COMPUTER ENGINEERING

This study presents the design and implementation of a smart, lightweight, federated deep learning system that integrates solar-powered cameras for automated attendance, unauthorized entry prevention and real-time cyber threat detection in academic environments. Using TensorFlow Lite, Python, and a Flask-based web interface, the model achieved high accuracy in facial recognition while maintaining low computational and energy costs. A structured SQLite3 database supported efficient local data handling, while solar energy integration enabled autonomous and sustainable operation. This project validates the potential of combining renewable energy, artificial intelligence, and federated learning to enhance classroom management and IoT security in low resource settings.

This project presents the design and implementation of an automated waste management system utilizing an Arduino Uno microcontroller, ultrasonic sensors, and a servo motor to enhance efficiency and hygiene in waste disposal. The system continuously monitors the fill level of a waste bin using an ultrasonic sensor, which provides real-time data to the Arduino. When the sensor detects that the bin is nearing capacity or a user is present, the Arduino activates a servo motor to automatically open and close the bin lid, enabling touchless operation and reducing the risk of contamination. Powered by a 9V replaceable battery, the system is portable and well-suited for environments with unreliable electricity supply. Rapid lid response, with positive user feedback regarding convenience and hygiene. The project highlights the potential for scalable, low-cost smart waste solutions in both urban and rural settings, and lays the groundwork for future enhancements such as IoT connectivity, renewable energy integration, and automated waste sorting for improved sustainability and resource management

The convergence of Artificial Intelligence (AI) with Digital One Health systems, which integrate human, animal, and environmental health data into unified platforms, has introduced unprecedented cybersecurity challenges that conventional frameworks fail to sufficiently address. This project proposes the design, implementation, and evaluation of an adaptive AI- driven security framework tailored specifically to safeguard sensitive cross-sector health data within complex hybrid cloud and on-premise infrastructures. The framework leverages foundational cybersecurity principles, such as Zero Trust Architecture and layered defensein-depth, combined with advanced AI-powered anomaly detection models including Long Short-Term Memory (LSTM) Autoencoders and Isolation Forest algorithms to enable realtime identification and mitigation of emerging threats. Employing open-source technologies integral to the system architecture; OPNsense for perimeter firewall and VPN, WireGuard for encrypted communication, Suricata IDS/IPS for intrusion detection and prevention, Wazuh Security Information and Event Management (SIEM) for centralized log aggregation, Keycloak for Multi-Factor Authentication (MFA), and Debezium for Database Activity Monitoring (DAM), the framework demonstrates a holistic approach to securing Digital One Health ecosystems. Central to this effort is the development of a robust AI-powered security dashboard based on the ELK stack (Elasticsearch, Logstash, Kibana), which provides unified visualization, real-time alerting, and compliance monitoring. The research adopts a design science approach, systematically assessing security vulnerabilities unique to hybrid Digital One Health infrastructures and implementing layered controls to address gaps in cross-sector interoperability and AI-specific risks such as adversarial attacks and data poisoning. Validation is performed using comprehensive threat simulations that mimic real-world attack vectors including SQL injection, lateral movement, privilege escalation, and Distributed Denial-of-Service (DDoS) scenarios. Quantitative performance metrics reveal a high detection rate (approximately 97%), significant reductions in both mean time to detect (MTTD) threats by 70% and false positive alerts by 58%, thereby illustrating improved operational efficiency and accuracy compared to traditional static security measures. Despite promising results, challenges remain around optimizing AI model training data quality, managing performance overhead in resource-constrained settings, and balancing stringent security controls with healthcare delivery imperatives. The study advocates for future enhancements including federated learning models to preserve data privacy, lightweight AI implementations for broader accessibility, automated compliance frameworks
to navigate complex regulatory environments, and expanded user training programs to address human-centric security gaps. This project contributes a scalable, ethically aligned cybersecurity framework specifically engineered for AI-powered Digital One Health systems, addressing the multifaceted demands of securing sensitive, interconnected health data across diverse environments. By merging advanced AI capabilities with proven cybersecurity practices and a centralized management platform, it advances the resilience and trustworthiness of modern healthcare ecosystems, fostering innovation while safeguarding critical digital health infrastructure globally.

