International Journal of Communication Networks and Information Security (IJCNIS)

Secure and Reliable AI-Based Resource Allocation in Edge Computing for Smart Logistics

Varun Kumar Shetty Vijjagiri, Ajay Chinta, Nivetha R — 2026-05-03

Growing adoption of edge computing in smart logistics has made it possible to make real-time decisions and provide low-latency services. However, the biggest challenge in solving such problems is providing secure, reliable, and adaptive allocation of resources in distributed edge environments due to dynamic workloads, the limitations of available resources, and a potential security loophole. This paper, therefore, proposes an AI-based adaptive resource allocation framework for Eu-rope, integrating security-aware orchestration mechanisms for optimizing system performance in edge-enabled smart logistics systems. The proposed framework utilizes machine learning algorithms for dynamically assigning computational resources among distributed edge nodes with trust-aware and secure data handling mechanisms to reduce the risks such as inconsistency in data, node failures, and unauthorized access. A mathematical model is formed that captures the tradeoff between reliability and latency and resource utilization under different system conditions. In order to assess how well the proposed approach works, simulation-based experiments are performed on realistic logistics scenarios. The results show that the proposed method results in an improvement of system responsiveness by up to 35% with no degradation under dynamic load conditions, which relies on high reliability. Additionally, the mixing of security-conscious techniques leads to improved fault tolerance and better tolerability at the systems in the event that they are plagued by possible threats. As compared to the traditional cloud-centric and non-adaptive frameworks, the overall system efficiency improvement is 22% in the proposed framework. The results demonstrate the need to combine AI-driven adaptive control with security mechanisms in edge computing environments as a powerful basis for designing secure, resilient, and intelligent smart logistics infrastructures.

A Hybrid Ensemble Learning Framework for Interpretable Thyroid Disease Classification Using SHAP and LIME

Mohd Bilal, Mr. Rahul Ranjan, Dr. Shish Ahmad — 2026-05-08

Thyroid diseases are considered as one of the most diagnosed endocrine diseases globally with a role in metabolic functioning, cardiovascular function and hormone regulation. Early diagnosis is difficult due to combination of different biochemical labo ratory tests and complication of clinical symptoms. Recent advances in artificial intelligence (AI) and machine learning (ML) have enabled development of automated diagnostic models which assist in the prognostication and decision making. The aim of this paper is to provide an in-depth systematic review of application of machine learning and deep learning techniques to diagnose and prognose thyroid diseases. Fundamental models, such as Random Forest, Support Vector Machine, Gradient Boosting, Artificial Neural Networks and Convolutional Neural Networks are also reviewed. Again, feature selection methods such as XGBoost and SelectKBest are discussed. Models like XGBoost, while very accurate (80-98 percent), are not explainable to be used in a medical environment. To address the limitation, explainable AI models such as SHAP and LIME are used in the proposed architecture. The algorithms provide insight in terms of features and visual representations that could help explain the trend of disease. Some of the problems identified in the paper include the lack of balance in the data sets, lack of clinical validation and lack of consistent evaluation methods and suggesting the hybrid ensemble-based solution to improve both the performance and understanding.

Real-Time Multimodal Emotion Recognition Using EEG and HRV Signals with Attention-Based Deep Learning

Reethu Joseph ,M.s Indhumol C — 2026-05-17

In affective computing, emotion recognition hasgrown in importance, especially for applications in adaptiveintelligent systems and mental healthcare. Despite extensiveresearch, facial expression and speech-based techniquesfrequently fall short in accurately capturing internal emotionalstates because of voluntary and environmental factors

Voice-Enabled AI System for Interactive Interview Preparation

Sunitha T S ,Ms .Renuka Babu T,Dr. C Sahaya Kingsly — 2026-05-17

The Voice-Enabled AI System forInteractive Interview Preparation is anintelligent, interactive system designed tosimulate real-world technical interviews
using ArtificialIntelligence.

Intelligent Sales Process Automation and Optimization using Salesforce

Anandu N ,Neethu Maria John ,Dr. C Sahaya Kingsly — 2026-05-17

Most enterprise sales teams suffer from significantoperational inefficiencies, with research estimating that 64% ofmanual working hours are consumed by non-revenueadministrative tasks, leading to high forecasting variances of 30% to 40% and severe pipeline leakage

Driver Drowsiness Detection using Deep Learning

Abhijith S ,Dr. Simy Mary Kurian ,Dr. C Sahaya Kingsly — 2026-05-17

Drowsy driving kills. Every year, road crashes claimaround 1.35 million lives globally, and a significant number of those accidents happen simply because a driver nodded off at the wheel.This paper describes a system we built to tackle that problem

Earthquake-prediction-using-Machine-learning-models

Preejamol Prasad,Ms.Riya Rajan — 2026-05-17

Earthquakes are among the most devastatingnatural disasters, causing loss of life, destruction of infrastructure, and serious economic damage. Becausethey happen suddenly and are difficult to predict,preparing for them is extremely challenging.

