The use of unmanned aerial vehicles (UAVs) for smart and speedy logistics is still relatively nascent compared to traditional delivery methods. However, it is witnessing sporadic and steady growth due to booming demands, technological advancement, and regulatory support. The intelligence and integrity of UAV systems depend largely on the underlying...

This paper presents BFL-Net, a blockchain-assisted federated learning (FL) framework integrated with a large language model (LLM) for accurate energy prediction in smart industries. BFL-Net enables decentralized training with FL while preserving data privacy and enhances predictions by processing unstructured data from the Gumi Industrial Dataset t...

In the era of digital transformation, securing data on metaverse platforms poses significant challenges. This paper proposes Multiparty Space Sharing and Authentication (MSSA), a novel approach for secure user login and location access control within specialized metaverse platforms. MSSA leverages Quantum Multiparty Secret Computation (QMSC) integr...

This paper investigates the impact of two adaptive client selection mechanisms commonly employed in federated learning (FL) for dynamic networks, such as the Internet of Medical Things (IoMT), to enhance global model performance. Specifically, the study focuses on a performance-based client selection algorithm and a deep reinforcement learning (DRL...

This work proposes a security model called Pure WalletS (PW-S). Where the S stands for secure, this model aims to enhance the entire security of offline transactions based on the Pure Wallet (PW) algorithm. In this model, Security is considered at both the front-end and smart contract level to ensure enhanced security. Blockchain transaction data a...

The convergence of blockchain technology and biochemistry offers a transformative solution to significant challenges in protein folding research. Traditional methods for data sharing in protein folding research often suffer from issues such as data tampering, slow processing, and lack of transparency, which hinder reproducibility and collaborative...

This work presents a roadmap for integrating Open Radio Access Networks (O-RAN) and Artificial Intelligence (AI) into 6G networks, focusing on intrusion and fault detection. O-RAN introduces flexibility in network design by enabling disaggregated and vendor-neutral components, but it also poses security challenges due to open interfaces and multi-v...

In today's era of global trade, efficient marine logistics are pivotal for international trading. However, conventional logistic frameworks have significant difficulties in traceability, transparency, and conflict resolution, which have become more prominent due to increased globalization. This study introduces an innovative method that uses blockc...

The adoption of Federated Learning (FL) for decentralised model training has enabled workload distribution across multiple client devices. However, extreme values and outliers in the model updates from distributed devices often challenge this approach. This study introduces the Adaptive Trim strategy (AdaTrim), which leverages a trimmed mean aggreg...

Securing vehicular communication networks is crucial as increased connectivity exposes these networks to various cyber threats, particularly Distributed Denial-of-Service (DDoS). This growing vulnerability necessitates a robust framework to identify and mitigate such attacks, giving rise to distributed learning. Federated Learning has shown promise...

In the new era of technology, daily human activities are becoming more challenging in terms of monitoring complex scenes and backgrounds. To understand the scenes and activities from human life logs, human-object interaction (HOI) is important in terms of visual relationship detection and human pose estimation. Activities understanding and interact...

Independent human living systems require smart, intelligent, and sustainable online monitoring so that an individual can be assisted timely. Apart from ambient assisted living, the task of monitoring human activities plays an important role in different fields including virtual reality, surveillance security, and human interaction with robots. Such...

Detecting and capturing 3D human structures from the intensity-based image sequences is an inherently arguable problem, which attracted attention of several researchers especially in real-time activity recognition (Real-AR). These Real-AR systems have been significantly enhanced by using depth intensity sensors that gives maximum information, in sp...

More adaptable and user-independent techniques are required for multi-sensors based daily locomotion detection (MS-DLD). This research study proposes a couple of locomotion detection methods using body-worn multi-sensors to successfully categorize several locomotion transitions. This research presents both standard and state-of-the-art methods for...

The sensitivity and exclusivity attached to personal health records make such records a prime target for cyber intruders, as unauthorized access causes unfathomable repudiation and public defamation. In reality, most medical records are micro-managed by different healthcare providers, exposing them to various security issues, especially unauthorize...

Service advisors are a crucial contributor to the profit and loss of an automotive dealership. It is thus crucial that their performances are continuously evaluated to make decisions that ensure continued profitability. Artificial intelligence (AI) algorithms like the finite mixture of regression (FMR) have been previously explored as a tool for be...

