		FeedBack \| Contact Us \| Links \| Site Map

Welcome to Journal of Theoretical and Applied Information Technology

Submit Paper / Call for Papers

Journal receives papers in continuous flow and we will consider articles from a wide range of Information Technology disciplines encompassing the most basic research to the most innovative technologies. Please submit your papers electronically to our submission system at http://jatit.org/submit_paper.php in an MSWord, Pdf or compatible format so that they may be evaluated for publication in the upcoming issue. This journal uses a blinded review process; please remember to include all your personal identifiable information in the manuscript before submitting it for review, we will edit the necessary information at our side. Submissions to JATIT should be full research / review papers (properly indicated below main title).

Journal of Theoretical and Applied Information Technology
December 2024 | Vol. 102 No.23

Title:	STRATIFICATION OF BREAST CANCER IMAGES BY UTILIZING SPATIAL INFORMATION AND DEEP LEARNING MODEL
Author:	RATHLAVATH KALAVATHI, Dr M SWAMY DAS
Abstract:	For the purpose of health care imaging, accurate breast cancer (BC) identification and classification is a critical task because of breast cancer tissue is too complex. BC is primary reason for women death with cancer. Due to potential strongest of deep learning in extraction of dominant features. In this paper, hybrid model is proposed for automated identification and treatment of BC namely EDNet-SVM i.e. Encoder-Decoder Net with SVM i.e., Support Vector Mechanism. The deep potential characteristics are extricated from EDNet and then breast cancer classification is done using SVM. EDNet is composition of Encoder and decoder and features are derived using long short term memory (LSTM). The segmented images are constructed using histogram equalization and morphological operations and feed to the EDNet for deriving optimal features from the breast cancer images which includes local features. The proposed EDNet-SVM model shown the high accuracy of 98.44% on considered freely available dataset BreastHis dataset and 96.34% on MIAS dataset. This paper listed performance values of three kernels of SVM (linear, cubic and Gaussian). The test outcomes concluded that the proposed model is superior than the existing state of art models.
Keywords:	Identification, Breast cancer, Support vector mechanism, long short term memory, local features.
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	DYNAMIC PROGRAMMING-ENHANCED ENERGY-EFFICIENT TASK SCHEDULING IN EDGE-CLOUD ENVIRONMENTS
Author:	DR.SARAVANAN.M.S, MRS. MADHAVI KARUMUDI
Abstract:	In the wake of modern Internet of Things (IoT) use cases and workflow applications, edge computing plays a crucial role in the faster execution of specific tasks. Therefore, the edge-cloud environment offers computing resources required by modern applications. Using cloud resources for every task may lead to violation of Service Level Agreements (SLAs), as well as causing impediments to the execution of tasks with deadlines. In this research, we suggested a system concept and architecture for effective task scheduling in an edge-cloud setting to address this issue. An algorithm called Dynamic Programming based Energy Efficient Task Scheduling (DPEETS) is introduced for effective work scheduling in edge-cloud environments. Our algorithm exploits dynamic programming, hamming distance termination, and randomization to optimize decisions made in the edge cloud about task scheduling. This suggested algorithm uses a dynamic programming table to keep track of iterations and eliminate unnecessary computations. DPEETS considers multiple objectives such as deadline, energy efficiency, and latency of task execution. The algorithm heuristically converges in a few iterations, leading to energy-efficient scheduling of tasks in the edge cloud. A simulation study has revealed that DPEETS outperforms many existing heuristic algorithms.
Keywords:	Task Scheduling, Cloud Computing, Edge Computing, Dynamic Programming, Energy-Efficient Task Scheduling
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	A DISTRIBUTED ADAPTIVE ALGORITHM FOR EFFICIENT LOCALIZATION OF SENSOR NODES IN AD HOC NETWORKS
Author:	PINAMALA SRUTHI, DR. Y AMBICA, DR. B.N.V. UMA SHANKAR, PRATHAP ABBAREDDY, VASAVI OLETI
Abstract:	Node localization in a wireless ad hoc network of sensor nodes is essential for making informed decisions and maximizing the network's effectiveness and efficiency. Localization is often based on different measures with centralized distributed approaches. With empirical study, we found that the distributed approach to node localization is more robust than its centralized counterpart. Therefore, in this paper, we proposed an algorithm for sensor node localization in a distributed approach. Our algorithm, Distributed Optimized and Adaptive Node Localization (DOANL), exploits a collaborative method that can achieve efficient localization even in a smaller number of beacons. The foundation of our methodology is the Angle of Arrival (AOA) and Time of Arrival (TOA). Node and beacon deployment locations have an impact on the localization process as well. This work presents the Distributed Optimized and Adaptive Node Localization (DOANL) algorithm, which enhances localization accuracy in distributed wireless ad hoc networks by leveraging AOA and TOA with minimal beacons and minimizing error propagation, outperforming state-of-the-art. This paper also investigates this proposition by analyzing node deployment and network connectivity. We also found that error propagation across the network limits the accuracy of the proposed algorithm. To address this problem, we further optimized the DOANL algorithm by minimizing error propagation across the network. The suggested approach outperforms state-of-the-art ones in wireless ad hoc network localization, according to an empirical research conducted with an NS-3 simulation study.
Keywords:	Wireless Ad Hoc Network, Wireless Sensor Network, Node Localization, Distributed Adaptive Localization, Error Propagation.
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	ADVANCED INSAR TECHNIQUES FOR LANDSLIDE DETECTION AND RISK ASSESSMENT: A CASE STUDY OF THE TAZA-AL HOCEIMA EXPRESSWAY IN MOROCCO
Author:	ISHAK HBIAK , OMAR BACHIR ALAMI , ANOUAR KESMAT , NIA MAJDA, SAID RHOUZLANE , MERIEM HABIBELLAH
Abstract:	The Taza-Al Hoceima expressway is a vital infrastructure that facilitates the integration of the region into Morocco's economic dynamics. Several challenges related to the difficult geotechnical context of the northern region arose during the construction period, and additional challenges continue to emerge concerning the sustainability of this infrastructure. Indeed, the complex geology and topography of the Taza province make it prone to land movements, posing risks to critical infrastructure. The objective of this study is to employ advanced Interferometric Synthetic Aperture Radar (InSAR) techniques to detect, monitor, and measure land movements, particularly landslides, along the infrastructure, using Sentinel-1 radar data covering the period from 2015 to 2020. Our work also focuses on explaining the instability movements measured during the study period by identifying their main causative factors. The study's findings also enable the identification of certain high-risk instability zones within our geologically complex study area, which require long-term monitoring.
