Abstract This paper presents a secure blockchain-enabled Automatic Number Plate Recognition (ANPR) framework for Intelligent Transportation Systems (ITS), combining lightweight YOLO-based license plate detection, OCR-based character recognition, and Ethereum smart contracts for tamper-resistant vehicle data management. The study investigates the inference performance of recent object detection architectures, namely YOLOv11, YOLOv12, and YOLO26, on Raspberry Pi 5 and NVIDIA Jetson Nano edge devices. Furthermore, optimizations for ONNX and TensorRT are explored to enhance inference efficiency on embedded hardware. Experimental results demonstrate complementary strengths among the evaluated models. YOLOv11n achieved the best overall detection performance, obtaining the highest recall (0.945), F1-score (0.9645), and mAP@0.5:0.95 (0.692), while YOLOv11n and YOLOv12n achieved identical precision (0.985) and mAP@0.5 (0.963). YOLO26 demonstrates the best deployment efficiency, reducing inference latency to 33.4 ms on Raspberry Pi (ONNX) and 38.9 ms on Jetson Nano, making it suitable for edge applications. Model optimization significantly accelerates inference, reducing YOLOv11 latency on Raspberry Pi from 98.2 ms to 38.3 ms after ONNX conversion. The results demonstrate that the proposed framework effectively balances detection accuracy, computational efficiency, and data security, making it a promising solution for smart parking and next-generation ITS applications.
Abstract Medical images such as computed tomography (CT), magnetic resonance imaging (MRI), and X-ray scans are commonly sent through health care networks for remote medical diagnostics and consultations. While sending, the said images become susceptible to being exposed to noise, tampered, and attacked, which might not seem visually perceptible yet could cause significant medical errors. In this study, we present AegisSentinel, an end-to-end solution that uses Hybrid Fibonacci Q-Matrix Encryption with SHA Key (HFQE-S) to secure medical images, a Hierarchical Auto-Associative Polynomial CNN (HAPCNN) to detect transmission attacks, and Crayfish Optimization Algorithm (COA) to tune the latter model. The algorithm was tested using 8,550 labeled CT scan images from the TCGA-LUAD dataset across six classes (clean and five types of attacks). The proposed encryption scheme yielded an entropy value of 7.9973 bits owing to a lossless decryption process with infinite PSNR. COA-tuned HAPCNN yielded 96.52% accuracy with 96.49% F1 score in classification while performing encryption and decryption within 2.87 ms and 2.35 ms, respectively.
Abstract This paper examines the application of the Generative Pre-trained Transformer (GPT) model for misinformation detection as part of Zero-Trust security models in Industry 5.0. As the transition to Industry 5.0 gains momentum, characterized by heightened interconnectedness and human-machine interaction, the necessity for rigorous security measures becomes critical. Disinformation poses significant threats, potentially undermining operational efficiency and safety, while causing substantial financial and reputational damage. To address these risks, we propose leveraging GPT, a natural language processing tool, to develop an effective misinformation detection system that enhances data integrity and trust in human-machine interactions within Industry 5.0. This paper discusses integrating a GPT-based misinformation detection system with existing security measures in Industry 5.0, showcasing its potential to strengthen a Zero-Trust security framework. We provide a detailed architectural specification of how the GPT-3 classifier integrates as a Policy Decision Point (PDP) within a Zero Trust Network Access (ZTNA) framework. Through case studies and simulations, we demonstrate the value of this machine learning model in protecting Industry 5.0 environments from misinformation-related threats. We also present inference latency analysis across GPT-3 variants to evaluate suitability for real-time industrial deployment. The research emphasizes the need for continuous vigilance and adaptable security strategies, highlighting the crucial role of GPT-based misinformation detection in maintaining a secure and reliable environment for human-machine interactions within the rapidly evolving context of Industry 5.0.
Abstract This abstract summarizes the application of System Readiness Level (SRL) as a governing metric for evaluating the maturity of System Security through the lenses of integration, interoperability, and standardization. In the development of complex, multi-domain systems, individual component maturity (Technology Readiness Level) often fails to account for the security vulnerabilities that emerge during subsystem convergence. This paper/discussion examines the System Readiness Level (SRL)—a composite index derived from the interaction between Technology Readiness Levels (TRL) and Integration Readiness Levels (IRL)—as a critical framework for System Security. By quantifying the maturity of secure interfaces, SRL provides a systematic method for ensuring that security controls are not only technically sound but also integrated into the system architecture without creating new attack vectors. Furthermore, the framework enhances interoperability by standardizing the criteria for secure data exchange across disparate platforms. Through the adoption of SRL, organizations can transition from fragmented security assessments to a standardized, holistic maturity model, ultimately reducing acquisition risk and ensuring that the final system is operationally ready to defend against evolving cyber threats.
