Volume 17, Issue 5, October 2026

Efficiency in Cloud Data Centers Via a Hybrid Methodology of VM Consolidation, Clustering and Power-State Switching
Pages: 1-10 (10) | [Full Text] PDF (720K)
Anam Ghaffar, Sana Yasin, Zunaira Altaf, Faiza Shafaqat, Fuhmida Suduf
Riphah International University, Faisalabad, Pakistan
Government College University, Faisalabad, Pakistan
University of Agriculture, Faisalabad, Pakistan

Abstract - Power consumption in data centers is a significant concern due to the ever-increasing demand for computing resources and the impact of high energy consumption. For the improvement and betterment in datacenters many algorithms are introduced. Researchers work for better and efficient data centers, but there is still space for the improvement of power consumption. Due to handling large amounts of data, data centers shift on virtual machines and big data concepts. This research focuses on power-saving techniques in data centers, specifically exploring the benefits of virtualization, sleep mode, and server clustering without compromising on its performance and quality of service (QOS). This is a hybrid technique in which virtual machines can share the load of physical servers due to which costs of cooling system for physical servers and their power become less then Server clustering that involves grouping multiple servers together to work collaboratively, allowing for workload distribution and improved resource utilization and then we apply sleep mode on clustered servers that shifts the load of a server to its neighbor’s server. The results demonstrate that the combination of virtualization, sleep mode, and server clustering can lead to significant power savings in data centers. We observe that the proposed technique shows significant advantages over the existing techniques so far.
 Index Terms - Virtual Machines (VMs), Cloud Datacenters, Quality of Service (QoS), Clustering and Sleep Mode
Comparative Evaluation of Operating Systems in Cloud Computing: Virtualization, Security, and Performance
Pages: 11-20 (10) | [Full Text] PDF (475K)
Lea Yue, Kung Yi
Department of Computer Science and Information Technology (CS&IT), Korea

Abstract - Operating systems play a fundamental role in modern computing environments by managing hardware resources, supporting application execution, and providing essential security mechanisms. With the rapid growth of cloud computing and virtualization technologies, operating systems have evolved to support virtualized resources, distributed computing, data sharing, and secure access to services over the Internet. This paper presents a comparative analysis of different operating systems with particular emphasis on their roles in cloud computing, virtualization, performance, and security. The study examines major virtualization approaches and evaluates how operating systems manage virtual machines, processing resources, memory, storage, and network services in cloud environments. In addition, the paper investigates security challenges affecting computer and mobile operating systems, including unauthorized access, malware, data breaches, privilege escalation, and other threats that can compromise system processes and resources. Various security mechanisms and potential solutions are also discussed to strengthen operating-system security in virtualized and cloud-based environments. Furthermore, the study explores the relationship between cloud computing and virtualization, highlighting how virtualized resource pools enable flexible processing, storage, and resource sharing through network-based services. The comparative analysis provides an overview of the capabilities, limitations, virtualization support, and security considerations of different operating systems and identifies key factors for developing secure and efficient cloud computing environments.
 Index Terms - Comparative Analysis, Operating Systems, Virtualization, Security and Performance
Road Features Extraction from Aerial Images Using Deep Learning
Pages: 21-29 (9) | [Full Text] PDF (719K)
Waseem Akram, Aasim Ghaffar, Anjum Ali, Muhammad Jehanzeb, Muhammad Rizwan Amin, Awais Rasool
Department of Software Engineering, The University of Lahore, Pakistan
Department of Physics, Astronomy and Mathematics, University of Hertfordshire
Department of Software Engineering, Riphah International University Faisalabad, Pakistan

Abstract - Roads are very important for transportation. Now, there is a dire need for urban development due to the increase of world population. Roads are the main part of the transportation system. Therefore, in an emergency case the extraction of roads is the main thing. In the past many methods of research were used to find the roads. An advanced method that researchers use in our research has better results than past methods. Therefore, there are many issues faced with different road management. In some cases, there is research on some roads that research is very difficult to extract from the satellite or aerial image in past. Road extraction from the aerial image is used for smart city and road monitoring. This is also used for emergency response, in the difficult areas where roads are not research prepared, and this technique gives us extraction of alternative roads. That is also used for traffic management, low cost, less time-consuming and more accuracy. Two main phases are used firstly; the use of deep learning advanced model that can extract road more accurately than others can methods that are used for the predicting of road segments. Secondly, research applies some advance-thinning algorithm to find the road centerlines from the complex urban areas. That is Automated Road extraction from satellite or aerial image provide the very efficient and accurate results in any urban and rural areas. The purpose method provides good percentage results in any type of road. That extract road with low cost and less time consuming. In this research road extraction from a arial image accuracy is 98%, Precision 92%, Recall 94%, F-Score is 96%.
 Index Terms - Road Extraction from Aerial Images, Deep Learning, Satellite Image Analysis, Remote Sensing and Image Segmentation
Recent Advances in Malware Detection Techniques for IoT Systems: A Systematic Review
Pages: 30-40 (11) | [Full Text] PDF (676K)
Muhammad Sajjad Ur Rehman, Muhammad Junaid Arshad
Department of Computer Science, University of Engineering and Technology, Lahore

