| 1 |
Author(s):
Ekoro Ekoro Igo (Ph.D.), Moses A. Agana (Ph.D.).
Page No : 1-14
|
Healthcare Sensor –Based Physiological Parameters Data Aggregation and Analytics Scheme for Monitoring of Patients in Internet of Medical Things (IoMT) Enabled E-Healthcare Platform.
Abstract
Inadequate mechanism for efficient patient health monitoring and data aggregation mechanism in healthcare services has posed serious bottleneck to healthcare delivery the world over. Healthcare sensors have become an accessible means for the communication of data from patients to medical personnel. The use of Healthcare sensors is also an invention in medical practice which involves measuring vital physiological parameters of patients with the objectives of detecting disorders to mitigating them and preventing severe complications. In this research, a system architecture for effective tracking of patients’ health by deploying the Internet of Medical things (IoMT) as an approach is developed. The methodology used for this research was the Design Science research methodology. The choice of the methodology was born out of the fact that the methodology involves the construction and evaluation of the prototype (artefacts) that address a considerably acknowledged problem. In this adopted approach, real-time data are sent to a local server through communication channels (Wi-Fi) and then transmitted to the Internet of Things (IoT) server through a designated network route using a Wi-Fi module. For efficient transmission of vital signs to the cloud, a Blynk IoT- server was used as platform. Two types of sensors DS18B20 an AD8232 ECG were deployed in monitoring and tracking body temperature and heart rate respectively, the XAMPP server was used as the local server platform. The outcome of this research include a developed artefacts and a mechanism where real-time numerical data is communicated to the report platform which is further transformed into ordered numbers. The novelty in this work involves collating real-time data from sensors attached to patients in IoMT-enabled environments, thereafter a simple linear regression model was deployed to convert these real-time data to ordered numbers which indicate the level of severity of the vital signals collated. These ordered numbers are converted to visuals to enhance ease of view. To the best of our knowledge, this approach has not been implemented by previous research studies reviewed. Based on our findings, we recommend that: the prototype system be developed and deployed in healthcare facilities and the designed web applications be plugged into an existing web domains of healthcare facilities to enhance timely intervention by the healthcare experts.
| 2 |
Author(s):
Adeyemi J. O., Ogunlere S. O., Akwaronwu B. G..
Page No : 15-50
|
Real-Time Detection of Examination Malpractices Using Convolutional Neural Networks And Video Surveillance: A Systematic Review with Meta-Analysis.
Abstract
This research project develops a system for automatically detecting cheating and identifying students in order to improve exam integrity while addressing the shortcomings of traditional monitoring methods. The technology detects and captures cheating pupils in real time using both machine learning and manual tactics. A study and analysis were conducted to provide evidence-based recommendations for designing effective automated cheating detection systems in educational settings. According to the PICOS framework, the research is aimed at students who struggle with exam cheating (Population), focuses on developing a detection system (Intervention), compares traditional monitoring techniques to the new system (Comparison), seeks to improve accuracy and fairness in identifying cheating (Outcome), and collects evidence using systematic review and meta-analysis methods (Study Design). The literature search followed PRISMA criteria and includes papers from the Scopus and Google Scholar databases from 2013 to 2024. The inclusion criteria included research papers that investigated exam participants, instances of cheating, and the application of new technologies such as deep learning and machine learning. Articles that were not about examination malpractices or did not use advanced technological tools were rejected based on particular criteria. A total of 37 articles were reviewed. The findings demonstrate how new technology may significantly increase the credibility and dependability of tests, ensuring academic honesty.
| 3 |
Author(s):
Bright Gazie Akwaronwu, Innocent Uche Akwaronwu, Oluwabamise Joseph Adeniyi.
Page No : 51-70
|
Machine Learning for Detecting DoS Attack: A Comparative Approach.