Keywords:	Geotechnics, Remote sensing, Landslide, Synthetic Aperture Radar, PCA
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	PREDICTING DAILY RIVER FLOW USING LONG SHORT-TERM MEMORY (LSTM): A DEEP LEARNING APPROACH
Author:	ABDEL KARIM BAAREH, MONTHER TARAWNEH, FAISAL ALZYOUD, IBRAHIM ALTARAWNI MOHAMMED AMIN ALMAIAH, ROMMEL ALALI, TAYSEER ALKHDOUR, MAHMAOD ALRAWAD, THEYAZN H. H ALDHYANI, RAMI SHEHAB
Abstract:	Deep learning is a large-scale model in machine learning that uses multi-layer neural networks, which mimic the complex structure of the human brain. Combinations of layers and neurons in artificial neural networks create excellent practical applications. This paper explores the use of Long Short-Term Memory (LSTM), a subset of recurrent neural networks, known for its complexity and versatility in deep learning. The primary goal involves using LSTM to address the challenge of predicting daily river flow for two prominent rivers in the USA. Real datasets related to the daily flow of the Black and Gila Rivers were used and divided into different sets for training and testing. A comparative analysis was performed between the training and test sets, and error metrics were evaluated to confirm the effectiveness of the LSTM model. The experimental results collected from this study using LSTM were remarkably good, and showed significantly low error values, demonstrating its effectiveness in predicting river flow.
Keywords:	Forecasting, Deep Learning, Long Short-Term Memory, Artificial Neural Network, and Recurrent Neural Networks
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	A DWT AND PATTERN RECOGNITION APPROACH FOR FAULT DETECTION AND CLASSIFICATION IN TRANSMISSION NETWORKS
Author:	G PRABHAKAR RAJU, V SRINIVASA RAO, G ASHOK KUMAR, RAMU BHUKYA, PRADEEP JANGIR, SARIHADDU KAVITHA, B VARAPRASAD RAO
Abstract:	The reliable and efficient operation of high-voltage transmission lines is essential for ensuring the stability and quality of electrical power distribution. To address this concern, this research paper presents an in-depth study on the application of wavelet transform for detecting and classifying faults in high-voltage transmission lines. Fault detection and classification are crucial tasks in the maintenance and operation of power systems to minimize downtime and ensure the safety of personnel and equipment. The wavelet transform has proven to be a powerful tool for analyzing transient signals in electrical systems, making it a valuable technique for fault detection and classification. This paper provides a comprehensive review of wavelet transform theory, its application to fault detection, and classification algorithms. Additionally, it discusses various case studies and practical implementations, highlighting the advantages and limitations of wavelet-based techniques. The results demonstrate the effectiveness of wavelet transform in enhancing the reliability and efficiency of high-voltage transmission line monitoring and maintenance.
Keywords:	High-Voltage Transmission Lines, Wavelet Transform, Faults, Power Systems
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	OPTIMIZING CROP RECOMMENDATION SYSTEMS USING ADVANCED DEEP LEARNING TECHNIQUES
Author:	PAPPALA MOHAN RAO, KUNJAM NAGESWAR RAO
Abstract:	The agricultural sector plays a significant role in economic development, especially for developing nations, but farmers experience hurdles such as climate change interventions, soil management, and crop yield choices. To this end, this study turns to Deep Learning (DL) to solve these problems and develop a holistic approach that employs a range of DL approaches for crop selection based on soil condition, climate, and past farming records. Predicted models and methods: These are Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), and Transformers; precision agriculture through crop prediction; data-driven decision-making on resource allocation to boost farm revenue; and soil management through the prediction of the soil pH and recommended crops. This plan of action is designed to help farmers make the most beneficial decision to enhance the farming results.
Keywords:	Agricultural, Deep Learning (DL), Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), Transformers, crop recommendation.
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	THE IMPACT OF ARTIFICIAL INTELLIGENCE ON HUMAN RESOURCE MANAGEMENT PROCESSES
Author:	ZHENGYAO FENG, JONGCHANG AHN
Abstract:	This article explores the impact of Artificial Intelligence (AI) technology on Human Resource Management (HRM) processes, aiming to reveal the role of AI in improving HRM efficiency and enhancing employee experience and engagement. Firstly, a literature review indicates that AI technology can significantly improve the efficiency of recruitment, training, performance management, and employee support. Additionally, AI technology can enhance employee work experience and engagement by utilizing personalized training and intelligent employee care systems. Subsequently, it conducted a questionnaire survey among 206 employees engaged in HRM or other management roles, and used quantitative research methods to analyze the survey data statistically. The analysis revealed that respondents generally believe AI positively enhances HRM efficiency and improves employee experience and engagement. They also showed a high acceptance and adaptability to AI technology, although there were some concerns about its potential risks. Multiple linear regression analysis showed that perceived risk has a significant positive impact on job performance, while the influence of other factors is not significant. The article suggests that companies should focus on managing perceived risks when applying AI technology to ensure it has a positive impact on business outcomes, and it also emphasizes the importance of improving employee experience and engagement. This research provides theoretical support and practical guidance for the application of AI technology in the HRM field, and it also points out possible directions for future research.
Keywords:	Artificial Intelligence (AI), Human Resource Management (HRM), Efficiency, Employee Experience, Perceived Risk
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	USING ARTIFICIAL INTELLIGENCE TO IMPROVE TAX SECURITY AND CONTROL OVER TAX AVOIDANCE SCHEMES
Author:	YEVHEN TSIKALO, OLEKSANDR ZINEVYCH, DENYS OSIPENKO, VIKTORIYA KULYK, OLENA LAGOVSKA
Abstract:	Tax security is a critical area in modern states in view of its importance for ensuring the state budgets’ replenishment and reducing the shadow economy size. At the same time, a serious obstacle to tax security is the development of tax evasion schemes using modern technologies. The aim of the research is to analyse the impact of artificial intelligence (AI) on the volume of losses from tax abuse in European countries. The work uses the methods of statistical analysis, case study, regression and correlation analysis. The conducted research confirmed the initial assumption about the impact of the AI development and the amount of losses arising from tax abuse. This was achieved by identifying a statistically significant impact of the Intercept, Talent, Infrastructure and Commercial variables on the Annual tax loss: Corporate tax abuse (% of GDP). The relationship is inverse with Intercept and Talent, and positive with Infrastructure and Commercial. It follows that the development of infrastructure and the active commercial use of AI is accompanied by an increased volume of tax losses from corporate tax abuse. Increased level of human capital development, expressed through the Talent indicator, is accompanied by a decreased amount of tax losses. Accordingly, the human capital development can become one of the main success factors along with the establishment of legal restrictions and the use of technology-based countermeasures. The research findings can be useful in the process of updating the legal framework regarding the limitations of the AI use and the development of an appropriate ethical framework.
