International Journal of Communication Networks and Information Security (IJCNIS)
Not a member yet
1021 research outputs found
Sort by
DESIGN AND IMPLEMENTATION OF MACHINE LEARNING METHODOLOGIES FOR DDOS ATTACK DETECTION AND CLOUD SECURITY
No full textCloud computing is a model for servicing software and hardware through the internet. Sincerely implementing the principle of cloud computing, users can operate the data and apps using any device. It is impossible to imagine an individual who has access to the web though it is possible to live without a certain gadget. The benefits of cloud computing include the issues of scalability, virtualization, accessibility of the users, low cost of infrastructure, and flexibility. Its drawback is that it can be blocked or attacked with a distributed denial of service attack. Distributed denial of service attacks are done by several computers simultaneously to launch an attack on a specific resource, website, or server. The result of which is a denial of service to end users. Frequent connection requests heard with false ones, and an unusual amount of messages and twisted packets make the system slow or shut it down. This cannot allow real people and the services they call for to be of service since it shuts them out. BC Online computing This article focuses on the use of the machine learning algorithms to detect distributed denial of service (DDoS) attacks. There are two conducted techniques in this research and the datasets used are from the NSL-KDD. On one side we find the Learning Vector Quantization (LVQ) filter, while on the other side there is the Principal Component Analysis (PCA) that it is a dimensionality reduction technique. Concerning the detection of DDoS attacks, the characteristics that were chosen from each of the mentioned approaches was categorized using Decision Tree (DT), the Naïve Bayes NB), and the Support Vector Machine (SVM). Thus, we contrasted the consequences of different categorizations. Thus, LVQ-based DT outperforms other types of DT in the case of attack identification
Predicting the Impact of Organizational Learning, Ethical Leadership, Government Support, and Corporate Governance on the Implementation of Sustainable Corporate Social Innovation: Evidence from Small and Medium Enterprises (Smes) in Malaysia
No full textSustainable corporate social innovation is important in encouraging organisations to create innovative goods and services that address social, environmental and business benefits. Sustainable corporate social innovation has prominently been debated in both academia and the public recently. However, sustainable corporate social innovation in Malaysia is new and has no clear establishment. It is important to understand the factors that improve a company’s social innovation practice as it improves the well-being of the community and is business oriented. This study proposes the model of the relationship of organisational learning, ethical leadership, government support, and corporate governance on sustainable corporate social innovation in small and medium-sized enterprises (SMEs) in Malaysia. Organisational learning, ethical leadership, government support, and corporate governance plays a strategic tool and have been proposed in the field of modern management for social innovation in stabilising organisational success. The proposed model can serve as a valuable tool for researchers exploring the factors influencing OL, EL, GS, CGI and CSI within SMEs, as well as for practitioners seeking to benchmark their SMEs practices
Determining Real-Time Emotion Using Convolutional Neural Networks in Machine Learning
No full textThis cutting-edge research project focuses on the development of a real-time emotion recognition system leveraging the power of Convolutional Neural Networks (CNNs) in the field of machine learning. Emotions play a pivotal role in human communication and behavior, and this project aims to create a system that can accurately and swiftly detect emotions from various sources, such as images, videos, or audio data. The recognition of human emotions is a captivating and multidisciplinary field that explores the intricate tapestry of human feelings, encompassing psychology, neuroscience, computer science, and more. Emotions, being integral to our existence, influence our thoughts, actions, and relationships. This pursuit involves deciphering emotional cues from various sources, including facial expressions, vocal intonations, physiological responses, and textual content. The identification of human emotions has been extensively studied, but there are no effective techniques that can identify them correctly. In order to overcome this problem, the proposed method implements a real-time emotion recognition system by using machine learning. Convolutional neural networks (CNN) are used for both gender and emotion classification
Leveraging Machine Learning and Data Engineering for Enhanced Decision-Making in Enterprise Solutions