Deep Learning Based Early Autism Spectrum Disorder Detection Using Eye- Tracking Data

Diya Merin Babu,M.s Meenu S — 2026-05-17

AutismSpectrumDisorder(ASD) is a neurodevelopmental disorder in which early screening is animportant aspect that helps in improving the intervention outcomeand overall developmental trajectory of the individual.

Explainable, Interactive, and Learning-Oriented AI-Based Debugging Assistant

Arathy krishna ,Ms.Raihanath A S — 2026-05-17

Artificial Intelligence (AI)-aided software develop-ment supported by Large Language Models (LLMs) is rapidly transforming modern software engineering by improving pro-ductivity and code quality. Among these applications, AI-baseddebugging assistants have gained attention for their ability

Deep Learning Based Prostate MRI Segmentation and Cancer Detection

Anagha Suresh ,Ms. Navya Davis — 2026-05-17

Prostate cancer is one of the most prevalent malignancies affecting men globally, where timely detection plays a crucial role in improving clinical outcomes. This work presentsan artificial intelligence–based framework for detecting prostate

Smart Stroke Detection System Using Machine Learning and Neuroimages

Anaha Babu,Dr. C Sahaya Kingsly,Ms. Binchu Mol Abraham — 2026-05-17

Stroke remains one of the leading causes of mortalityand long-term neurological disability worldwide. A significant number of severe stroke cases occur due to delayed diagnosis orincorrect interpretation of early warning signs.

LSTM-Based Stock Price Prediction Using Deep Learning Techniques

Sreelakshmi PT ,Ms. Surabhi S Nair — 2026-05-17

Forecasting the stock market is difficult due to its extreme volatility and nonlinearity. For financial organisations and investors, accurate stock price forecasting is crucial. This research suggests a Long Short-Term Memory (LSTM) neural network with deep learning for stock value prediction. The model is trained and assessed using historical stock data from Alphabet Inc.

CipherCrypto: A Hybrid LWE-Based Post-Quantum Framework for Secure and Efficient Multimedia Encryption

Abhiram T K,Jayakrishnan B,Dr. C Sahaya Kingsly — 2026-05-19

The high pace of quantum computing development isa serious risk to the traditional cryptographic systems like RSA and Elliptic Curve Cryptography (ECC) that are based onmathematical problems that can be solved by quantumalgorithms. This requires creation of quantum-resistant security

BLOCKCHAIN-BASED FEDERATED LEARNING FOR SECURE MEDICAL IMAGING

Soumya Srivastava, Ms. Nida Khan, Dr. Syed Hauider Abbas — 2026-05-15

The increasing use of Artificial Intelligence in healthcare has transformed medical imaging diagnostics by enabling automated disease detection, classification, and prediction with high accuracy. However, traditional centralized deep learning systems require collecting sensitive patient data from multiple healthcare institutions into a single server, which raises serious concerns regarding privacy, security, and regulatory compliance. Federated Learning has emerged as a decentralized machine learning approach that allows multiple healthcare organizations to collaboratively train deep learning models without sharing raw patient data. Although Federated Learning improves privacy preservation, it still suffers from vulnerabilities such as malicious model updates, poisoning attacks, unauthorized access, and lack of transparency during model aggregation. To overcome these limitations, this research proposes an implementation-oriented Blockchain-Based Federated Learning framework for secure medical imaging applications. The proposed architecture integrates blockchain technology with federated learning to provide secure model sharing, decentralized verification, tamper-resistant communication, and trustworthy aggregation mechanisms. Smart contracts are employed for validating model updates before aggregation, while blockchain ensures transparency and immutability across participating institutions. Experimental evaluation was performed on medical imaging datasets including Chest X-ray and Brain MRI datasets using Convolutional Neural Network models. The experimental results demonstrated that the proposed framework achieved 96.8% classification accuracy, reduced communication overhead by 22%, and significantly improved privacy preservation and attack resistance compared to traditional federated learning approaches. The proposed Blockchain-Based Federated Learning framework therefore provides a secure, scalable, and efficient solution for next-generation intelligent healthcare systems and distributed medical image analysis.