Keywords:	Artificial Intelligence, Corporate Tax Abuse, Tax Evasion, Tax Fraud, Tax Security.
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	IMPROVING USABILITY THROUGH A GAMIFICATION-BASED E-LEARNING SYSTEM WITH MDA FRAMEWORK
Author:	SALOMO H TIMOTHY PARDEDE , TANTY OKTAVIA
Abstract:	This research aims to develop a gamified e-learning platform of PT. Adhi Karya, a construction engineering company in Indonesia, using the mechanics, dynamics, and aesthetics (MDA) framework. The effect of gamification on the e-learning system was examined using a questionnaire that measured usability, which consisted of efficiency, errors, learnability, memorability, and satisfaction. Purposive sampling technique was employed to recruit the participants, and validity and reliability tests were conducted before the questionnaire was distributed to the participants. Our results show the successful implementation of gamification into PT Adhi Karya's e-learning system, where the majority of the respondents considered that the system had good usability, demonstrating the suitability of the framework for gamification-based e-learning system.
Keywords:	E-Learning System, Gamification, MDA Framework, Usability
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	FEATURE REDUCTION AND STROKE PREDICTION USING SPARSE SUBSPACE CLUSTERING AUTOENCODER ON CLINICAL DATA
Author:	DURGA DEVI. P, K. AKILA
Abstract:	The use of deep learning techniques for feature reduction presents a novel way to improve stroke prediction, as demonstrated by the article "Feature Reduction and Stroke Prediction using Sparse Subspace Clustering Autoencoder on Clinical Data". The volume and complexity of clinical data might be challenging for traditional approaches to process, which can result in unsatisfactory prediction results. However, this work offers a novel method that sorts through a lot of clinical data to find and improve the most relevant parts using deep learning algorithms. By employing state-of-the-art neural network topologies, the proposed approach effectively identifies significant stroke-related predictors, enabling more accurate and quick risk assessments. Early stroke prediction is essential for emergency treatment and consequence avoidance. Clinical data used to predict stroke is often high-dimensional, making evaluation and interpretation challenging. This study proposes a novel feature reduction technique for clinical data utilizing the sparse subspace clustering autoencoder (SSC-AE) to predict strokes. The SSC-AE approach is contrasted with other deep learning algorithms, such as CNNs, RNNs, and LSTMs. The SSC-AE technique works better than other algorithms in terms of reconstruction error, clustering performance, and stroke prediction accuracy, according to experimental results. The proposed approach might improve stroke prediction models' efficiency and accuracy.
Keywords:	Neural Network Algorithms, Deep Learning, Stroke Prediction, Feature Reduction, Clinical Data Analysis.
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	EMPIRICAL INVESTIGATIONS TO DETECTION OF LIVER CANCER USING MOBILEUNET
Author:	K. RAMANJANEYULU, P. SAROJA, P. VENU MADHAV, SRINU PYLA, VENKATA BHUJANGA RAO MADAMANCHI, VENKATESWARARAO CHEEKATI
Abstract:	Identification of cancer tissue and biopsy is a procedure which is nevertheless however a delicate and complex process as of now. Cancer CT images can be segmented in order to help with the medical plan interventional and clinical response assessment of liver lesions. Mobile U-Net, which has been shaped and implemented as a reliable tool, is used for the segmentation of liver tumour and to address the existing liver cancer problem. Liver lesions in Computerized tomography may be used for assessing the volume and location of tumours, predicting their treatment and assessing condition of the patient. This is the amendment of the U-Net model structural design which takes been designed specifically for the usage in mobile platforms. The deep learning structure decodes the perception both by putting some enlightenment on what characteristics are involved in internal layer examination and forecast and by illuminating the part of a working model of a trained deep neural networks beforehand.
Keywords:	Deep learning ,U-Net, Mobile Optimization, Liver Tumor segmentation, Computed Tomography
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	LEVERAGING RESNET-152 AND WEB TECHNOLOGY FOR RAPID COVID-19 DIAGNOSIS FROM X-RAY IMAGE
Author:	JENNY LING SIAW NIE, SOO SEE CHAI, KOK LUONG GOH, KIM ON CHIN
Abstract:	In December 2019, the SARS-CoV-2 virus gave rise to COVID-19, which was first detected in Wuhan, China. The virus has infected over 700 million individuals on Earth. This virus can spread through direct and indirect contact, making humans vulnerable even in small places or through food consumption. The pandemic highlighted challenges, including a shortage of radiologists and the time-intensive interpretation of X-ray images, leading to discrepancies and delays. To address this, a classification model based on X-ray images became crucial for COVID-19 identification. Proposing a web-based system integrating convolutional neural network (CNN) models, particularly the ResNet-152 model, aims to enhance precision in monitoring and diagnosing COVID-19. After fine-tuning a pre-trained ResNet-152 model using transfer learning on a COVID-19 dataset and adding a classification head, a COVID-19-specific classification model is created. In this project, the pre-trained COVID-19 ResNet-152 model achieved 86.84% accuracy, 89.95% sensitivity and 77.27% specificity. The model is then integrated into the system, which enables healthcare professionals to upload and receive a clear visualisation of the COVID-19 classification results via Application Programming Interface (API) endpoints. This platform enables healthcare professionals to login, upload, search, and classify COVID-19 diagnoses based on the uploaded X-ray pictures, providing an intuitive interface and a user-friendly system. Leveraging advanced image processing and deep learning, the system has the potential to expedite accurate diagnoses and alleviate the workload on healthcare professionals, ensuring swift and accurate detection of COVID-19 cases.
