Lahore Garrison University Research Journal of Computer Science and Information Technology
Not a member yet
227 research outputs found
Sort by
Security Assessment in Software Defined Networks (SDN): Vulnerabilities, Challenges and Research Prospects
Full text linkConventional internet architecture facilitating users for different services and applications but facing number of challenges like network management, QoS management for network virtualization, IP multicasting, deployment of IPV6, crucial security measures, end to end connectivity, inter and inter domain routing. To meet the demand of these services due to rapid growth of technologies and traffic, an emerging network architecture termed as Software Defined Network (SDN) with programmable technology has brought unprecedented management to control networks. Due to separation of data, control and application planes software defined network provides cost effective, openness, centralized automation, programmable features as per users own demands and high resilience to network administrators. OpenFlow evolved as a first standard protocol for software defined network control and data planes communication to meet changing business requirements. Although, Software Defined Network brings enormous advancements in networks to support business applications but it is severely affected with cyber-attacks at data, control and application planes. Middle boxes plays a significant role to manage network effectiveness and provide adequate security control from external and internal security threats but require proper management and configuration otherwise it leads to devastating effect. The main contribution of this paper are: (1) It explores various potential security attacks at SDN layers and inconsistent policies, (2) It provides various security concerns in SDN planes and preventive measures against these prevailing attacks, (3) It discusses security threats challenges and research opportunities in software defined network keeping in view critical security controls like spam detector, IDS/IPS, Firewall and policy management. This article assists researchers to comprehend security concerns and state of the art development and challenges explored by scientific community in SDN
A Machine Learning (ML) based Forecasting Model for Covid-19 Patients
No full textThis research paper introduces a Machine Learning (ML) based forecasting model for COVID-19 cases, to investigate the performance of the Machine Learning algorithms and develop a new procedure to improve prediction efficiency. Utilizing the Multilayer Perceptron (MLP), Linear Regression (LR), K-Nearest Neighbours (KNN), Support Vector Machines (SVM), and the proposed approach, the study considers the COVID-19 data to forecast case numbers. The study contextualizes its findings through a systematic methodology of dataset compilation, algorithm interpretation, and framework development. It uses measures such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) to evaluate the predictive performance. The trend in COVID-19 data and predictions from different algorithms is shown through graphical illustrations. The same goes for the proposed framework predictions. This study demonstrates that the proposed LR approach and the framework outperform previous MLP, KNN, and SVM models, suggesting the relevance of explainable and robust modeling solutions for COVID-19 risk assessment. Our suggested framework, which particularly outperforms individual algorithms by averaging out their results after combining them, reduces prediction errors significantly. Discussion of implications of forecasts for interventions, resource allocation, and policy decisions is done, with the need for accurate forecasting in the pandemic response highlighted. Further research works can be expected to improve the framework, incorporate new machine learning (ML) techniques, and deploy real-time adaptive modeling systems. Collaboration between researchers, policymakers, and healthcare practitioners is crucial for the adoption of research results into practice and the promotion of evidence-based decision-making in the context of outbreak response and preparedness. Generally, this study shows the progress of the field of epidemiological forecasting as well as provides information that is important in fighting COVID-19 and future epidemics
ADVANCED IMAGING TECHNIQUES IN CT SCANS FOR EARLY LUNG CANCER DETECTION USING NEURAL NETWORKS: A STUDY OF PAKISTAN’S HEALTHCARE SECTOR
Full text linkDue to multiple healthcare system limitations in Pakistani medicare, lung cancer continues to represent one of the biggest health threats of this era despite various image processing methods proving insufficient for accurate diagnosis. An unreliable clinical approach exists in early diagnosis because existing techniques yield excessively misleading positive and negative results. Conventional diagnosis tools mainly use two-dimensional imaging techniques that both set limitations on observing complete tumor dimensions and lead to misidentifying cancer characteristics. The need for advanced detection techniques exists because current methods require improvement to detect tumors precisely at an early stage. Early detection of lung cancer remains vital to achieve better treatment results along with improved survival rates