Abstract
Denial-of-Service (DoS) attacks has been a critical challenge in cybersecurity, disrupting the availability of network services and causing significant operational and economic losses. To ascertain the most suitable approaches to mitigate to dilemma, this study compares the effectiveness of some selected machine learning models in identifying denial-of-service (DoS) attacks. Two ensemble learning models, Random Forest (RF) and Extreme Gradient Boosting (XGB), showed remarkable accuracy and dependability. RF performed almost perfectly on criteria including accuracy (99%), precision (99%), and recall (99%). Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) stood out as Deep Learning models, capturing complex patterns with CNN achieving an accuracy of 98% and a perfect AUC score of 1.00. The models utilized the Recursive Feature Elimination (RFE) to select significant features and ensured proper data balancing techniques for robust model training and evaluation, minimizing overfitting and enhancing generalization. The results highlight RF and CNN as the best-performing models, with RF offering interpretability and computational efficiency, while CNN excels in handling unstructured and complex datasets. This study underscores the need for context-driven model selection and suggests exploring hybrid approaches that integrate the strengths of ML and DL for improved DoS attack detection. Future work should aim to enhance scalability and adaptability for real-world cybersecurity applications.
| 4 |
Author(s):
Idowu Peter Sunday, Folashade Y. Ayankoya (Ph.D.).
Page No : 71-98
|
Enhanced Keylogger Classification Using Deep Learning: A Systematic Review.
Abstract
This systematic review explores methodologies for detecting and mitigating keyloggers, pervasive cybersecurity threats that surreptitiously capture keystrokes. After conducting thorough database searches, 26 relevant studies were found, showing a wide range of methods including machine learning algorithms, heuristic techniques, and behavior-based strategies. The review underscores the efficacy of combining proactive and reactive measures in countering keylogger threats, with machine learning algorithms exhibiting varying degrees of success. Significantly, creating interfaces that are easy for users to use is found to be a crucial element in improving user knowledge and making it easier to take quick action.
However, the analysis also points out some drawbacks, such as the lack of extended verification for suggested approaches and variations in how algorithms are designed in different research. These findings underscore the imperative for ongoing innovation and collaboration among practitioners and policymakers to develop standardized protocols and address emerging threats comprehensively. Overall, this review offers valuable information on how to detect and prevent keyloggers, which can help direct future research in this important area.
| 5 |
Author(s):
Folashade Y. Ayankoya (Ph.D.), Shade O. Kuyoro (Prof.), Ernest E. O. (Ph.D.), Peter S. Idowu.
Page No : 99-129
|
Machine Learning in Predicting Lung Cancer Outcomes from Xenograft Models: A Systematic Review and Meta-Analysis: A Systematic Review.
Abstract
The predictive ability of machine learning algorithms in determining treatment outcomes for lung cancer is investigated in this work utilizing patient-derived xenograft (PDX) mouse models. The review focuses on panitumumab, an epidermal growth factor receptor (EGFR)-targeted drug, across a variety of xenograft models.
In addition to the original study, a systematic evaluation was undertaken between 2009 and 2024 to evaluate machine learning applications for predicting lung cancer outcomes using xenograft models. Decision trees, neural networks, and support vector machines were found to be the most commonly investigated models. The evaluation stresses methodological variety, computational techniques, and the types of data used. Inclusion criteria include research that combines many procedures and data types, whereas exclusion criteria target studies that lack detailed approach information or are presented in non-English languages.
The findings reveal differing degrees of accuracy among models and give insights on areas for development. Patients' clinical records, gene expression datasets, and molecular profiling are all primary data sources. The systematic review intends to uncover patterns, obstacles, and overall predictive model performance, which will serve as a platform for future research to improve lung cancer prediction and treatment results.
| 6 |
Author(s):
Mugerwa Joseph, Ajaegbu Chigozirim, Oyerinde Emmanuel , Awodele Simon Olufikayo .