Keywords:	COVID-19, Classification, X-ray, ResNet-152 Model, Deep Learning
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	IDENTIFICATION AND CLASSIFICATION OF MANGO LEAVES DISEASE ON GREY WOLF OPTIMIZATION OF MULTIVARIATE GATED RECURRENT NETWORK
Author:	AMIRTHA PREEYA V , DR.S.PRAVINTH RAJA
Abstract:	Plant provides essential nutrients and energy for daily life on reducing sickness, stress and anxiety through its process of cleaning air. Further it provides many medicinal and health benefits to the society to treat various degenerative disease. Especially mango leaves are potential source of minerals like nitrogen, potassium, calcium, magnesium and vitamins. In Particular, these most beneficial plant were exposed to various disease in its growing regions due to several environment changes. Hence it becomes mandatory to protect and prevent the plant against the disease using precautionary measures. Even though there exist large solutions to combat these issues still it requires a strong solutions to examine those disease and categorize it into various forms to establish a effective countermeasures. Many researches has been initiated using machine learning and deep learning to detect the disease and its types but those approaches fails to discriminate diseases effectively in early stages. In this work, a new deep learning model entitled as Multivariate Gated Recurrent Network composed of Input layer, hidden layer , abstract layer and output layer contains various state like hidden state, forgot state, update state and Reset state is employed to analyze the disease region of the image in detail towards effective discrimination in early stages. In particular abstraction layer has been combined with graph cut segmentation algorithm to improves the efficiency of the network by searching and grouping of seed points which represents the infected region of leafs has common attributes. Those common attributes generate the best features for classification. Further optimization of gated recurrent network is carried out using grey wolf optimization to eliminate the network over fitting issues on increasing the speed and accuracy of the network to identify and classify the regions infected of different diseases on the plant leafs on different diseases perspectives. Experimental analysis of proposed architecture on cross fold validation of plant village dataset explains the model accuracy and efficiency with respect to execution performance in terms of training accuracy and validation accuracy respectively for disease identification. GWO and multivariate GRU work together to automate disease diagnosis with low computing overhead, advancing accurate, real-time agricultural disease management. This research advances our understanding of plant disease detection while providing a scalable, reliable, and optimal approach that may be modified for comparable horticultural and agricultural uses.
Keywords:	Mango Leaf Disease, Deep learning, Recurrent Neural Network, Multivariate Gated Recurrent Network, Grey Wolf Optimization algorithm.
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	AUGMENTED CLUTCH REALITY MOBILE: INNOVATIVE EDUCATIONAL MEDIA FOR MODERN ENGINEERING
Author:	MUSLIM, AMBIYAR, ARWIZET KARUDIN, MUHAMMAD SYAFIQ HAZWAN RUSLAN, HSU-CHAN KUO, NUZUL HIDAYAT, RIDO PUTRA
Abstract:	TOne of the interactive technologies that can be applied to improve the quality of education in the era of the 4.0 revolution is augmented reality (AR). The purpose of this study is to develop interactive media augmented clutch reality based on mobile (ACRMobi) in the powertrain course. In addition, it analyzes the responses of lecturers and students to the implementation of ACRmobi as a modern vocational education media in the 21st century. The 4D model is used as a media product development model while the instruments used are validation and response questionnaires to the media developed. It uses descriptive qualitative and quantitative analysis techniques to calculate the average score by looking at the criteria for the results. The findings obtained in the form of ACRmobi media that was developed are valid based on expert assessments and practical based on lecturer responses and student responses after being implemented at one of the vocational education institutions. Experts who act as validators report that ACRMobi is an interactive media that can be applied to vocational education. ACRMobi contains elements that can increase student motivation because it is technology-based. Integrated learning ACRMobi is more motivating and more flexible in its application.
Keywords:	Augmented Reality, Powertrain, 4D Model, ACRMobi, Quality Education
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	SOLVING PROBLEMS OF BIG DATA INFRASTRUCTURE BY USING BLOCKCHAIN
Author:	MOHAMED ESMAIL, MOHAMED A. EL-DOSUKY, TAHER T. HAMZA
Abstract:	Big data faces challenges like substructure security, data confidentiality and data administration. With the inception of blockchain, massive data security circulation has become possible. The paper surveys previous work, focusing on security challenges in big data models, a summary of blockchain services in a big data environment, and the challenges of research in big data working together with blockchain. Proposed methodology begins with the mathematical foundation of big data-blockchain mapping. Then the proposed big data blockchain infrastructure is proposed. The infrastructure allows many operations such as separating and storing data, querying the separated stored data, block validation, and gossip protocol. This work established an archetype system using python programming language to confirm that the suggested concept of isolating and information storage is applicative to big data controlling systems on blockchain technologies. Finally, forged block attack is scrutinized. The methodology integrates big data with blockchain through a proposed infrastructure, validated via a Python prototype, enabling data management and addressing forged block attacks.
Keywords:	Big data, Blockchain, Security
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	OPTIMIZING DIABETES DIAGNOSIS: ADGB WITH HYPERBAND FOR ENHANCED PREDICTIVE ACCURACY
Author:	SWAPNA DONEPUDI, G.N.V.G. SIRISHA, PAPPULA MADHAVI, S PHANI PRAVEEN, DESHINTA ARROVA DEWI, MUSTAFA JABER, Massila Kamalrudin
Abstract:	This study introduces an innovative machine-learning framework to enhance diabetes prediction accuracy and model interpretability. The methodology begins with multiple imputations by chained equations (MICE) to address missing data and ensure a complete dataset for analysis. To tackle class imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) is employed. Z-score outlier detection is utilized to remove outliers, further improving model robustness. A hybrid feature selection method hybrid GWAN combining Grey Wolf Optimizer (GWO) and ANOVA optimizes selecting relevant features, balancing predictive power with model simplicity. The core of the framework is the Adaptive Boosted Gradient Boosting Machine (ADGB), an ensemble learning model that merges the strengths of AdaBoost and Gradient Boosting Machines (GBM). Hyperparameter optimization through the Hyperband algorithm fine-tunes the model, achieving a high prediction accuracy of 97.84%. This comprehensive approach not only improves accuracy but also enhances the precision, recall, and F1 score of the predictive model. By integrating these advanced techniques, the framework demonstrates significant potential in early diabetes diagnosis, emphasizing the importance of ensemble methods in healthcare data analysis and the necessity of accurate, interpretable models for developing reliable diagnostic tools.
