Journal of Information Security System and Cyber Criminology Research

An Intelligent Palm Vein-based Human Recognition Framework

Ananthi G — Fri, 06 Feb 2026 00:00:00 +0000

The work is about recognizing people based on the veins in their palms. In many cases, authentication is a major issue to reduce crime and eliminate the possibility of spoofing the originality, so palm-vein for human recognition is used. Palm veins are more difficult to forge than fingerprints and palm-prints because they are an internal biometric of living human bodies. Extrinsic biometrics, such as fingerprints, ears, iris, and palm print, are vulnerable to forgery attacks, putting users’ privacy and protection at risk. Due to the acquisition of live palms and the necessity of user approval, palm vein traits are difficult to forge or reproduce. Aging, as well as bruises, scars, and tattoos, seems to have little effect on palm vein identification. The main goal of the palm-vein-based human recognition scheme is recognition accuracy. The palm vein-based human recognition mechanism is divided into five modules. The first module is pre-processing, which involves determining the ROI of the palm. The second module is feature extraction, which uses the wave atom transform (WAT) method to extract the required feature of the palm vein. The third module is function randomization, in which the extracted feature is randomized using a user-specific secret key, preventing hacking. The fourth module is template creation, in which each template is produced for verification. The fifth module is the identification module, which determines whether or not a person is genuine or an imposter by comparing database values. The CASIA multispectral palm print database is used for this project.

Toward Sustainable Big Data Analytics: A Review of Privacy-preserving Federated Learning

Kangana Soni, Nitika Singhi — Mon, 16 Feb 2026 00:00:00 +0000

The accelerated growth of data-driven systems in health services, the banking sector, automated transport networks, infrastructures, and the internet of things has increased issues linked to data privacy, secure data exchange, growth capability and extended durability of data analysis systems. Traditional centrally controlled massive data analytical processing, which collects unprocessed information at a core central system, encounters privacy disclosure threats, governance-based limitations, elevated data transmission operational load, and heavy power utilization, causing these systems to become gradually inefficient in massive and diverse settings. Federated learning (FL) has developed as a sustainable and privacy-preserving framework by supporting non-centralized predictive model learning while maintaining data at on-site endpoints, thereby decreasing data transfer and enabling legal adherence. This study provides a brief overview of privacy-preserving big data analytics utilizing federated learning with a focus on data privacy protection approaches decentralized analysis, and secure data sharing frameworks. Crucial methods, including differential privacy, secure aggregation, homomorphic encryption, and secure multi-party computation, are evaluated to measure their efficiency in minimizing data exposure from learning model modifications. The overview emphasizes essential exchanges between confidentiality effectiveness, data transmission optimization and scalability and examines federated learning with conventional single-server analysis from a sustainability viewpoint. Lastly, essential investigation shortcomings are determined: restricted applied implementations, insufficient management of non-IID data, absence of consistent assessment metrics and poor interoperability with current big data frameworks. This research intends to facilitate the stated design of secure, scalable, and sustainable big data analytics architectures.

Cross-domain Ethical AI: A Systematic Review of Sector-specific Challenges in Healthcare, Finance, and Criminal Justice Applications

Akshaya Punnamaraju, P. Devi Sravanthi, Manas Kumar Yogi — Thu, 19 Feb 2026 00:00:00 +0000

Artificial intelligence (AI) is increasingly embedded in high-stakes domains such as healthcare, criminal justice, and finance, delivering significant operational and societal benefits while simultaneously introducing complex ethical challenges. This study critically examines the ethical implications of AI deployment across these sectors using four foundational ethical principles: beneficence, non-maleficence, autonomy, and justice. In healthcare, AI-driven systems support early diagnosis, personalized treatment, and clinical decision-making; however, they raise serious concerns related to patient autonomy, informed consent, data privacy, and cybersecurity. Within the criminal justice system, predictive analytics and risk assessment tools influence policing strategies and sentencing decisions, offering efficiency gains but also amplifying risks of algorithmic bias, opacity, and the reinforcement of historical inequalities. In the financial sector, applications such as algorithmic trading, fraud detection, and automated credit scoring enhance speed and accuracy, yet may compromise transparency, accountability, and equitable access to financial services. Through a comparative, cross-domain analysis, this study identifies common ethical challenges including bias mitigation, explainability, and accountability while also underscoring sector-specific risks and governance requirements. The findings emphasize that responsible AI adoption necessitates comprehensive regulatory frameworks, interdisciplinary collaboration, stakeholder oversight, and continuous ethical auditing to ensure that technological innovation remains aligned with human values, social justice, and long-term societal well-being.

A Secure and Transparent Evidence Management System Based on Blockchain Technology

S. Uma, Gokul G — Wed, 08 Apr 2026 00:00:00 +0000

Digital evidence plays a crucial role in modern criminal investigations and legal proceedings. However, traditional evidence management systems rely on centralized databases, which are vulnerable to security threats such as unauthorized access, data tampering, and data loss. These issues can compromise the integrity and authenticity of digital evidence. To address these challenges, this research proposes an evidence management system using blockchain technology. Blockchain provides a decentralized and immutable ledger that ensures transparency and security in storing digital evidence records. In the proposed system, whenever digital evidence such as images, videos, or documents is uploaded, a cryptographic hash value is generated and stored on the blockchain using smart contracts. This hash acts as a unique digital fingerprint of the evidence file. During the verification process, the system generates a new hash value and compares it with the stored hash to confirm the authenticity of the evidence. The use of blockchain technology ensures that evidence records remain tamper-proof and traceable, thereby improving the reliability and trustworthiness of digital investigations and legal processes.

Enhancing OTP Transaction Security through Federated Learning in IoT Environments

Wed, 08 Apr 2026 00:00:00 +0000

We live in a digitalized age where banks breathe data and hospitals pulse through networks. The internet of things (IoT) has transmitted and transformed the normal norm into a digitized ecosystems, automated and algorithmically connected. Yet, in this architecture of convenience lies an invisible vulnerability. As devices multiply, so do the attacks; as connectivity expands, so does the cyber fragility. Among the most pervasive threats in this digital intimacy is the one-time password (OTP) fraud—a subtle yet devastating breach of trust engineered through deception, phishing, and manipulation, which are risks at stake. A single click, tap, submit or accept push button fractures and damages the boundary between privacy and harsh truth exposed, granting unauthorized access to sensitive data and financial credentials. This research reimagines cybersecurity not as a centralized surveillance, but as a decentralized intel. In this research, we propose an IoT-based monitoring framework powered by federated learning (FL), a paradigm shift from data extraction to data sovereignty. Within this architecture, models are trained locally across distributed IoT nodes. Only encrypted model updates are aggregated, ensuring privacy preservation while enabling intelligence. This approach represents more than a technical upgrade; it is a philosophical repositioning of cybersecurity. We acknowledge in the research that in a post-digital society, privacy is not merely a feature – it is a fundamental design principle. The prototype features: data localization and sovereignty, reduction in centralized breach risks, real-time fraud detection, scalable privacy-preserving intelligence, and ethical AI deployment within the ecosystems. Overall, by synthesizing IoT infrastructure with federated learning algorithms, this research demonstrates that security and privacy coexist through distributed cognition. In an era where information is currency and identity is code, the fusion of federated learning and IoT-based monitoring systems provides a resilient, secure, and safe solution.