Abstract Communication technologies in modern power systems have rendered energy grids vulnerable to False Data Injection Attacks (FDIA). This study evaluates FDIA detection on the IEEE 118-bus test system using pandapower. A synthetic dataset of 2000 samples was partitioned into 75%/25% training/test splits. Three models — LSTM networks, Isolation Forest and One-Class SVM — were comparatively assessed. LSTM achieved the highest performance with 97.0% recall, F1-score of 0.965 and AUC of 0.987, outperforming Isolation Forest (93.0%) and One-Class SVM (90.0%). These results confirm that deep learning approaches are highly effective for smart grid cybersecurity and compatible with IEC 61850-based infrastructures.
Reinforcement learning has emerged as a promising paradigm for adaptive Internet of Things (IoT) security due to its ability to optimize sequential defense decisions in dynamic environments. Yet, the vast majority of reinforcement learning-based IoT security solutions take benign learning conditions and impose regulatory imperatives as static post-processing constraints. You are aware that these assumptions allow adversarial exploitation of the learning loop with respect to state observations, reward feedback, and policy updates while still adhering to the rules. To address this limitation, in this paper, we propose an adversarially robust and regulation-aware reinforcement learning framework that explicitly models the learning process itself as a possible attack surface. By integrating bounded adversarial threat modeling and dynamic, compliance-aware policy optimization, the approach leverages uncertainty-aware reward shaping and constrained robust optimization to maintain policy stability and regulatory compliance. Experimental assessment over benchmark IoT intrusion detection datasets shows that the framework achieves up to 35% more stability in policy, reduces security degradation by more than 30% under adversarial conditions, and dramatically lowers compliance violation rates compared to classical architectures relying solely on reinforcement learning. These results emphasize how the security of the learning process itself is essential to the establishment of trust-based and regulation-compliant IoT security systems.
A fair tendering system is essential in public administration to ensure the proper use of taxpayers' money. The sealed-bid quotation method remains preferred for minor purchases due to its simplicity. However, it has more geographical restrictions and is more vulnerable to fraud than the e-tendering system. E-tendering methods are more technology-dependent and are used for major purchases. However, both approaches require quotation submission to the procurement host. Therefore, this research proposes a lightweight and secure bidding system that requires no quotation submission at the time of participation. Quotation disclosure is required at the time of bid opening. Until the bid opening event, the quotation remains in the respective bidder's possession, and there is no risk of data leakage. Therefore, this system can be used to secure bidding in real-time.
The high rate of healthcare digitalization and the extensive use of electronic healthcare records (EHRs) have heightened the issues of data privacy, interoperability, and the ability of the system to scale. Blockchain has emerged as a promising paradigm of decentralizing trust and improving security in healthcare information systems, and its real implementation is still divided. The paper highlights a systematic review of 94 peer-reviewed articles published in the years 2019-2025, which investigate architectural designs, privacy designs, scalability designs, federated learning designs, cross-chain interoperability designs, and novel cryptography designs in health care blockchain systems. The review contributes to the research in two ways: (i) a full taxonomy of blockchain design methods of healthcare applications and (ii) an organized discussion of the gaps in the research and future trends. The results demonstrate that hybrid constructions of lightweight zero-knowledge proofs, federated learning, adaptive consensus mechanisms, and cross-chain frameworks have better potential in privacy, scalability, and regulatory compliance than blockchain-based EHR solutions.
Wireless Body Area Networks (WBANs) are crucial for advancing healthcare monitoring, enabling continuous data collection directly from patients. Efficient data management strategies are essential to maximize the utility of the sensitive health data these networks gather. Existing WBAN systems often face significant challenges in balancing data privacy with network performance, potentially compromising patient safety and system effectiveness. Traditional data aggregation methods in WBANs generally emphasize enhancing transmission speeds and reducing energy usage, but they frequently neglect critical aspects of privacy protection and data integrity. This study evaluates advanced data aggregation techniques need to prioritize privacy preservation while enhancing system efficiency, specifically addressing the limitations in latency and energy consumption inherent to existing methods. The efficacy of the projected solutions is assessed using various datasets, including the comprehensive MIMIC-III and PhysioNet for real-world healthcare scenarios, and custom simulated WBAN data to test under controlled conditions. Comparative evaluation of existing techniques show that significantly improve network performance, achieving packet delivery ratios up to 99%, reducing latency to as low as 80 ms, and minimizing energy consumption to 0.33 J, all while maintaining high levels of data integrity and privacy. The advanced data aggregation strategies analysed in this study, which surveys and comparatively evaluates techniques effectively address the critical challenges in WBANs, demonstrating substantial enhancements in both privacy and operational efficiency. This balanced approach paves the way for broader adoption and more reliable deployment of WBAN technologies in sensitive health monitoring applications.
