International Journal of Scientific Research in Network Security and Communication
Not a member yet
272 research outputs found
Sort by
S-RMCSA: A Security-Aware Routing and Spectrum Assignment Framework with QKD and Physical-Layer Encryption in SDM-EONs
Get PDFWe present S-RMCSA, a new security-aware routing and spectrum assignment framework for space-division multiplexing elastic optical networks (SDM-EONs). Unlike traditional approaches that focus only on blocking and crosstalk, S-RMCSA integrates quantum key distribution (QKD) and physical-layer encryption directly into network resource allocation. Each traffic demand is assigned a security profile that decides whether a dedicated QKD lane is placed on an isolated fiber core and whether chaos masking or constellation twisting is applied for added protection. A multi-objective cost function balances crosstalk, fragmentation, encryption overhead, and survivability. Simulation results on the NSFNET with 7-core MCF links show that S-RMCSA lowers blocking probability by up to 29%, improves crosstalk margin by over 5 dB, and maintains QKD key rates above 10 kbps over 500 km. Importantly, it secures more than 97% of critical demands with minimal spectral overhead, enabling confidential and resilient next-generation optical networks
Real-Time Intrusion Detection in Controller Area Networks: An Evaluation of Current Methods and Future Directions
Get PDFController Area Networks (CANs) are critical components of modern vehicles and industrial systems, facilitating communication between various electronic control units. However, the widespread connectivity and lack of inherent security measures make CANs vulnerable to cyber-attacks. Intrusion detection systems (IDS) safeguard CANs by detecting and mitigating potential attacks. This paper presents a comprehensive analysis of current methods for the real-time detection of attacks in CANs. The IDSs based on different input data modalities are evaluated based on their effectiveness, accuracy, and efficiency. The analysis highlights the strengths and limitations of each method, providing valuable insights for researchers and practitioners in developing robust and reliable intrusion detection systems for CANs. The findings suggest that the lightweight strategy in IDS is widely accepted for real-time application due to its computational simplicity and model structure. Furthermore, the paper identifies future directions to enhance the security of CANs and ensure their resilience against evolving threats
Empowering Students: Building an Integrated Application for Enhanced Productivity, Efficiency and Creativity
Get PDFIn the realm of education, students often encounter a myriad of challenges when it comes to managing their academic tasks efficiently and expressing their creativity effectively. This research paper delves into developing and implementing an integrated application designed specifically for students to streamline their workflows, enhance productivity, and foster creativity. By examining the features, functionalities, and potential impact of this application, we explore its capacity to address the diverse needs of students and revolutionize their academic experiences. By thoroughly examining and presenting empirical data, this study highlights how technology can revolutionize the future of education.
 
Design Of Microstrip Patch Antenna For Sixth Generation Frequency Band
Get PDFThe rapid advancement of technology and our growing reliance on it in our daily lives have led to a growing demand from users for higher data transfer speeds. The This work presents the design of rectangular microstrip single patch antennas as well as 1x2, 2x2, and 1x4 array patch antennas for 6G applications that operate in the 100 GHz–300 GHz frequency range. A rectangular patch antenna with copper conductivity material developed using Durod5880 unique substrate materials serves as the centrepiece of this arrangement. The dimensions are calculated using mathematical formulas and software, and this design was optimized and simulated using CST Studio Suite. Therefore, the assessed characteristics—return loss, bandwidth, gain, directivity, sidelobe magnitude, angular breadth (3 dB), input impedance, radiation efficiency, and VSWR—give adequate performance for the specified antennas. The array antenna is used to optimize the designated antenna in order to attain optimal performance.
 
EDeLeaR: Edge-based Deep Learning with Resource Awareness for Efficient Model Training and Inference for IoT and Edge Devices
Get PDFDeep learning has emerged as a powerful technique for processing and extracting insights from complex data. However, the resource-constrained nature of edge devices poses significant challenges to the deployment of deep learning models at the network edge. This research proposes a novel algorithm called EDeLeaR, which stands for Edge-based Deep Learning with Resource-awareness, to enable efficient model training and inference in edge computing environments. EDeLeaR leverages adaptive resource allocation and optimization techniques to maximize the utilization of limited computational resources while preserving model accuracy and minimizing latency. This paper presents the design, implementation, and evaluation of EDeLeaR, showcasing its effectiveness through comprehensive experiments on real-world edge devices.
 
Credit Card Fraud Identification Using Calibrated K nearest Neighbor
Get PDFCredit card fraud has become a significant concern in today`s digital world, leading to substantial financial losses for individuals and businesses alike. Detecting fraudulent transactions accurately and efficiently is crucial for maintaining the security of financial systems. The proposed method combines the power of KNN, a popular classification algorithm, with calibration techniques to enhance the fraud identification performance. Calibration is employed to adjust the probabilities assigned by the KNN algorithm, allowing for more accurate classification decisions and better control over the false positive rate. To evaluate the effectiveness of the proposed approach, comprehensive experiments are conducted on a benchmark credit card fraud dataset. The results demonstrate that the calibrated KNN method outperforms the traditional KNN classifier in terms of both accuracy and other performance parameters. The calibrated KNN approach achieves higher fraud detection rates and produces well-calibrated probability estimates, reducing the risk of false alarms or missed fraud cases. This research contributes to the advancement of credit card fraud detection systems and provides valuable insights for financial institutions and individuals concerned with safeguarding against fraudulent activities
Copy-Move Image Forgery Detection Using CNN and SIFT Algorithm
Get PDFThe widespread use of digital image alteration emphasizes how urgently reliable detection methods are needed to protect the originality and integrity of visual content. In this paper, we present an original technique for copy-move image forgery detection that combines Scale-Invariant Feature Transform (SIFT) with Convolutional Neural Networks . Our approach uses SIFT for reliable key-point descriptor extraction and Error Level Analysis (ELA) preprocessing to improve potentially changed regions. In parallel, a CNN model is trained using characteristics extracted from ELA representations to distinguish between modified and unmanipulated images. Although our hybrid methodology shows promising results in identifying copy-move forgeries, it is important to recognize the limits of current methods and systems.These drawbacks include the inability to grasp the results, scalability problems, dependency on handcrafted characteristics, computational complexity, limited generalization, partial copy-move vulnerabilities, and lack of interpretability. Our suggested method`s incorporation of SIFT is essential for identifying forgeries, especially in situations where copy-move manipulation is involved. By offering robust and unique descriptors that are independent of scale, rotation, and translation, SIFT features provide precise recognition of replicated areas in a picture. This method improves the model`s capacity to identify minute changes and visualise the location of forgery by utilizing SIFT in conjunction with CNN. This helps to maintain the visual authenticity and reliability of digital content.
 