Keywords:	Grey Wolf Optimizer, Gradient Boosting Machines, Synthetic Minority, Public Health
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	ENHANCING CYBER ATTACK DETECTION IN NETWORK TRAFFIC USING ADAPTIVE REGRESSION TECHNIQUES
Author:	Dr.TALLURI.SUNIL KUMAR, P.JYOTHI, Dr.RAJESH KUMAR VERMA, 4PADMINI DEBBARMA, Dr.N.V.S.PAVAN KUMAR, I.NAGA PADMAJA, HYMAVATHI THOTTATHYL, Dr.M.SATHISH KUMAR
Abstract:	Cyber attack detection is pivotal for preempting threats, securing data, and safeguarding critical systems against breaches in our digitally reliant world, ensuring uninterrupted operations and user privacy. Timely detection of cyber attacks is paramount to prevent potential damages, financial losses, and reputational harm inflicted upon individuals, organizations, and critical infrastructure.The proposed algorithm, "AdaptoReg," introduces a novel approach to cyber attack detection within network traffic using the NSL-KDD dataset. By integrating adaptive regression techniques inspired by Lasso Regression and Ridge Regression, this algorithm aims to dynamically adapt to diverse attack patterns while maintaining robustness against evolving cyber threats. Through feature engineering and an ensemble strategy reminiscent of Random Forest Regression, "AdaptoReg" identifies anomalies in network behavior, offering a comprehensive solution for detecting and flagging potential cyber attacks. The algorithm undergoes rigorous evaluation, demonstrating its effectiveness in accurately identifying malicious activities and highlighting its potential as a valuable tool in enhancing network security and mitigating cyber risks.
Keywords:	Cybersecurity, Network Intrusion Detection, Adaptive Regression, Anomaly Detection, NSL-KDD Dataset
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	A HYBRID LION PRIDE AND BAT ALGORITHM (HLPBA) FOR OPTIMAL SPOT AND SIZE OF EV CHARGING STATIONS IN DISTRIBUTION NETWORKS
Author:	RAGALEELA DALAPATI RAO, PADMANABHA RAJU CHINDA, KUMAR CHERUKUPALLI
Abstract:	A dramatic increase in the number of EVs on city streets has led to a dramatic increase in the need for well-placed EV charging stations inside distribution networks. Electric vehicle charging station (EVCS) installation is crucial to avoid power losses, voltage instability, and network overload. In response to these issues, this study presents a new algorithm called Hybrid Lion Pride and Bat Algorithm (HLPBA) for determining the best size and location of electric vehicle charging stations (EVCS) in radial distribution networks. By integrating the strengths of the Bat Algorithm (BA) for global exploration and the Lion Pride Optimization Algorithm (LPOA) for local exploitation, the HLPBA is able to strike a good balance in its search for optimal solutions. The suggested algorithm's goal is to keep the voltage stable across the network while minimizing total active power losses. By implementing the HLPBA into the IEEE 33-bus radial distribution system, power losses are reduced by 72.5% and the minimum bus voltage is improved to 0.98 p.u. The results show that the HLPBA outperforms more conventional optimization methods like Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), therefore making it a great choice for distribution systems' EVCS placement.
Keywords:	Electric Vehicle Charging Stations (EVCS); Distribution Network Optimization; Power Loss Minimization, Voltage Stabilit; Hybrid Optimization Algorithm; Lion Pride Optimization Algorithm (LPOA); Bat Algorithm (BA); Hybrid Lion Pride and Bat Algorithm (HLPBA).
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	A HYBRID APPROACH COMBINING LONG SHORT-TERM MEMORY AND RANDOM FOREST FEATURE SELECTION FOR NETWORK INTRUSION DETECTION
Author:	HANDRIZAL, FAUZAN NURAHMADI, ALBERTMAN PUTRA BARASA
Abstract:	Network intrusion detection is essential for identifying suspicious network activity. However, with technological advancements and the increasing need for data, data processing becomes a significant challenge in intrusion detection. This research explores a detection system using Long Short-Term Memory (LSTM) combined with Random Forest-based feature selection (RFUTE) to enhance performance. RFUTE reduced data dimensions by selecting 29 features for binary and 30 for multiclass classification from 43 total features. Tested on the NF-UQ-NIDS-v2 dataset, the system showed significant improvements, achieving 99% accuracy in binary classification and 95% in multiclass classification. Additionally, the AUC-ROC curve of the model with RFUTE showed better performance than the model without RFUTE, with an increase in AUC to 0.999 for binary classification and 0.9987 for multiclass classification. These findings demonstrate that the integration of Long Short-Term Memory networks with Random Forest-based feature selection significantly improves the accuracy and predictive performance of network intrusion detection systems, particularly in large-scale and complex environments. This research contributes to the development of a more accurate and efficient model for network intrusion detection by integrating Long Short-Term Memory networks with Random Forest-based feature selection. This model improves the ability to identify network threats quickly and responsively.
Keywords:	Network Intrusion Detection, Binary Classification, Multiclass Classification, LSTM, Random Forest Feature Selection (RFUTE), Feature Selection.
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	DETECTION OF BREAST MASSES USING WATERSHED AND GLCM TEXTURE FEATURES
Author:	MOHAMMED RMILI , MOUSTAPHA M. SALECK , ABDELMAJID EL MOUTAOUAKKIL , ABDELLATIF SIWANE
Abstract:	Nowadays, breast cancer is the primary cause of the increase in the rate of death among women in both developed and developing countries. Examining mammograms to find signs of breast lesions is a challenging task that radiologists have to perform often. Thus, it is crucial to implement image analysis techniques to detect and outline breast lesions, as they provide essential morphological data that can assist in ensuring an accurate diagnosis. In this paper, we propose an efficient split-and-merge approach for tumor segmentation in digital mammogram images, using a watershed algorithm and texture features. First, we adopt the median filter technique to reduce noise and enhance the quality of the mammograms. Second, we apply Contrast-Limited Adaptive Histogram Equalization (CLAHE) to improve the interpretability of texture analysis results. Third, we exploit the watershed algorithm to split the image into homogenous subregions. Finally, we use the Gray Level Co-occurrence Matrices (GLCM) technique to localize the suspected subregions and merge these subregions to detect and outline breast lesions. The mini-MIAS database is used to evaluate the efficiency of the proposed approach. The experimental results demonstrate that the proposed hybrid method achieves a highly satisfactory success rate compared to state-of-the-art breast lesion segmentation methods in mammogram images.