IIoT and cyber-physical systems transform industries through intelligent and adaptable automation. As connectivity increases, cybersecurity dangers escalate. Certain methods for ensuring safety are ineffective in decentralized industrial environments. This study examines how blockchain safeguards the transmission of IIoT and Industry 4.0 data, creates immutable records, and regulates access via smart contracts. We examine blockchain for device authentication, data integrity, and decentralized access management. The paper provides practical examples of how blockchain may enhance transparency, trustworthiness, and resilience in many sectors.
Cloud-based electronic health records (EHRs) are increasingly vulnerable to insider threats, privacy issues, and fine-grained access. Modern state-of-the-art techniques, such Ciphertext-Policy Attribute-Based Encryption (CP-ABE), some-times have scaling problems in actual healthcare settings, high computational costs, and restricted revocation possibilities. In order to overcome these drawbacks, we provide a hybrid security system that integrates symmetric Fernet encryption with CP-ABE, together with a Trusted Third Party (TTP) for key management, policy enforcement, and dynamic role-based access. A time-bound access control mecha-nism, user revocation, and real-time access auditing are all supported by the system. Under a 500-user load with 50 transactions per second, our solution, which was cre-ated using Flask and React.js on a cloud-backed Ubuntu server, exhibits minimal latency-150 ms for registration, 120 ms for login, and 200 ms for 1 MB EHR upload. According to experimental findings, the suggested approach outperforms current methods in terms of cryptographic efficiency and computing cost. Because of this, our platform is useful for actual EHR implementations in addition to being scalable and safe.
This paper presents a novel task scheduling framework for edge cloud environments by integrating Federated Reinforcement Learning (FRL) with an Adaptive Swarm Intelligence based approach. The need for intelligent scheduling algorithms is increasing day by day for executing low latency and energy efficient tasks in distributed loT applications. Preserving data privacy and optimizing performance are two important objectives expected from these techniques. The proposed framework employs FRL to train localized agents at edge nodes for making scheduling decisions while collaboratively improving a global model without raw data exchange. A modified Artificial Bee Colony (ABC) algorithm is integrated to further improve the task allocation process. This algorithm dynamically adapts to resource states and workload characteristics across edge and cloud tiers. The proposed hybrid system jointly optimizes make span, energy consumption and resource utilization. Experimental results on simulated edge cloud workloads demonstrate significant improvements in scheduling efficiency, latency reduction and load balancing compared to existing centralized RL and bio-inspired methods.
This paper proposes a novel taxonomy of physical social engineering attacks that require physical & direct interaction with the target. It addresses the classification challenge found in the literature, where approximately 35% of non-SEA cases are identified. This helps in understanding diverse methodologies employed by attackers. This paper presents a comprehensive state-of-the-art study of each attack's concept, background, methodology, and impact on individuals or organisations, to estimate the severity of the attack. Also, social and psychological factors associated with attacks, such as trust, obedience, respect, and kindness, are exploited, which requires a critical need to enhance awareness and integrate psychology with security measures. This emphasises considering the human element in cybersecurity solutions and their understanding.
Physically Related Functions (PReFs) are fundamentally similar to strong Physically Unclonable Functions (PUFs). Their study was driven by the goal of realizing "Cryptophasia in hardware"-a phenomenon where two hardware devices can securely establish an authenticated key without the need for long-term cryptographic key storage. Chatterjee et al. demonstrated that devices using PReFs can achieve Authenticated Key Exchange (AKE) without the burden of intensive cryptographic computations, making this approach desirable for lightweight environments. However, to ensure forward secrecy, they employed an elliptic curve Diffie-Hellman protocol, with PReFs replacing long-term keys typically used in traditional public key-based AKE systems. This paper proposes a method to eliminate dependence on conventional public key cryptography for achieving forward-secure AKE under minimal and practical assumptions. Our forward-secure AKE protocol is constructed using PReFs, error-correction codes, and collision-resistant hash functions, and is proven secure under CK and eCK security models.