No full textThis comprehensive study explores the integration of machine learning (ML) and data engineering techniques to enhance decision-making processes in enterprise solutions. As organizations grapple with increasingly complex data landscapes, the need for sophisticated analytical tools and methodologies has become paramount. This research investigates how ML algorithms, coupled with robust data engineering practices, can be leveraged to extract actionable insights, improve operational efficiency, and drive strategic decision-making across various business domains. Through a combination of literature review, case studies, and empirical analysis, we demonstrate the transformative potential of these technologies in areas such as predictive analytics, customer behavior modeling, supply chain optimization, and risk management. Our findings highlight the critical success factors, challenges, and best practices in implementing ML-driven decision support systems within enterprise environments. Furthermore, we propose a novel framework for integrating ML and data engineering processes that addresses common pitfalls and maximizes the value derived from organizational data assets. This research contributes to the growing body of knowledge on data-driven decision-making and provides practical guidelines for enterprises seeking to harness the power of ML and data engineering to gain a competitive edge in today's data-rich business landscape
An efficient hybrid software-defined networking (HSDN) approach is proposed to optimise the distribution of network traffic in traffic engineering
No full textTraffic engineering (TE) is a very efficient technique for optimizing the distribution of network traffic, leading to improved performance of a hybrid software-defined network (SDN). Traditionally, TE systems have mostly used heuristic methods to centrally optimize the setting of link weights or traffic splitting ratios while dealing with static traffic demand. It is important to realise that as the network grows and management gets trickier, centralised traffic engineering (TE) methods have a hard time keeping up with the huge amount of data they have to process and take a long time to find the best way to route traffic when there are problems or changes in the demand for traffic on the network. Aim to improve the implementation of dynamic and efficient routing in traffic engineering (TE). ring (TE). Efficient hybrid SDN (hSDN) schemes are crucial for the preservation of global information and resource allocation to several applications running in the network. These schemes specifically focus on topology identification, traffic categorization, energy management, and load balancing algorithms. Therefore, in order to enhance network performance by enhancing traffic engineering (TE), it becomes crucial to provide appropriate resources to applications in the network. To accomplish a worldwide optimization goal, An interactive setting for training routing agents with access to partial link use data. To improve the distribution of credit in a multi-agent system, A differential reward assignment method. The purpose of this mechanism is to motivate agents to make choices that are more optimal. The comprehensive simulations conducted on real traffic traces demonstrate the superiority of improving traffic engineering (TE) performance, especially in scenarios when traffic demands vary or network outages happen
THE DEVELOPMENT OF AN EXPERT SYSTEM FOR THE IDENTIFICATION OF CAUSES, PREDICTIONS, AND REMEDIES IN REGARD TO AIRCRAFT DAMAGE
No full textIt is necessary to do routine maintenance, repairs, and upgrades on aircraft whenever possible in order to guarantee that operations run smoothly and to keep the pavements in a satisfactory state. It is possible to prevent aircraft breakdowns and ensure safety by performing periodic maintenance on an aircraft system and detecting failures or defects in the system at an early stage. Consequently, there is a requirement to incorporate cutting-edge technological systems into the process of aircraft repair. The purpose of this article is to contribute to the development of an expert system that can forecast failures, identify the factors that lead to failures, and offer solutions to aircraft failures. The probability tree was utilized in the research project to forecast faults in a selection of aircrafts, specifically the Boeing Aircraft (Year: 2016) and Airbus (Year: 2014) Model (BX2V3). These faults were then diagnosed by the expert system that was developed using the C++ programming language. The purpose of this system was to identify aircraft faults and offer solutions for a variety of faults that were identified
Machine Learning Driven Smart Wearable System to Monitoring and Prediction of CVD
No full textCardiovascular disease (CVD) affects large number of people every year, and its prevalence is rising abruptly, conferring by WHO. Research is being proposed by a smart wearable system, is a state-of-art tool for Online Medical Teleconsultation Service (OMTS). is a framework used to monitor and predict CVD, based on risk factors identified by the features of data set, while data elicited from the system, various classifier algorithms, like, K-Nearest Neighbors (KNN),Naïve Bayes (NB) and Optimization algorithms like, CAT Swarm Optimized and Bayesian Optimize - Support Vector Machine (BO-SVM) being are used to predict CVD effectively. The system is well performed with CSO algorithm, with accuracy of 92.8%, precision of 100%, and sensitivity of 75%, compared with other algorithms.  