Keywords:	Digital mammogram, tumor segmentation, Watershed algorithm, texture features, Split-and-Merge
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	EXPLORING ENHANCEMENT IN CONFIDENCE-BASED ASSESSMENT: A SYSTEMATIC LITERATURE REVIEW
Author:	NUR MAISARAH NOR AZHARLUDIN, KHYRINA AIRIN FARIZA ABU SAMAH, MOHAMAD FAIZ BIN DZULKALNINE, RASEEDA HAMZAH
Abstract:	Confidence-based assessment (CBA) is a technique designed to evaluate an individual’s degree of confidence or expectation regarding their response to discern their true level of knowledge. In this methodology, individuals assign confidence scores to their answers, indicating their level of certainty about the correctness of their choices. This approach enhances understanding an individual’s abilities or comprehension by distinguishing between correct responses with high confidence and those with low confidence. Consequently, evaluators gain a more comprehensive understanding of an individual’s competence by examining their cognitive processes and self-awareness. Despite its potential, there is a lack of systematic reviews focusing on enhancing CBA. This study addresses this gap by conducting a systematic literature review (SLR) on improving CBA methodologies. The present study follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, systematically analysing 11 articles published between 2019 and 2024. These articles were selected from three primary databases—Scopus, Web of Science, and ScienceDirect—and one supplementary database, Google Scholar. The review of these studies identified four major themes: sector, purpose, algorithm, and methods used. The findings of this SLR provide valuable insights into the current state of CBA research and suggest directions for future studies. In conclusion, this research offers significant benefits for scholars in the CBA field, providing a reference for enhancing the application and understanding of CBA.
Keywords:	Confidence, Confidence-Based Assessment, Confidence-Based Learning, Enhancement, Systematic Literature Review
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	DUAL Q-LEARNING WITH GRAPH NEURAL NETWORKS: A NOVEL APPROACH TO ANIMAL DETECTION IN CHALLENGING ECOSYSTEMS
Author:	JOHNWESILY CHAPPIDI, DIVYA MEENA SUNDARAM
Abstract:	Detecting wild animals is crucial to prevent road accidents caused by their crossings and mitigate intrusions into residential areas. Existing methods often struggle with complex spatial contexts and environmental variability. This study introduces an integrated approach using Graph Neural Networks (GNNs), advanced Q Learning, Multi-Attribute Utility Theory (MAHP) with deep learning, and Generative Adversarial Networks (GANs) for data augmentation. The model enhances spatial awareness with GraphSAGE and Graph Attention Networks (GAT), employs Deep Q-Networks (DQN) for adaptive decision-making, integrates MAHP with custom CNNs for nuanced attribute evaluation, and utilizes Conditional GANs for synthetic data generation. Comparative evaluations show substantial enhancements in accuracy, precision, recall, speed, AUC, and specificity, establishing new benchmarks for wildlife detection in challenging conditions. This research advances automated wildlife monitoring, which is crucial for biodiversity conservation and addressing ecological challenges through integrated computational techniques.
Keywords:	Graph Neural Networks, Advanced Q Learning, Multi-Attribute Utility Theory, Generative Adversarial Networks
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	PREDICTING THE WEAVABILITY OF A NEW WOOLLEN FABRIC USING FUZZY LOGIC
Author:	M. EL BAKKALI , R. MESSNAOUI , M. ELKHAOUDI , O. CHERKAOUI , Pr ATIKA RIVENQ MENHAJ, Pr AZIZ SOULHI
Abstract:	Weaving saturation can lead to several undesirable problems, such as problems with loom performance, premature wear of mechanical parts and loss of expensive raw materials. Therefore, when designing and creating new woollen fabrics, it is crucial to adjust yarn densities and qualities according to the weaves to pass weavability tests. This study focuses on the development of a practical fuzzy logic model to predict the saturation of new 100% Wool fabrics. To validate this fuzzy model, an experimental part was carried out. The fabric samples used in this study came from three different weave types (plain, twill and satin) and included five weft counts (Nm) and nine different densities. The results obtained using the fuzzy logic model developed were compared with the experimental values. The predictions generated by the fuzzy logic model were found to be satisfactory and accurate, demonstrating its effectiveness for predicting the saturation of a new 100% wool fabric.
Keywords:	Saturation Index, Weaving, Fuzzy Logic, Modelling, Weavability.
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	ENHANCED LDA MODEL FOR SENTIMENT ANALYSIS (ELDASA)
Author:	ALURI SUGUNA RATNA PRIYA , A . MARY SOWJANYA
Abstract:	In the contemporary era, participative Internet communication or Social media platforms are widely used by people from all walks of life. Sharing opinions on various products and services has become common over social media. Unlike the conventional approach, opinions freely expressed over social media are goldmines to businesses. Analysing public sentiments has the potential to leverage business intelligence. Many researchers exploited Natural Language Processing (NLP) and Machine Learning (ML) to mine and ascertain opinions in online user-generated reviews. However, processing large text corpora is still challenging and prone to deteriorated performance. In this paper, we proposed a Generative Framework using an enhanced version of Latent Dirichlet Allocation Model considering sentiment polarities and latent aspects and we developed an algorithm named Enhanced Dirichlet Allocation Model for Sentiment Analysis (ELDASA) to realize the framework. This model is supported by a learning-based approach with ML toward the identification of sentiments and classifying them. Our empirical study using three social media datasets, consisting of reviews on hotels, music, and games, revealed that the proposed algorithm supports effective sentiment analysis.
Keywords:	Latent Dirichlet Allocation, Machine Learning, Sentiment Classification, Sentiment Analysis, Natural Language Processing.