An AI-Based Implementation Framework for Early Cancer Detection Using Machine Learning Techniques

Ms. Rashida Bano, Mrs. Noorishta Hashmi — 2026-05-15

Early detection of cancer remains a critical challenge in modern healthcare, significantly influencing patient survival rates, treatment effectiveness, and overall healthcare costs. With the rapid advancement of Artificial Intelligence (AI), automated diagnostic systems have emerged as powerful tools to assist clinicians in accurate and timely decision-making. This paper proposes a comprehensive, implementation-oriented AI-based framework for early cancer detection by leveraging both machine learning (ML) and deep learning (DL) techniques. The proposed framework is designed as a multi-stage pipeline that systematically processes medical imaging data, including MRI, CT scans, and histopathological images. It incorporates essential stages such as data acquisition, preprocessing, image enhancement, segmentation, feature extraction, and classification. Advanced preprocessing techniques, including noise reduction, normalization, and contrast enhancement, are applied to improve data quality. Tumor regions are effectively isolated using segmentation methods, enabling precise analysis of the region of interest.

For classification, the framework integrates traditional machine learning algorithms such as Support Vector Machine (SVM) and Random Forest (RF), along with deep learning-based Convolutional Neural Networks (CNN). Comparative analysis across these models is conducted using standard evaluation metrics such as accuracy, precision, recall, F1-score, and Area Under Curve (AUC). Experimental results demonstrate that the CNN-based model significantly outperforms conventional approaches, achieving higher accuracy, robustness, and generalization capability due to its automatic feature extraction and hierarchical learning ability. Furthermore, the proposed system is designed with scalability and real-time clinical applicability in mind, making it suitable for deployment in healthcare environments. The framework not only reduces human intervention and diagnostic errors but also enhances early-stage cancer detection efficiency. Future enhancements include the integration of explainable AI techniques and deployment in cloud-based and IoT-enabled healthcare systems.

BLOCKCHAIN-BACKED DYNAMIC SWITCHING ALGORITHMS FOR HYBRID QKD–PQC NETWORKS

Sarah Samia — 2026-05-15

The rapid advancement of quantum computing has introduced significant security challenges for conventional cryptographic systems used in modern communication networks. Traditional encryption algorithms such as RSA and ECC are vulnerable to quantum attacks due to the computational power of emerging quantum computers. Quantum Key Distribution (QKD) provides theoretically secure communication based on the principles of quantum mechanics, while Post-Quantum Cryptography (PQC) offers software-based cryptographic resilience against quantum-enabled threats using mathematically complex algorithms. However, standalone QKD and PQC systems suffer from limitations including scalability issues, infrastructure dependency, computational overhead, communication latency, and adaptability challenges under dynamic network environments.

This research proposes a Blockchain-Backed Dynamic Switching Algorithm for Hybrid QKD–PQC Networks that intelligently switches between QKD and PQC communication mechanisms based on real-time network conditions, channel quality, attack probability, communication latency, and security requirements. Blockchain technology is integrated into the framework to provide decentralized authentication, immutable transaction management, tamper-proof communication logging, and secure switching validation through smart contracts. The proposed hybrid framework enhances communication security, improves switching efficiency, reduces latency, increases throughput stability, and strengthens resistance against cyberattacks and unauthorized network access. Experimental evaluation demonstrates that the proposed framework significantly outperforms conventional cryptographic systems in terms of security efficiency, adaptive switching accuracy, communication reliability, throughput optimization, and attack mitigation. The integration of blockchain-assisted dynamic switching with hybrid QKD–PQC communication provides a scalable and intelligent solution for next-generation quantum-safe communication infrastructures in healthcare, finance, cloud computing, military systems, and critical industrial networks.

AN INVESTIGATION INTO PATIENT ATTITUDES TOWARD DATA PRIVACY AND CYBERSECURITY IN DIGITAL MEDICAL SYSTEMS

2026-06-03

Background: The use of digital medical systems such as electronic health records (EHRs), telemedicine, and online health platforms has changed the way healthcare is delivered. However, the advancements have also triggered more fears regarding the problem of information security and privacy. The patient's attitude towards the problem is the key to establishing trust and supporting the ethical application of digital health technologies.

Objective: The paper investigates patient awareness, perception, and the level of trust in digital medical systems in terms of privacy as well as cybersecurity of data. It will be necessary to identify what variables will influence the patient's confidence in using digital healthcare platforms and how demographic factors will influence attitudes.