because lung cancer affects a large number of people worldwide. The discovery of Computed Tomography (CT) scans through technological advances made them essential for performing early-stage lung cancer detection. The research evaluates the acceptance level along with the diagnostic effectiveness of state-of-the-art imaging methods in CT scans used to discover lung cancer at early stages in Pakistan's medical space. Late-stage diagnosis of lung cancer leads to minimal treatment success because of its low survival rates. The proposed research method enables the detection of cancer cells at an early stage. A 3D convolutional neural network applies its analysis to the LIDIC-IDRI dataset to enhance diagnostic precision. The created model reached an accuracy level of 98.22%. A 3D convolutional neural network (CNN) analysis model is introduced in this research to raise the precision of lung cancer diagnosis detections. An implemented 3D CNN model examines volumetric CT data during analysis to provide superior malignant spot detection when compared to traditional 2D imaging systems. This method makes diagnoses more reliable through the LIDC-IDRI dataset while minimizing the wrong classification of patients. Additionally, the study will compare the use of CT scan technology in Pakistan with other countries, offering insights into how different healthcare systems have integrated these technologies
Countering DDoS threats: leveraging ensemble methods for detection and mitigation
Full text linkCloud computing is the modern concept of distributing numerous services through the Internet, such as web applications, databases, and programmers that operate on several servers. As Cloud computing technologies evolve, increasing susceptibility to attack may result from service outages during data storage and transmission. The most common sort of assault against Cloud settings is distributed denial-of-service (DDoS). Several approaches for detecting and mitigating these attacks have been offered, however they are ineffective. In this research, we propose a method for detecting and mitigating DDoS attacks in their early phases, considering top-layer advances at the application layer and the TCP handshake mechanism. This study employs a variety of ensemble-based machine learning approaches to classify incoming data as legitimate or malicious to respond to DDoS attacks at the application layer. Furthermore, the double TCP connection concept is used to prevent DDoS. Experiments show that the stacked voting system detects DDOS attacks with the best F-score of 99.9%
Guardians of the iot realm: a comparative analysis of cryptographic security solutions for bolstering iot device networks
Full text linkThe "Internet of Things" (IoT) is an emerging technology that allows electronic devices and sensors to connect with one another over the Internet. The IoT is a network of smart devices such as sensors that are connected through cables or wirelessly. Today, IoT devices are commonly used in medical science, industrial automation, and smart home automation, among other applications. Easy accessibility and the open-source environment of IoT devices are major threats to privacy and security. In this paper, we comprehensively discuss areas of IoT, including architecture, details of IoT layers with respect to security algorithms, protocols, attacks, and their mitigation. The primary purpose of this research is to provide an efficient and cryptographically proven algorithm scheme with appropriate hardware to improve IoT device network security. The presented research enhances the network security of IoT devices by presenting an algorithmic framework fortified by cryptography and supported by the finest hardware options. After evaluating several cryptographic algorithms and hardware options, a more secure solution for IoT infrastructure is proposed. The article in particular, proposes a detailed strategy that addresses security at every level of the IoT system. This research is an important step towards ensuring the integrity and security of IoT devices in a networked environment, with the primary goal of increasing the protection of sensitive data and interactions
UNRAVELING RANSOMWARE IN THE DIGITAL BATTLEFIELD: THREAT ANALYSIS AND COUNTERMEASURES
Full text linkRansomware is a well-known form of malware known for causing severe and permanent damage to its targets. Timely identification of such attacks is important to mitigate the consequences of these attacks. According to Data Breach Investigation Report (DBIR), since 2021, ransomware attacks have grown 17% yearly. It is widely considered a major cybersecurity threat at individual and organizational levels. There are several techniques that organizations can use to manage ransomware, such as backup, network segmentation, HR education, endpoint protection, and advanced threat hunting. It’s worth noting that only some techniques are foolproof, and a comprehensive defense strategy often involves combining multiple techniques. Ransomware has been used in the context of the Russia-Ukraine war, primarily by Russian-backed cybercriminal groups. These groups have targeted Ukrainian infrastructure and businesses with ransomware attacks, encrypting their victims’ data and demanding payment to unlock the data. These attacks have caused significant disruptions and financial losses for the targeted organizations. The paper aims to study the ransomware technique and summarize the most prominent threat actors involved in the war. We have chosen one of the well-known malwares,” HermeticRansom”, performed its thorough analysis and created a Yara rule for its detection