Page No : 130-140
|
An Efficient Mac-Based ICMP Verification Algorithm for Early Detection of Bandwidth-Depleting DDOS Attacks
Abstract
Distributed Denial-of-Service (DDoS) attacks continue to pose a significant threat to the availability and reliability of online services. This paper presents a novel detection algorithm that leverages Message Authentication Code (MAC)-based verification of ICMP traffic to identify and block bandwidth-depleting DDoS attacks. Unlike threshold-based or machine learning-dependent techniques, the proposed algorithm uses IP and MAC address correlation to validate the legitimacy of packets, effectively filtering spoofed traffic in real time. The approach was implemented and tested using the NS-2 simulation environment. Results demonstrate an average detection accuracy of 88.89%, with zero false positives and negligible resource overhead. The proposed method offers a lightweight and effective solution suitable for deployment in edge and enterprise networks. This research contributes a simple yet robust technique to the existing portfolio of DDoS mitigation strategies.
| 7 |
Author(s):
Philip-Kpae F. O., Ataisi A. S..
Page No : 141-162
|
Optimizing Sustainable Power Delivery in Sensor Networks through Hybrid Energy Harvesting for Autonomous Energy Storage Solutions.
Abstract
The increasing proliferation of low-power embedded systems and Internet of Things (IoT) devices, especially in remote or energy-constrained environments, has intensified the demand for sustainable and autonomous energy solutions. Traditional battery-powered systems face limitations in longevity, maintenance, and environmental adaptability, prompting the exploration of alternative energy harvesting techniques. This study investigates four energy harvesting methods—solar, vibration, radio frequency (RF), and capacitor-based storage—to evaluate their performance and suitability for reliable power delivery in such applications. MATLAB simulations were employed to model the energy output of each technique across a normalized input range (0 to 1), ensuring a consistent basis for comparison. Results showed that solar energy harvesting, using an 18% efficient photovoltaic panel with a 0.02 m² surface area, and vibration-based systems with 150 N/m stiffness and 0.002 m displacement, both achieved a normalized peak output of 1.0. RF harvesting, utilizing 1 W transmission power and a gain of 2 at both ends, performed less efficiently with a peak below 0.4. Capacitor storage, modeled with a 0.01 F capacitor over a 1V to 5V voltage range, demonstrated a parabolic output curve with peak performance at mid-range inputs. These results highlight the potential of integrating solar and vibration energy harvesting methods, complemented by capacitor-based buffering, to form robust hybrid power systems. The study contributes to knowledge by providing a comparative performance analysis of multiple harvesting techniques under standardized conditions and offers design insights for engineers developing energy-resilient embedded and sensor-based applications. The implications suggest that hybrid energy systems, supported by smart energy management strategies, can significantly enhance the operational autonomy, reliability, and sustainability of next-generation IoT deployments.
| 8 |
Author(s):
Francis Abei, Benson Mirou, Mohsen Aghaeiboorkheili.
Page No : 163-173
|
Traffic Congestion – The Problem Affecting the Flow of Traffic in Cities.
Abstract
Papua New Guinea has a lot of serious issues within its community. One particular issue that is becoming an increasing bigger problem is that of traffic congestion. This problem is affecting cities such as Port Moresby, Goroka, and Lae. This paper aims to examine the origin, effects, and possible solutions for this issue. The increase in vehicles on the road, very bad road infrastructure, ineffective public transportation systems, and poor city planning are some of the causes that affect the flow of traffic, as shown within the study. Long traffic congestions then can affect other sectors of the society causing losses in finance, health hazards, and safety issues. The methods used within this study include gathering information from earlier studies and simulating the flow of traffic within the city of Port Moresby using the VISSIM software with certain parameters. Practical solutions that can be carried out include having better city planning, effective public transport systems, more stringent traffic laws, improved road maintenance, and better road infrastructure which are all recommended within this study. The city planners and government departments must work together to address this issue. This study emphasizes the need for sustainable transportation solutions in order to raise the standard of living for people within this society.