The increasing complexity of modern programming environments has created a growing need for intelligent, real-time support systems capable of assisting students and developers in understanding, writing, and debugging code. This project presents the design and implementation of an AI-powered chatbot web application that leverages Large Language Model (LLM)-based Natural Language Processing (NLP) and Remote Code Execution APIs to address programming queries interactively. The system integrates Google’s Gemini API for natural language understanding and JDoodle API for live code execution within a secure, microservicebased architecture. The chatbot provides conversational assistance, real-time code validation, debugging support, and algorithmic explanations through a responsive web interface built with React, TypeScript, Tailwind CSS, and Vite. The backend is implemented using Node.js, Express.js, and MongoDB, with modular microservices for authentication, chatbot intelligence, and code execution, all orchestrated through an API Gateway.This architecture ensures scalability, maintainability, and independent service deployment, while also improving accessibility for users in resource-constrained environments. The system demonstrates how combining LLMs with execution APIs can create a more reliable and contextaware programming assistant than traditional static or rule-based systems. The outcome of this project is a robust, user-friendly, and intelligent chatbot that enhances learning efficiency, developer productivity, and accessibility to programming support, particularly for students in developing regions. It contributes to ongoing research in AI-driven education, NLP, and intelligent tutoring systems, offering a sustainable model for future AIintegrated learning platforms.

This project presents the design and implementation of an Internet of Things (IoT) based water quality monitoring system that enables real-time assessment of key water parameters using low-cost sensors and microcontroller technology. The system employs an ESP32 microcontroller integrated with pH, turbidity, Total Dissolved Solids (TDS), and temperature sensors to measure the chemical and physical characteristics of water. The sensed data are processed, displayed locally on a 20×4 Liquid Crystal Display (LCD), and transmitted wirelessly to a custom web interface for remote visualization and storage.The proposed system addresses the limitations of traditional laboratory-based water testing methods, which are typically expensive, time-consuming, and prone to human error. By leveraging IoT connectivity, the system provides continuous, accurate, and cost-effective monitoring of water quality, suitable for domestic, educational, and small-scale environmental applications. Experimental results show that the developed prototype effectively measures and reports water parameters with acceptable precision and reliability.Overall, the system demonstrates a practical approach to automating water quality analysis, promoting accessibility, sustainability, and public health through real-time digital monitoring.

The global transition towards sustainable energy and the decarbonization of power and transport sectors have positioned Lithium-ion Battery-based Energy Storage Systems (Li￾ion ESS) as a critical technological enabler. Despite their dominance, characterized by high energy density and efficiency, these systems face persistent challenges related to performance trade-offs, safety, economic viability, and sustainability, with distinct implications for developing regions like Sub-Saharan Africa. This study undertakes a comprehensive review of Energy Storage Systems for Lithium-ion Batteries to evaluate recent advancements, identify key challenges, and map future research directions. Employing a systematic qualitative review methodology, this research analyzes and synthesizes findings from peer-reviewed literature, technical reports, and conference proceedings published between 2020 and 2025, sourced from databases including IEEE Xplore, ScienceDirect, and SpringerLink. The analysis is structured around a comparative framework that evaluates dominant lithium-ion chemistries—specifically Lithium Nickel Manganese Cobalt Oxide (NMC), Lithium Iron Phosphate (LFP), and Lithium Nickel Cobalt Aluminum Oxide (NCA)—across key parameters: energy density, safety, cycle life, cost, and application suitability. The review reveals a critical market bifurcation. NMC and NCA chemistries remain preferred for applications demanding high energy density, such as premium electric vehicles. In contrast, LFP has emerged as the superior choice for stationary storage and an increasing segment of the EV market, owing to its exceptional safety, long cycle life, cost-effectiveness, and cobalt-free supply chain. The study further highlights that system￾level components, particularly the Thermal Management System (TMS) and Battery Management System (BMS), are paramount in determining overall ESS performance and lifetime. From a regional perspective, the adoption of Li-ion ESS in Sub-Saharan Africa, and Nigeria specifically, is hampered by infrastructural deficits, reliance on imports, high costs, and an underdeveloped recycling framework. The study concludes that the future of Li-ion ESS lies in application-specific vii optimization, the integration of digital technologies like AI for predictive management, and the development of robust circular economy models. It recommends intensified research into solid-state batteries, the standardization of second-life battery protocols, and the formulation of policies that support local capacity building and sustainable deployment in developing economies to harness the full potential of lithium-ion energy storage for a clean energy future.