The growing occurrence of cyber attacks requires advancing more complex encryption methods to protect sensitive communication in marine tactical networks. The study introduces Quantum DNA Partial Permutation (QDPP), a cryptographic scheme that combines quantum cryptography, DNA computing, and permutation-based methods to secure confidential communi...

This study proposes a novel framework that combines instance segmentation and digital twin technology to improve military surveillance. Integrating precise object identification with dynamic environmental modeling addresses critical surveillance challenges. This study presents early results of a digital twin system showcasing a military base with m...

Mass shootings, terrorism, and small firearm trafficking account for the rise in causes of death and grave injuries in most countries. However, one of the critical security concerns is the detection, screening, and recognition of small and locally handcrafted firearms. Deep learning though has played promising roles, more research options are still...

With its networked military equipment, the Internet of Military Things (IoMT) is bothered by severe security issues, hampering data accessibility and service delivery. Distributed Denial of Service (DDoS) incidents are risks that disrupt mission-critical activities. To detect DDoS assaults within IoMT data caching systems, this article investigated...

Maritime borders hold significant importance for a nation's economic and political stability. Ensuring the security of these extensive maritime frontiers is a task of priority importance for naval forces, particularly in managing and mitigating unauthorized entries by foreign vessels into international maritime zones. To simplify and automate this...

With the proliferation of the Internet of Things, artificial intelligence, and virtual reality, securing Metaverse E-commerce against malware has become increasingly critical. This study introduces a novel framework integrating Semantic Communication (SemComm) with Deep Learning for Intrusion Detection Systems (IDS) in Metaverse E-commerce. Leverag...

The immersive metaverse environment offers distinct social interactions and opportunities, yet it also presents significant challenges in securely managing misbehavior, including hate speech, bullying, and harassment. Existing solutions primarily focus on detecting such behavior through artificial intelligence but lack robust mechanisms for managem...

Artificial intelligence (AI) positively remodels industrial processes, notably inventory management (IM), from planning, scheduling, and optimization to logistics. Intelligent technologies such as AI have enabled innovative processes in the production line of manufacturing execution systems (MES), particularly in predicting IM. This study proposes...

Federated Learning (FL) has emerged as a transformative artificial intelligence paradigm, facilitating knowledge sharing among distributed edge devices while upholding data privacy. However, dynamic networks and resource-constrained devices such as drones, face challenges like power outages and network contingencies, leading to the straggler effect...

This work proposes a Blockchain-enabled Organ Matching System (BOMS) designed to manage the process of matching, storing, and sharing information. Biological factors are incorporated into matching and the cross-matching process is implemented into the smart contracts. Privacy is guaranteed by using patient-associated blockchain addresses, without t...

This study introduces FedQ, a quantum-based security scheme for defending against cyber attacks in the Internet of Medical Things (IoMT). By integrating quantum key distribution (QKD) with privacy-preserving federated learning (FL), FedQ ensures secure communication between hospital management units (HMUs) and the FL server. QKD encrypts FL transmi...

This paper addresses the rising vulnerability of the medical Internet of Things (MIoT) to attacks like spoofing and data alteration. Using the focal loss function to enhance neural network models boosts detection accuracy and reduces false positives. Experimental findings on recent MIoT data demonstrate that the focal loss function notably reduces...

The Internet of Vehicles (IoV) emphasizes the crucial role of Intrusion Detection Systems (IDS) in strengthening security with Machine learning (ML) algorithms, promising enhanced IDS performance by offering real-time anomaly detection capabilities. This study evaluates ML algorithms for accurately detecting intra-vehicular data falsification. Comb...

This work compares neural fitting optimization algorithms for predicting household electricity consumption. Leveraging gas consumption, income, season, and occupants as inputs, Bayesian Regularization, Scaled Conjugate Gradient, and Levenberg-Marquardt were assessed. The findings illuminate the most effective algorithm for forecasting accurate and...

The Joint All-Domain Command and Control (JADC2) initiative prioritizes proactive decision-making through operational terrain analysis. It integrates a vast military Internet of Things (IoT) network with Artificial Intelligence (AI) to enhance efficiency. However, this integration raises security concerns, particularly regarding distributed denial...