Plant leaf vein and outline feature extraction using fractal and computer vision approaches
Full text linkThe study focuses on utilizing plant leaf characteristics for plant identification and disease detection. Leaves are pivotal for gathering information about plants. Leveraging computer vision and smart agricultural technologies, the proposed model discerns venation and texture features in various plant leaves. This research utilized a modified dataset derived from the Flavia leaf image dataset, comprising images of 32 different plant species. The dataset was divided into two subsets (one with 1907 images and another with 1000 images) to differentiate between tuned and untuned image processing. Techniques such as GLCM, LBP, Gabor filters, Fractal Dimension, and box counting were employed to extract leaf texture features, including venation patterns. The study conducted four experiments with training and testing splits of 70/30 and 80/20. A novel method combining SVM with fractal dimension analysis was benchmarked against six classifiers (Random Forest, KNN, DNN, Naïve Bayes, Decision Tree, and SVM), achieving an impressive accuracy of 88% and a Fractal Dimension of 1.8709. This research holds significant potential for advancing digital and modern agriculture, particularly in the early detection of plant diseases and accurate plant identification
Factors Affecting Acceptance of Health Information Systems in Pakistan
No full textFor decades, information technology's (IT) acceptance has continued to be a source of contention. Technology Acceptance Models, such as TAM, are utilized to forecast user behaviour. There is little data on the predictive capacity of health information system consumers' approval (HIS). The research focuses on healthcare professionals' Perceived Ease of Use (PEOU) and Perceived Usefulness (PU). All Factors are positive and significant relationships with each other. The first stage is to determine the level of adoption, followed by stakeholders' desire to develop the HIS to improve healthcare services in Pakistan
Cotton Leaf Disease Classification
No full textThe cotton industry is a significant agricultural sector that has a profound impact on the nation's economy. To gauge a nation's economic performance, it is crucial to examine both the quality and quantity of its agricultural production. Early diagnosis of leaf diseases may lead to higher revenues in manufacturing. Various image-processing techniques have been developed throughout the years to identify illnesses that damage leaves. Advancements in technology are accelerating the process, although it is still in its initial phases. The agriculture business has a major challenge in dealing with the rise of leaf diseases. Many diseases, such as powdery mildew, army worm, bacterial blight, target spot, and aphids, may affect cotton plants. Extensive observations may be time-consuming, expensive, and sometimes inaccurate for the producers involved. We suggest utilizing machine learning techniques like Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Random Forest (RF), Logistic Regression, and Convolutional Neural Network (CNN) to automatically detect diseases on cotton leaves. This method is designed for the agricultural sector to distinguish between healthy and sick leaves. The study found that CNN performs best in image classification as it has the greatest accuracy percentage of 99.1%
The IMPROVING CLASSIFICATION ACCURACY ON BREAST HISTOPATHOLOGY IMAGES DATASET USING TRANSFER LEARNING
No full textAt the present time, one of the main causes of mortality for women is Breast Cancer (BC). The pathologist still faces several difficulties in accurately diagnosing cancer. Invasive Ductal Carcinoma (IDC), the common kind of BC, has been categorized in this study. Numerous innovative strategies have been used in the realm of medical research for the categorization of IDC. However, there are a number of issues with the BC classification approach, including vanishing gradient, class imbalance, data overfitting, low accuracy rate, and latency to discover cancer cells in patients. Therefore, creating a precise and well-structured method for IDC categorization is essential. In order to address these issues, a productive technique has been put out, in which the classification model is specified as the TransResCNN model, or transfer learning applied to the CNN model of the pre-trained residual network (ResNet). The most widely used techniques for handling large datasets are transfer learning and data augmentation. In order to assess the model's performance, an image-based confusion matrix is used to classify IDC. A number of assessment criteria, including recall, F1-score, accuracy, and precision, have also been used. Upon comparing our suggested research with other current studies, it was found to have the greatest accuracy (91.66%) and F1-score (94.22%). The examined study demonstrates that, in comparison to earlier research investigations, our suggested technique produced better results