With the rapid growth of online users, protecting user privacy in access control scenarios has become a critical challenge in the field of information security. Group signatures serve as a fundamental cryptographic primitive that enables users to sign on behalf of a group, providing anonymity while supporting traceability of signers. However, traditional group signature schemes, which rely on number-theoretic assumptions vulnerable to quantum algorithms, face significant security threats in the advent of quantum computing. In this paper, we propose a module lattice-based group signature scheme (ML-GS). ML-GS leverages FIPS 204 standard, a recently standardized signature scheme by the National Institute of Standards Technology (NIST), and integrates a dualrejection-sampling signature mechanism with the K-PKE encryption scheme from FIPS 203 standard, forming a "sign-hybrid-encrypt" hybrid structure to ensure both efficiency and traceability. In the key generation phase, we introduce a module Gaussian preimage sampling algorithm that reduces public key size and supports dynamic user enrollment. The security of ML-GS is formally proven in the random oracle model under the Module Learning with Error (MLWE) and Module Short Integer Solution (MSIS) assumptions. Experimental results demonstrate that, compared to existing scheme with equivalent security level, the ML-GS scheme achieves significant improvements in both time and storage overhead.
The advent of quantum computers brings about the need for cryptosystems that can withstand quantum attacks. The QC-MDPC based McEliece cryptosystem is one such post-quantum cryptographic scheme, offering robust security yet posing significant challenges in efficient decoding. Central to these challenges is the selection of an optimal threshold for bit-flipping decoding algorithms. This paper presents a novel approach that applies Particle Swarm Optimization (PSO) to dynamically establish optimal thresholds, thereby aiming to minimize the number of iterations needed for successful decoding. We demonstrate the effectiveness of our method through rigorous simulations, underscoring its potential to enhance the efficiency of the McEliece cryptosystem and comparable post-quantum cryptographic schemes. This research could mark a substantial step towards greater practicality in the realm of post-quantum cryptography.
Traditional visual secret sharing (VSS) schemes that encrypt secret images into noise-like images have been extended to multiple secret sharing (MSS) schemes to share multiple secret images using meaningful shares. Due to poor visual quality, such schemes are vulnerable to attacks from intruders. In this paper, we propose an efficient (n, n) MSS scheme for sharing n secret images, where n cover images are embedded in n meaningful shares. The proposed scheme provides improved visual quality with higher sharing capacity. Security is enhanced by meaningful share images. Thus, the proposed scheme offers enhanced quality and efficiency while satisfying security criteria.
Effective risk management is crucial for businesses to mitigate a wide array of threats. This paper conducts a bibliometric analysis of information security risk management literature spanning two decades, from 2000 to early 2025. Using the Web of Science (WoS) database, 418 English-language journal articles were reviewed. Visualizations using VOS viewer and Power BI offer insights into publication trends, co-citations, keywords, authors, journals and contributing countries. The analysis identifies four thematic clusters reflecting key areas in information security risk management research. The findings highlight significant growth in this field over the past 25 years, with expectations for continued international expansion and knowledge advancement.
The blockchain was initially designed to secure cryptocurrency transactions, a distributed and unalterable registry technology. After its innovative application in the world of cryptocurrencies, solutions based on this technology are now offered to deal with various issues in various sectors, such as the agricultural sector. It is currently one of the most disruptive technologies. This article explores the convergence of Hyperledger Aries blockchain technology and artificial intelligence to improve the management of agricultural disease detection data. We present an innovative system that guarantees complete traceability of each data point, from initial detection to analysis results. Blockchain ensures the transparency and immutability of information, while artificial intelligence, integrated into the detection process, accurately distinguishes infected sheets from healthy ones. This approach provides a robust solution for sustainable agriculture, enabling rapid and targeted response to disease threats while ensuring data integrity.
Securing wireless sensor networks (WSNs) is challenging due to the resource constraints of SNs, including limited computational power, memory, and energy capacity. Cryptographic key management is central to ensuring data confidentiality, integrity, and authenticity in WSNs. This paper explores key management schemes categorized by their design principles, including symmetric, asymmetric, hybrid, and hierarchical approaches. The strengths, limitations, and ideal use cases of these schemes are analyzed to identify their suitability for various applications. The study also addresses critical challenges such as energy efficiency, scalability, resilience to node capture, and the emerging need for post-quantum cryptography in securing future networks. The findings aim to guide the development of optimized key management solutions tailored to the evolving requirements of WSNs, ensuring robust security and operational efficiency.