Abstract - The Internet of Things (IoT) has grown tremendously in the past decade, with billions of heterogeneous resource-constrained devices being deployed in fields such as smart homes, healthcare, industrial automation, and critical infrastructure. This massive growth has also resulted in a tremendous attack surface going largely unsupervised and unmonitored, making IoT ecosystems a desirable target for malware operators. Traditional security approaches, especially signature based Anti-Virus engines, are not designed for and suitable to the limited hardware and diverse architecture commonly found in IoT deployments and were never intended to detect or prevent new and unknown “zero day” threats. This paper aims to systematically review the IoT environment proposed malware detection techniques from 2015 to 2025. We share the literature search and screening process to obtain 38 primary studies, provide a taxonomy of detection paradigms, and perform a comparative analysis of signature-based, anomaly-based, classical machine learning, deep learning, and federated learning approaches. The review also analyzes publicly available benchmark datasets and presents typical performance results in the form of radar charts and bar charts, and discusses the main open challenges, such as limited resources, biases in the datasets, adversarial evasion, concept drift, and privacy, as well as providing some detailed lines for future research. The paper aims to be a comprehensive and evidence-based guidebook for researchers and practitioners working on malware detection for IoT deployments.
 Index Terms - Internet of Things, IoT Security, Malware Detection, Intrusion Detection Systems, Machine Learning, Deep Learning, Federated Learning, Anomaly Detection, Botnet and Network Traffic Analysis
D-KAP: A Transfer-Learning and Handcrafted Features Framework for Mango Leaves Disease Classification and Detection
Pages: 41-48 (8) | [Full Text] PDF (694K)
Malka Shahid, Fiza Noreen, Fuhmida Suduf, Iram Shahzadi, Tayyba Anwar
National Textile University, Faisalabad, Pakistan
Riphah International University, Faisalabad, Pakistan
Govt. College University, Faisalabad, Pakistan
University of Agriculture, Faisalabad, Pakistan

Abstract - Agriculture is an important sector in the world economy but plant diseases pose a serious constraint on productivity and quality of crop plants. Mangoes are an important crop in Pakistan, and diseases such as Mango Scab, Mango Rot, and Alternaria leaf blotch can impact yield. Manual inspection by the farmers is labor intensive, time consuming, and has a high risk of errors, and automated solutions are needed. This research presents the D-KAP framework, an automated system designed for the early detection and classification of mango leaf diseases using deep learning. A set of 419 high-resolution images was specifically collected from the Uch Sharif district of Bahawalpur and included healthy as well as diseased leaves at different environmental conditions. The study adopts the transfer learning using the Convolutional Neural Network ( CNN) architectures ResNet-50 and Xception. Image preprocessing is used to improve feature visibility. The experimental results show that the ResNet-50 model had the highest classification accuracy of 80% compared to Xception model with a classification accuracy of 78.1%. “The framework also investigates handcrafted feature extraction techniques, including Speeded-Up Robust Features (SURF), Histogram of Oriented Gradients (HOG), and Local Binary Pattern (LBP), and evaluates their performance under increasing dataset complexity. The proposed framework offers a comprehensive and effective solution for farmers, allowing them to manage various diseases effectively and potentially boost their crop yields.
 Index Terms - Mango Disease, Deep Learning, CNN, ResNet-50, Xception and Image Processing
Hyperspectral Imaging and Machine Learning for Detection and Classification of Major Fungal Diseases in Rice
Pages: 49-55 (7) | [Full Text] PDF (682K)
Zahid Iqbal, Sadia Hameed, Anam Ghaffar, Azka Aziz, Humaira Sadaf, Rida Noor
University of Suffolk, Ipswich, United Kingdom
Department of Computer Science & IT, The University of Lahore, Pakistan
Department of Computer Science, Riphah International University Faisalabad, Pakistan
Department of Software Engineering, The University of Lahore, Pakistan
Department of Software Engineering, Government College University Faisalabad, Pakistan

Abstract - Rice production is severely affected by fungal diseases, making early and accurate disease identification important for timely disease management. Traditional identification mainly relies on visual symptoms and expert knowledge, which can be time-consuming, subjective, and difficult to apply at large scale. This study investigates hyperspectral imaging (HSI) combined with machine learning for pixel-level detection and classification of major fungal diseases in rice. Unlike studies that focus on a single rice disease or a limited number of disease categories, the present study comparatively evaluates Artificial Neural Network, Support Vector Machine and Random Forest using hyperspectral spectral data to distinguish three major fungal diseases brown spot, blast, and sheath blight while including healthy rice tissue as a separate control class. A total of 150 diseased leaf samples were collected, with 50 samples for each disease. Pixel-wise data extracted from the hyperspectral images comprised 1472 control pixels, 669 blast pixels, 693 brown spot pixels and 1169 sheath blight pixels. The hyperspectral data represent spectral responses of healthy and disease-affected rice tissues rather than direct spectral signatures of the pathogens. The HSI data were pre-processed to reduce noise and improve spectral reliability before classification. The experimental results showed that ANN achieved the highest classification accuracy of 98%, followed by SVM with 94% and RF with 88%. Overall, the results demonstrate that combining HSI with comparative machine learning classification can provide a rapid, non-destructive, and reliable approach for distinguishing major fungal diseases in rice and healthy tissue, with potential applications in precision agriculture. Overall, the combination of hyperspectral imaging and machine learning provides a rapid, non-destructive, cost-effective and reliable approach for detecting and classifying fungal diseases in rice. The proposed approach can support large-scale crop monitoring, facilitate early disease diagnosis, and assist farmers and agricultural experts in taking timely control measures. This technology has considerable potential for developing intelligent and automated disease management systems for precision agriculture.
 Index Terms - Rice Fungal Diseases, Hyperspectral Imaging, Artificial Neural Network, Support Vector Machine, Random Forest, Machine Learning, Disease Classification and Precision Agriculture

IJCST Managemet

Current Issue



     
           ISSN 2047-3338 (Online)                   

Recent Issues

Author Guidelines

IJCST RSS Feeds

News & Updates



   Google Scholar