Methods: The qualitative descriptive research design was used, which presupposed the use of a structured questionnaire with 283 patients from the public and private health organizations. The questionnaire was categorized into the awareness section, the concern section, the trust section, and the perception of control. The statistical tests consisted of descriptive statistics, reliability and validity tests (Cronbach's Alpha, KMO, and Bartlett's Test), inferential tests (t-tests, ANOVA, Kruskal-Wallis, Chi-Square, and regression). Statistical tests were performed to test relationships and predictors, both through inferential tests and correlation and regression.

Results: The data sample proved to be very reliable (Cronbach's Alpha = 0.90) and valid (KMO = 0.78; Bartlett’s ?² = 356.42, p < 0.05). Normality was confirmed (p > 0.05). The inferential results showed that there was a significant difference by gender (t = 2.31, p = 0.021) and education (F = 4.78, p = 0.003), which have an overriding influence on trust as well as awareness, respectively. Also, the degrees of concern differ greatly between the types of healthcare (H = 11.26, p = 0.009), and the concern about privacy was found to be related to digital use (2 = 18.53, p = 0.015). All the major variables: Awareness, Concern, Trust, and Control (r = 0.670.91) exhibited a positive correlation. Regression results revealed that Awareness (? = 0.412), Concern (? = 0.376), and Control (? = 0.298) positively impacted the patient trust prediction (? = 0.05).

Conclusion: The findings demonstrate that the perceived power and the attitude toward the belief of being responsible for data privacy significantly condition the confidence of the patients in digital healthcare systems. Exposure and educational background are also huge determinants of attitude. Further improving patient faith in online health services may be achieved via the enhancement of transparency, patient education, and improved online security guidelines.

Federated Threat Intelligence and Explainable Anomaly Detection for Distributed Network Environments in U.S. Critical Infrastructure

2026-06-10

Distributed network infrastructure, cloud computing, the Internet of Things (IoT)and edge based systems have contributed to a dramatic rise in cybersecurityvulnerabilities and the complexity of today's cyber threats. Traditional centralized Intrusion detection systems tend to have problems with scalability

Architecting Intelligence: A Scholarly Examination of SAP Transportation Management

Ajay Bhaktharahalli Nagesh Hemambika, Rajashiva Ramalingam — 2026-06-15

Competitive advantage in modern industry is increasingly determined by how effectively organizationsmanage the movement of goods. Within the broader ecosystem of enterprise software, few platformshave attracted as much technical complexity and strategic weight as SAP Transportation Management.This article undertakes a detailed scholarly investigation

The Online Learning Experience in Malaysian Public Universities: A Systematic Literature Review

Ahmed Saleh Ahmed ,Mohd Norasri Ismail — 2026-06-24

This article systematically explores the online learning experience in Malaysian public universities, synthesizingfindings from studies published in the past five years. Utilizing the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) framework, this review rigorously identified, selected, and analyzedrelevant research to identify key challenges, benefits

A Comprehensive Review of K-Sparse Autoencoder Techniques for Early Alzheimer’s Disease Diagnosis

Saumya Singh ,Somendra Tripathi — 2026-07-14

Alzheimer disease (AD) is a progressive neurodegenerative disorder, which is related to thedeterioration of cognitive functions, memory loss, and change in behavior that profoundly influence thequality of life of those who are affected. Timely intervention, proper management, and enhanced patientoutcomes should be the outcome of early and correct diagnosis. Traditional methods of diagnosis, such as clinical examination, neuropsychological testing, and manual review of neuroimaging data

Recent Advances in Ensemble Learning for Brain Tumor Detection and Segmentation: A Comprehensive Review

Sumit Yadav ,Somendra Tripathi — 2026-07-14

Brain tumours can be considered one of the most acute neurological conditions that haveserious impacts on the central nervous system, and may result in severe cognitive and functional disordersunless they are managed in the initial stages. Proper identification and categorization of brain tumors willbe critical in the development of appropriate treatment plans, diverse disease surveillance, and enhanced survival rate of patients. Medical imaging processes, especially Magnetic

Securing Healthcare Generative AI with Confidential Computing and Zero Trust Principles

Ms. Priyanka Pandey , Prof. (Dr.) Abhay Shukla — 2026-07-14

The adoption of Generative Artificial Intelligence (GenAI) in healthcare is constrained by criticalsecurity limitations that are not adequately addressed by conventional security frameworks—most notably the data-in-use vulnerability, wherein sensitive patient information and proprietary

The Impact of Artificial Intelligence on Big Data Analysis and Digital Transformation