Diabetes Predictor: Prediction Using Machine Learning Techniques
Get PDFAs the old saying goes, prevention is better than cure when it comes to health. The likelihood of saving lives can be greatly increased by anticipating diseases such as diabetes. Numerous variables, including age, obesity, lack of exercise, genetic predisposition, lifestyle, nutrition, and high blood pressure, can contribute to diabetes, an illness that is spreading quickly. With the help of machine learning techniques (MLT), healthcare professionals can now forecast patient outcomes using pre-existing data, which makes them indispensable tools. Several categorization machine learning methods are used in a diabetes prediction project to identify the most accurate model. This model takes into account extrinsic factors linked to diabetes risk in addition to conventional components like insulin, age, BMI, and glucose. Comprehending the natural glucose regulating process of the body is essential to understanding diabetes. The body uses glucose, which is obtained from foods high in carbohydrates, as its main energy source. The pancreas secretes insulin, which makes glucose easier for cells to use as fuel. On the other hand, diabetes is brought on by inadequate insulin synthesis or inadequate insulin use, which raise blood glucose levels. Here, skin thickness, number of conceptions, and pedigree function are additional characteristics that improve the model`s prediction power. These factors enhance the accuracy of diabetes risk assessment by adding to conventional markers and providing insightful information. Proactive illness prediction is made possible by utilizing MLT in the healthcare industry, especially for conditions like diabetes. The predicted accuracy of diabetes models can be greatly increased by incorporating both traditional and non-conventional risk indicators, such as skin thickness, number of pregnancies, and pedigree function. This will enable early intervention and better patient outcomes.
 
The Neural Frontier: AI`s Relentless Encroachment into the Human Mind
Get PDFThe intersection of neuroscience and artificial intelligence (AI) represents a turning point in technological progress, presenting previously unheard-of chances to better comprehend and improve human cognition while also posing significant ethical issues. The ethical implications of AI-driven neuro-technologies are examined in this work, "The Neural Frontier: AI`s Relentless Encroachment into the Human Mind," through a tripartite technique that includes a systematic literature review, an interdisciplinary ethical analysis, and a speculative risk assessment. Our findings highlight the importance of brain privacy and cognitive liberty as fundamental rights, highlighting the need for strong governance frameworks and data security measures. The possibility of AI systems perpetuating biases in neuroscientific applications, the potential for unequal access to cognitive augmentation to exacerbate societal disparities, and the blurring of moral duty as AI influences human cognition are among the major ethical problems that we uncover. The study makes the case for the creation of frameworks for equitable cognitive augmentation, neuro-ethically-aligned AI, and adaptive governance models. It highlights how important it is to collaborate across disciplines, involve the public, and incorporate neuro-ethics into scientific education. We are not just developing technology but also redefining the limits of human identity and awareness as we traverse this cerebral frontier, striking a balance between the enormous potential of AI in neuroscience and the necessity to protect human dignity, autonomy, and cognitive liberty. In an era of unparalleled technological advancement, this undertaking necessitates moral discernment, scientific probity, and a dedication to safeguarding the fundamental principles that characterize our humanity.
 
A Cloud-Based Machine Learning Approach for Blood Cell Classification using YOLOv5
Get PDFChecking blood cell counts is crucial for diagnosing health issues. Traditionally, this involves manually counting cells under a microscope, a slow and tiring process. This research explores a new method using machine learning. A machine learning approach for automatic identification and counting of three types of blood cells using ‘you only look once’ (YOLO) object detection and classification algorithm. YOLO framework has been trained with a modified configuration BCCD Dataset of blood smear image to automatically identify and count red blood cells, white blood cells, and platelets. Moreover, this study with other convolutional neural network architectures considering architecture complexity, reported accuracy, and running time with this framework and compare the accuracy of the models for blood cells detection. Overall, the computer-aided system of detection and counting enables us to count blood cells from smear images in less than a second, which is useful for practical applications. Among the state-of-the-arts object detection algorithms such as regions with convolutional neural network (R-CNN), you only look once (YOLO), we chose YOLO framework which is about three times faster than Faster R-CNN with VGG-16 architecture. YOLO uses a single neural network to predict bounding boxes and class probabilities directly from the full image in one evaluation. We retrained YOLO framework to automatically identify and count RBCs, WBCs, and platelets from blood smear images. Also, the trained model has been tested with images from another dataset to observe the precision and accuracy to be around 95% with the recall-confidence to be 0.99.