The growing interconnection between human, animal, and environmental health has emphasized the importance of integrated surveillance systems for early detection and coordinated response to zoonotic and environmentally linked diseases. However, in many low-resource settings such as Nigeria, disease monitoring remains fragmented across sectors, resulting in delayed reporting and weak cross-sectoral collaboration. This project presents the design and implementation of a Digital One Health Surveillance (DOHS) Dashboard, a web-based system developed to unify real-time disease surveillance across the three One Health domains—human, animal, and environmental health. The system was built using a React-based frontend and a RESTful backend API, integrating modules for authentication, case management, data analytics, and geospatial visualization. HTTP polling mechanisms were implemented to achieve real-time data synchronization without requiring WebSocket infrastructure, ensuring compatibility with constrained networks. The dashboard provides role-based access control, sector-specific reporting interfaces, Excel-based data export, and an interactive GIS map powered by React-Leaflet for spatial trend analysis. Data are securely transmitted via token-authenticated API calls, while the modular architecture allows future expansion and integration with existing national systems such as SORMAS or DHIS2. Deployment was designed for both cloud-based Virtual Private Servers (VPS) and onpremise servers within health facilities, supporting continuous operation even in areas with limited internet access. The system demonstrated efficient cross-sectoral data integration, realtime monitoring of case reports, and improved accessibility for public health officers, veterinary staff, and environmental personnel. The DOHS dashboard contributes to the operationalization of the One Health approach in Nigeria by providing a scalable, adaptable, and user-friendly digital infrastructure for integrated disease surveillance. It lays the foundation for data-driven decision-making and offers a replicable model for similar low-resource environments across Africa.

The increasing demand for reliable, sustainable, and decentralized energy systems has accelerated the adoption of microgrids, particularly in regions with unstable power infrastructure such as Nigeria. However, the effective design and management of microgrids remain challenging due to the variability of renewable energy sources and the lack of intelligent tools that integrate real-time environmental data, predictive analytics, and optimization techniques. This study addresses these challenges by developing an Intelligent Microgrid Management and Optimization System aimed at improving the planning, performance, and operational efficiency of renewable energy-based microgrids. The system was developed as a Python-based desktop application incorporating machine learning, optimization algorithms, and real-time data integration. Support Vector Regression (SVR) models were implemented to predict energy generation from multiple renewable sources using environmental inputs such as solar irradiance, temperature, and wind speed. A genetic algorithm was employed to optimize system component sizing, including solar panels and battery storage, based on cost and reliability objectives. Real-time weather data was retrieved via an API and integrated into a dynamic simulation engine, while a graphical user interface built with PyQt6 provided interactive visualization, monitoring, and control of system operations. The results demonstrate that the developed system achieves high predictive accuracy, with machine learning models attaining strong performance metrics (R² values above 0.88 across energy sources) and low prediction errors. The optimization module effectively identified cost-efficient and reliable microgrid configurations, while the simulation engine maintained stable energy balance and accurate battery state tracking under varying conditions. Overall, the system proved to be computationally efficient, user-friendly, and capable of supporting real-time decision-making, thereby offering a practical and scalable solution for intelligent microgrid design and management.

This study presents the design and implementation of a smart, lightweight, federated deep learning system that integrates solar-powered cameras for automated attendance, unauthorized entry prevention and real-time cyber threat detection in academic environments. Using TensorFlow Lite, Python, and a Flask-based web interface, the model achieved high accuracy in facial recognition while maintaining low computational and energy costs. A structured SQLite3 database supported efficient local data handling, while solar energy integration enabled autonomous and sustainable operation. This project validates the potential of combining renewable energy, artificial intelligence, and federated learning to enhance classroom management and IoT security in low resource settings.

This project presents the optimization of solar inverter efficiency using machine learning algorithms to improve power generation accuracy and system reliability under varying environmental conditions. Traditional solar inverter systems and Maximum Power Point Tracking (MPPT) methods often experience limitations in adapting to fluctuations in solar irradiance, temperature, and shading conditions, leading to reduced efficiency and energy loss. To address these challenges, this study developed and evaluated machine learning models capable of predicting and optimizing inverter performance in real time. Environmental and operational data including irradiance, temperature, day, hour, and inverter performance metrics were collected from the NASA and NSRDB datasets for the University of Benin region. Data preprocessing techniques such as normalization, interpolation, and feature engineering were applied before model training. Three machine learning models — Random Forest (RF), Gradient Boosting Machine (GBM), and Artificial Neural Network (ANN) — were implemented and evaluated using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Coefficient of Determination (R²). Results showed that the ANN model outperformed the other models with an MAE of 0.019, RMSE of 0.029, and R² value of 0.962. The optimized system achieved an efficiency improvement of 8.3% compared to conventional MPPT methods. The study further demonstrated the capability of machine learning algorithms to adapt to changing environmental conditions and improve solar inverter performance. The developed model was deployed using Django REST Framework for real-time prediction and monitoring. This research confirms that machine learning-based optimization can significantly enhance solar inverter efficiency, reduce energy losses, and contribute to sustainable and intelligent renewable energy systems.