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	NEUTROSOPHICAL MULTIPLE REGRESSION ENRICHED CHAOS DEEP BELIEF NETWORK FOR DYSLEXIA PREDICTION
Author:	VINODH M R , P.J. ARUL LEENA ROSE
Abstract:	Children with dyslexic can use the appropriate resources and specialised software to enhance their skills when problem is diagnosed early. Deep learning and machine learning techniques analyze dyslexia-related datasets from healthcare and educational sources, yet conventional models struggle with the inherent vagueness of dyslexia data, often represented in intervals. In this paper, generalized model of fuzzy and intuitionistic fuzzy known as neutrosophic multiple regression model is used for determining the degree of dependency among the independent and dependent variables of dyslexia dataset. In neutrosophic concept, each attribute is defined with the truthiness, falsity and indeterminacy membership, and are independent to each other. The correlation among the attributes are determined using neutrosophic least square error method. In existing deep neural networks using gradient based optimization ends up with the local minimum and results in early convergence. The proposed work deep belief network hyperparameters are scrutinized with chaos synchronization for classification. This work used two datasets, Dyslexic 12-4 dataset from Keel software repository and real time dataset collected from dyslexia schools of various districts are applied for performance comparison, the results proved that the proposed neutrosophic regression model produced highest rate of detection rate in dyslexia prediction compared with existing models. This study leverages advancements in AI to address the complex task of early dyslexia diagnosis. By employing neutrosophic multiple regression models alongside chaos deep belief networks, it achieves high prediction accuracy and handles data uncertainties effectively. This work significantly contributes to improve educational and healthcare outcomes for dyslexic children, underscoring the role of IT in learning disability interventions.
Keywords:	Dyslexia, Uncertainty, Chaos Theory, Neutrosophic Multiple Regression, Deep Belief Network
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	IMPROVING TEXT CLASSIFICATION IN FEDERATED LEARNING THROUGH TRANSFER LEARNING: A COMBINED APPROACH FOR ENHANCED CONVERGENCE AND PERSONALIZATION
Author:	Dr. MOHAMMED ABDUL WAJEED , Dr. ANNAVARAPU CHANDRA SEKHARA RAO , Dr. T.K. SHAIK SHAVALI , Dr. ABDUL RASOOL MD
Abstract:	Federated learning (FL) has become a prominent decentralized approach to machine learning, allowing collaborative model training across distributed clients without compromising sensitive data. However, the diversity of data among clients, especially in text classification tasks, often results in slower convergence and subpar performance. This study introduces a novel framework that combines transfer learning with federated learning to tackle the issues of slower convergence as the learning begins from scratch, due to which the parameters of the learning model are randomly initialized. By utilizing pre-trained language models, we propose a two-phase approach where clients first fine-tune a pre-trained model by taking consideration of the parameters already obtained, on their local text data before engaging in a federated averaging process. This integration facilitates quicker convergence, better generalization across varied client data, and improved model customization for individual clients. We assess our methodology on several text classification datasets with differing levels of data heterogeneity and demonstrate that our approach significantly enhances both overall accuracy and communication efficiency compared to conventional federated learning techniques. The findings highlight the potential of transfer learning in boosting the effectiveness and scalability of federated learning for real-world text classification applications.
Keywords:	Five: Federated Learning, Transfer Learning, Text Classification`, Federated Averaging, Heterogeneous data
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	PERFORMANCE ANALYSIS OF CONVENTIONAL AND DEEP LEARNING IMAGE FUSION METHODS
Author:	SURYA PRASADA RAO BORRA, RAJESH K PANAKALA, T.V. HYMA LAKSHMI, RAJA SEKHAR PITTALA, D.N.V. SATYA NARAYANA, GUNTI SURENDRA, A. GEETHA DEVI
Abstract:	This paper presents a comparative study of three different image fusing techniques such as hybrid image fusion designed using conventional image fusion technique, enhancement of low dose Computerized Tomography (CT) image and fusing with Magnetic Resonance Imaging (MRI) image, and Deep Learning (DL) based image-fusing technique. Firstly, it is observed that the hybrid method of image fusion outperforms single mode of image fusion such as Discrete Wavelet Transform (DWT) or Principal Component Analysis (PCA)-based image fusion. Secondly, it is worth noting that the performance of image fusion applied to enhanced low dose CT and MRI images are on par with fused normal dose CT & MRI images based on results obtained with the standard fusion quality metrics. Finally, the performance of Deep Learning (DL)-based image fusion developed using Convolution Neural Network (CNN) is evaluated. The Peak Signal to Noise Ratio (PSNR) of the DL based image fusion is more than 25 dB that of the conventional methods of image fusion. From that, it can be concluded that DL based image fusion technique will be the advanced form of medical image fusion to obtain a proper diagnosis of different diseases.
Keywords:	Discrete Wavelet Transform; Principal Component Analysis;Image Fusion; Deep Learning; Convolutional Neural Network, Multi Modal Images
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	ENHANCING SMART HOME SECURITY: A BLOCKCHAIN- INTEGRATED IOT APPLICATION WITH NFT AUTOMATION CONTRACTS
Author:	Mrs. SUDHA KAPUGANTI , Mr. SEETARAMANATH . M . N , Mr. V. KAMAKSHI PRASAD
Abstract:	With the ongoing growth of smart home technology, securing and managing IoT devices has become increasingly critical. This study introduces an innovative approach to smart home security by integrating blockchain technology with IoT devices, utilizing NFT-automated smart contracts to enhance security and management. The research aims to develop a solution that automates and strengthens the security protocols of smart home IoT devices while establishing a reliable framework for ownership management. By deploying a private blockchain, the system ensures data integrity, restricts access to authorized individuals, and reinforces device protection. NFT automation contracts further advance the system by creating a verifiable ownership structure, promoting secure and efficient device operations. Communication among IoT devices is optimized with the MQTT protocol, enabling efficient data transmission and consistent, rapid connectivity. Performance evaluations on real-time networks confirm the system’s usability, security, and operational efficiency, showcasing the value of blockchain and NFT technologies in enhancing the scalability and security of modern smart home systems. This research expands the potential of smart home technology, delivering a more secure, efficient, and scalable automation framework.
Keywords:	Message Queue Telemetry Transport (MQTT) Protocol, Blockchain Technology, Internet of Things, Smart Contracts, Home Automation, NFT-Automation.