Simran Kumari,Prof (Dr.) Abhay Shukla — 2026-07-14

The integration of Big Data analytics with Artificial Intelligence (AI) has emergedas a key driver of digital transformation across multiple domains. The rapid growth in data volume necessitates advanced computational techniques

A Predictive Deep learning framework for early warning & Risk Assessment of Skin Cancer

Ms. Priyanka Pandey,Prof (Dr.) Abhay Shukla — 2026-07-14

Skin cancer remains a major global health challenge, where timely identification and accurateclassification are essential for reducing mortality and treatment complexity. Despite notableprogress in learning–based medical image analysis, two critical challenges persist: the scarcity of large, diverse medical datasets and the difficulty of effectively optimizing trainable

Diabetes Prediction Using machine learning

Sakshi Verma,Dr. Abhay Shukla. — 2026-07-14

Diabetes mellitus is a chronic metabolic disorder that affects a large proportion of theglobal population and poses a significant public health challenge. Early identificationof individuals at risk is essential for preventing severe complications and improvinglong-term health outcomes. In this study, machine learning–based classification

Scalable Container - Based Backend Architecture for Social Web Applications Under High Concurrent Workloads

2026-07-23

Social web apps that serve small groups of people have a lot of trouble keeping up with high workloads,which can lead to slow response times, server errors, and resource exhaustion during peak usage. Thisresearch proposes and assesses a scalable container-based backend architecture implemented withina hosting panel environment, tailored for minimal latency andefficientresource utilization in socialnetworking platforms. The architecture uses separate Docker containers forthreebackendimplementations: Rust (Rocket), PHP (Laravel), and Node.js (Express.js). Each of these implementsthe same socialnetworkingfeatures with 20 CRUD routes that all use the same MariaDB database. AnNginx reverse proxy sends traffic to the rightdomain, and a React.js SPA frontend makes it easy toswitch between backends without affecting frontend performance

A Systematic Literature Review of Volunteered Geographic Information Quality Assessment: Methods, Challenges, and Applications

Abobaker Ali Ahmed Abdulwahab , Dr. Asmala Ahmad — 2026-07-23

Volunteered Geographic Information (VGI) has emerged as an important source of geospatialdata, enabling citizens to contribute spatial information through collaborative mappingplatforms such as OpenStreetMap (OSM). The increasing use of VGI in urban planning,transportation, environmental monitoring, disaster management, and humanitarian operations has raised significant concerns regarding data quality and reliability. Therefore,assessing the quality of VGI has become a critical area of research within GeographicInformation Science (GIScience).

Predictive DVFS and Compressed Memory Enabled RV32IMC RISC-V SoC Architecture for Energy-Efficient IoT Edge Computing

Bhavithra Ramu , Susithra Arulvelan, Prabavathi Sundhar, Suguna Sandhaseelan — 2026-07-23

This paper proposes a power-efficient RV32IMC-based RISC-V System-on-Chip (SoC) architecturefor Internet of Things (IoT) edge computing with limited power and memory resources. Traditionalcloud computing approaches with higher latency, larger bandwidth consumption, and privacy issues,along with higher energy consumption, make edge computing more suitable for real-time IoTapplications. But as edge devices, smart power management, memory optimization, and performancescalability in response to varying workloads are essential. To address these issues, the design includes three key elements: predictive Dynamic Voltage

A Low-Overhead Hierarchical Parent–Child CAM Based BIRA Architecture for Memory Repair

Kanteti Himabindu,Dr. K.Babulu,Dr.G.N.Swamy — 2026-07-23

As the densities of the memories in system on chip is increasing dramatically, test cost and yield improvement hasbecome the key challenge. Built in redundancy analysis (BIRA)is widely used to improve the yield by replacingthe faulty cells with redundant structures. The most common form of the BIRA is a module that stores and analysesof fault address with a two-dimensional spare architecture. BIRA Performance primarily determined by decreaseof Area overhead, optimum repair rate and the speed with which the analysis is performed. Efficiency of BIRA improves mainly due to the fault storing methodology and redundancy analysis procedure

A Blockchain based Framework for secure Electronic Health Records using AP PBFT and Integrated IDHG-ECC Encryption

Lakshmi Teja Tummala ,Dr.D.Kalyani — 2026-08-12

Electronic Health Records (EHRs) do not have astrong framework or an adaptable structure to guardhealth information while computerizing medicaltasks. Existing blockchain based EHR compete in achieving confidentiality, integrity, usability,lowresponse time in real healthcare scenarios. Thisendeavor presents a prototype secure EHR platformthat combines an IDHG-ECC-inspired elliptic curvecryptographic method with a hybrid blockchain and InterPlanetary File System