The Role of Electronic Management in Developing Health Services and Facilities at the Governmental Hospitals
No full textThe present study aimed at identifying the role of electronic management in developing health services and facilities at the governmental hospitals. The researcher relied on the descriptive analytical method for conducting the present study. A questionnaire was designed with two axes: the first axis (the reality of implementing the electronic management at the military hospital) with (24) statements, and the second axis (the quality of health services at the military hospital) with (24) statements. Participants of the study consisted of (262) employees at the military hospital in the city of Taif in the Kingdom of Saudi Arabia. Results of the research revealed that there was a statistically significant difference at the level of significance (05.0) between the electronic management applications and the level of development in health services and facilities at the governmental hospitals due to the variable of experience years. Also, there was a statistically significant difference at the level of significance (05.0) between the electronic management applications and the level of development in health services and facilities at the governmental hospitals due to the variable of scientific level. Moreover, there was a statistically significant difference at the level of significance (05.0) between the electronic management applications and the level of development in health services and facilities at the governmental hospitals due to the variable of training courses.  
Optimizing Energy Efficiency in Wireless Sensor Networks using Enhanced K-Means Cluster Head Selection
No full textIn Wireless Sensor Networks (WSNs), the efficient management of energy resources is critical to prolonging network lifespan, particularly given the challenges posed by the unpredictable mobility and communication demands of ad hoc mobile devices. Traditional methods for Cluster Head (CH) selection, which group nodes into clusters with designated leaders for data routing and management, often suffer from biases that favour certain nodes. This can lead to uneven energy depletion, with CHs exhausting their power more quickly due to increased responsibilities. To address this issue, this paper proposes an enhanced approach to CH selection using the K-means algorithm, ensuring a more balanced distribution of energy consumption across all nodes in the network. The proposed K-means-based CH selection algorithm incorporates several key parameters, including residual energy, node density, distance to the base station, and signal strength indicators. By integrating these factors, the algorithm ensures that CHs are selected not only based on their proximity to other nodes but also considering their remaining energy and network position. This results in more equitable CH rotations and prevents premature energy exhaustion, thereby extending the network's operational lifespan and maintaining overall performance. Through extensive simulations, the proposed method is evaluated against established CH selection protocols such as LEACH (Low-Energy Adaptive Clustering Hierarchy) and HEED (Hybrid Energy-Efficient Distributed). The analysis focuses on metrics like residual node energy, packet delivery ratio, throughput, and the number of live and dead nodes. The findings demonstrate that the proposed enhanced K-means algorithm outperforms these traditional methods, offering significant improvements in energy efficiency and network sustainability
Efficient Text Extraction Methodologies for Sentiment Analysis: Utilizing University of South Africa Students’ Email Communications as a Case Study
No full textAfter the advent of transformers, highlighted in the paper 'All You Need is Attention’ by Vaswani et al. [1], Large Language Models (LLMs) gained significant traction, notably for tasks like sentiment analysis due to their improved accuracy. Our study focuses on devising a systematic approach to extract textual data from student emails addressed to (University of South Africa) UNISA staff members via the Viva Engage platform. UNISA has customized Yammer, a Microsoft-owned platform typically used for enterprise collaboration and communication, into Viva Engage. Within this platform, students utilize it to express their concerns and opinions on various issues. Given that Viva Engage is intricately linked with UNISA staff emails, every staff member is promptly notified of incoming student messages. Our primary objective is to mine this data for sentiment analysis, aiming to discern the prevalent concerns among students. Specifically, our study delineates the methodologies employed for data extraction and preprocessing tailored for sentiment analysis utilizing LLMs. It is worth noting that this paper exclusively addresses the data extraction phase from Viva Engage emails. The subsequent sentiment analysis, utilizing a BERT LLM, is elaborated upon in a separate research endeavor