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	AI BASED CLUSTER HEAD BASED MOBILE ADHOC NETWORK FOR PERFORMANCE IMPROVEMENT
Author:	S.HEMALATHA , S.SHALINI , U.SATHYA , P. KUMARAVEL , S VAMSEE KRISHNA , PRAMODKUMAR H KULKARNI
Abstract:	The transmission of packets in Mobile Ad Hoc Networks (MANETs) often experiences delays due to changes in internal parameter values caused by intruders or attackers. Additionally, power consumption in wireless networks remains a challenging issue, as the depletion of a node's internal battery can disrupt the entire communication system. To address these challenges, various methods have been proposed to enhance transmission time and optimize MANET battery management, enabling better evaluation of performance metrics. This article introduces novel Artificial Intelligence (AI)-based techniques for improving transmission time and power efficiency in MANETs using cluster heads. Clustering nodes are dynamically formed within a region based on factors such as high forward time, low delay, residual power, mobility, and node connectivity. These cluster nodes play a pivotal role in selecting optimal transmission paths, thereby enhancing communication efficiency and power optimization. The proposed Forward Parameter-Based Wireless Routing Protocol (FPWP) was integrated into the on-demand AODV routing protocol, resulting in the FPWP-AODV protocol. Comparative simulations with existing AODV protocols demonstrated that FPWP-AODV significantly outperforms its counterparts, achieving superior results in terms of transmission time and power management, making it an effective solution for MANET performance improvement.
Keywords:	MANET, Cluster Head, Transmission Time , Route Management, Life Time, Forward Time, Forwarded Packet
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	DYNAMIC TABEL ELIMINATION MODEL FOR SECURE NETWORK CONNECTIVITY IN WIRELESS SENSOR NETWORK ENVIRONMENT
Author:	DR. MYLA THYAGARAJU, DR. VENKATESWARLU CHANDU, A.CHARAN TEJA, DR.CH.SAHYAJA, ANKAM DHILLI BABU, NOURIEN MOHAMMAD, DR.CH.V.RAMA KRISHNA RAO
Abstract:	Wireless Sensor Networks (WSNs) consist of spatially distributed autonomous sensors that monitor physical or environmental conditions, such as temperature, humidity, or pressure, and cooperatively pass their data through the network to a main location. As WSNs are widely deployed in critical applications, including military, healthcare, and industrial automation, network security becomes paramount. Ensuring the confidentiality, integrity, and availability of data in WSNs is challenging due to their inherent constraints, such as limited processing power, memory, and energy resources. Key security issues include protecting against unauthorized access, data tampering, eavesdropping, and denial of service attacks. This paper proposed a novel Dynamic Address Table Removal (DATR). The proposed DATR model performs Weighted Dynamic Routing (WDR) for the computation of routing path in the network. Upon the estimation of the routing path in the network Table removal method, the address of each node is altered for the data transmission in the network scenario. The proposed model DATR employs the cryptographic process with the Table removal for the secure data transmission in the network for the efficient data transmission and reception of the data in the network. Simulation results demonstrated that the proposed DATR model achieves a significant throughput of 92.56% with an attack detection rate of 93%. Through extensive simulations, DATR demonstrates significant performance gains over traditional protocols such as AODV and DSR. It achieves up to 20% higher throughput, maintains a packet delivery ratio exceeding 97%, and reduces energy consumption by approximately 10-15% across varying network sizes. The simulation results demonstrated that the proposed DATR model achieves a higher attack detection rate with higher network throughput compared with the conventional technique.
Keywords:	Networking, Security, Confidentiality, Dynamic Routing, Table Removal, Attack Detection
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	AN APPLICATION OF CLUSTERING TECHNIQUE FOR SELECTING SNP MOLECULAR MARKERS IN RICE GENOME
Author:	LUXSANAN PLOYWATTANAWONG , SUCHA SMANCHAT , SISSADES TONGSIMA
Abstract:	A selection of Single Nucleotide Polymorphism (SNP) molecular markers whose unique genotypic combinations represent individual rice breeds is a critical consideration in rice breeding programs. Due to the complexity of SNP data with unknown target phenotypes, identifying trait-associated markers presents a significant challenge. Existing research has applied both supervised and unsupervised techniques to genetic and protein datasets; however, little effort has been directed toward SNP analysis. To mitigate the time and difficulties involved in exploring and identifying important markers for biologists, we employ a clustering technique for selecting significant SNPs from the rice genome. The experimental dataset comprises genome-wide SNPs from 88 rice breeds, each containing 50,172 SNPs. We propose an iterative application of the K-means clustering method to cluster these rice breeds into an increasing number of clusters. To identify potentially important SNP markers, the frequency with which each SNP is closest to the centroid of its group is counted. The SNPs are then ranked based on this frequency. The results demonstrate that the proposed method can distinguishes certain SNPs that are more frequently closest to the centroids, potentially indicating their importance as biomarkers. These SNPs among thousands can be recommended for further investigation for biologists in wet experiments.
Keywords:	Rice Genome; K-means Clustering; Molecular Markers; Single Nucleotide Polymorphism; Bioinformatics
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text

Title:	MOBILE ADHOC NETWORK INTRUDER NODE DETECTION AND PREVENTION FOR EFFICIENT PACKET TRANSFERING
Author:	S.HEMALATHA , TAVANAM VENKATA RAO , S. SHALINI , SHRUTHI S NAIR , SURYA LAKSHMI KANTHAM VINTI , DR.G.KRISHNA MOHAN
Abstract:	Wireless networks, particularly those without infrastructure, are vulnerable to security threats. Mobile Adhoc Networks (MANETs) are especially vulnerable to security breaches, with intruders constituting a substantial risk. These intruders seek to degrade network performance by sending duplicate packets to surrounding nodes, increasing the burden on these nodes and reducing overall network performance. Numerous research efforts are focused on detecting and avoiding such invasions. This article focuses on intruders in MANET communication, outlining their strategies and the negative impact on network performance. This paper describes a study on identifying intruders in MANET routing traffic using the Watch Dog Algorithm and a threshold-based categorization technique. The study's goal is to verify whether the identified nodes are invaders by running simulations with NS2.34 and evaluating the outcomes using important parameters including attack rate and detection time, PDR, and END. The suggested WDBIC model outperforms the standard AODV protocol in a variety of MANET performance metrics. Specifically, the WDBIC model has a greater attack rate, a slightly smaller percentage of normal nodes across different node counts, detects attackers faster, and consistently gives superior packet delivery ratios across various transmission parameters. Additionally, the WDBIC model lowers end-to-end latency by 6.2% to 43.4% when compared to the AODV protocol. These findings show that the WDBIC model outperforms the classic AODV protocol in MANETs in terms of efficiency, detecting the attack.
Keywords:	Manet , Intruder Node, Packet, Intruder Detection, Intruder Detection
Source:	Journal of Theoretical and Applied Information Technology 15^th December 2024 -- Vol. 102. No. 23-- 2024
Full Text