ASTU ETD

Adama Science and Technology University Library Electronic Theses & Dissertations (ASTU- ETD) is a new digital institutional repository system that collects, preserve and distribute the scholarly output of the university, mainly postgraduate Electronic Thesis and Dissertation (ETD), articles, proceedings. The system is dedicated to help users to find all the information they might require in order to format and successfully submit their graduate thesis, dissertation and publications electronically. The user friendly web interface enables to maximize & optimize the resource sharing among ASTU different Colleges.

Colleges and Departments in ASTU

Select a Colleges or Departments to browse its collections.

Now showing 1 - 6 of 6

Recent Submissions

Evaluating The Impacts Of Land Use And Climate Variability On Water Availability In Hargeisa Watershed, SomalilandThesis
(ASTU, 2025-10) Abdirahman Ibrahim
Hargeisa, Somaliland, faces a critical water crisis driven by rapid urbanization and a volatile climate. This study provides the first comprehensive hydrological assessment for the region, aiming to quantify the impacts of historical land use change (2000–2024) and past climate patterns (1995–2024) on water availability, and to identify the resulting socio economic impacts on the community. To overcome the challenge of an ungauged watershed, a regionalization approach was employed using the Soil and Water Assessment Tool (SWAT+). The model was successfully calibrated and validated against a physically similar gauged watershed, achieving satisfactory performance (Calibration: R²=0.55, NSE=0.52; Validation: R²=0.53, NSE=0.50). This successful validation provided a scientifically sound foundation for quantifying the drivers of water scarcity. The climate analysis revealed a statistically significant warming trend (Tmax increased by 0.037°C/year and Tmin by 0.025°C/year), while precipitation showed high inter-annual variability but no significant long-term trend. The study integrated these hydrological simulations with an analysis of satellite-derived land use maps and socio-economic data from a survey of 134 residents. The findings reveal a profound landscape transformation: fueled by population growth, urban areas expanded by 377%, primarily converting natural grasslands. Hydrological simulations show this has critically impaired the watershed's ability to store water, increasing surface runoff by 24.1% while reducing groundwater percolation by 25%. This provides a direct scientific explanation for the hardship reported by 75% of residents, who find their water supply insufficient. The study concludes that Hargeisa’s water crisis is a direct consequence of uncontrolled urban expansion, a condition significantly worsened by the region’s inherent climate variability. This research provides a crucial, data-driven foundation for policy, highlighting the urgent need for integrated water management strategies, such as protecting groundwater recharge zones and implementing urban rainwater harvesting, to ensure Hargeisa's future water security.
Regional Determinants Of Road Traffic Accident In Oromia: Analyzing Vehicle - Related InterventionThesis
(2025-10) Efa Dega
The regional determinants of road traffic accidents in the Oromia region, has been shown to have issues with particularly regarding vehicle- related problems in Oromia region. This can lead to the causes of death and high injury with the Oromia region being one of the most affected areas. To address this, a comprehensive analysis of road traffic accident was conducted using Microsoft Excel, followed by optimization using Design Expert software. The analysis specifically focused on identifying vehicle-related problems that contribute to these regional determinants of road traffic accidents in the Oromia region. The analysis revealed several key findings regarding vehicle-related problems and their association with road traffic accidents in Oromia. Brake failures are the leading causes (30%), likely due to poor maintenance culture and lack of enforcement. There is also steering problems (30%) often arise from neglected repairs and misalignment, contributing significantly to loss of control. and finally, tires of (25%) are strongly linked to overloading and poor road conditions, increasing blowout risks. Therefore, this structured quantification helps prioritize brake and steering system improvements as the most critical areas for reducing Road traffic accidents in Oromia. The findings highlight the need for targeted interventions to address these problems and reduce occurrence of accident. The results underscore the critical importance of focusing on mechanical failures as primary intervention area. Targeted interventions focusing on vehicle condition, proper loading and maintenance systems can significantly reduce accident rates while supporting the region’s mobility needs and economic development. Brake failure, steering problem and tire system failures contribute significantly to Road traffic accidents in Oromia. Addressing these issues through better maintenances, enforcement, and driver education can reduce preventable accidents. Road traffic accidents in Oromia like in many other regions are often influenced by vehicle mechanical failures, particularly in critical systems such as brakes, tires and steering issues.
Predicting Task Offloading Requests and Resource Demands in IoT Edge Computing using Hybrid Deep Learning ModelThesis
(ASTU, 2025-10) Abdu Aliyi
The dynamic and heterogeneous nature of IoT edge environments, where billions of devices continuously generate diverse and time-varying workloads, necessitate predictive mechanisms capable of accurately forecasting both task offloading requests and resource demands. While existing prediction-based offloading studies have primarily focused on forecasting task volumes or a single resource and in some cases both CPU and memory), none have jointly predicted offloading requests with task-level features alongside their associated resource requirements. To address this gap, we propose a hybrid multi-task learning CNN-BiLSTM-Attention model. The CNN component extracts local temporal patterns, the BiLSTM captures long-range dependencies, and the Attention mechanism emphasizes the most informative time steps and features. The model jointly predicts two categorical offloading parameters (task priority and delay tolerance) and four continuous resource-demand metrics (CPU request, memory request, maximum CPU usage, and maximum memory usage). A multi-objective learning strategy was employed, with classification targets representing task-level semantics and regression targets estimating resource demands. Optimization was performed using a weighted combination of sparse categorical cross-entropy and mean squared error losses to effectively balance the heterogeneous objectives. The Google Cluster dataset was employed to train and evaluate the proposed model. Model performance was assessed using standard evaluation metrics, including MAE, MSE, RMSE, R2, MAPE and ordinal-aware accuracy for categorical targets. Post training quantization was performed for edge compatibility. The results indicate the superior performance of the proposed model consistently achieving the best results, with a minimum values of MAE =0.00014, MSE and RMSE values as low as 0.0001, and an R2 score of 0.99, alongside a minimum MAPE of 2.30%. For classification tasks, the model attained the highest accuracy of 0.98. Additionally, we have benchmarked deep learning and classical machine learning models. Furthermore, when compared with prior state-of-the-art studies on workload prediction using the same dataset, the proposed model improved prediction accuracy by 6.4% in terms of R2, thereby demonstrating its outperformance and advancement over existing approaches.
Human Face Emotion Recognition in Thermal Images Using cGAN and EfficientNet with Attention MechanismsThesis
(ASTU, 2025-11) Tahir Aman
Human use emotions to express their feelings and effectively interact with others. Human express emotions through hands, voice, gestures, and most importantly, facial expressions. Facial emotion recognition widely used in human-computer interaction, security, and healthcare. Traditionally based on posed visible images, such methods fail to capture natural emotions and are prone to lighting conditions. Visible spectrum imaging has several limitations, including sensitivity to lighting conditions, facial obstructions, and potential variations in facial expressions. Thermal imaging is less affected by obstructions of the face and changes in illumination. Thermal facial emotion recognition offers a robust alternative to conventional visible-spectrum methods by mitigating the effects of lighting variability and deceptive facial expressions. However, the use of thermal images for emotion recognition has not been extensively explored, primarily due to the limited availability of thermal image datasets. This research investigated the impact of enhanced preprocessing, synthetic data generation, and advanced deep learning architectures on thermal emotion classification. In this paper we used bilateral filters, Gaussian Blur and CLAHE for image preprocessing, data augmentation, cGANs for synthetic data generation, and EfficientNet-B5with CBAM for feature extraction. We utilized CBAM on the EfficientNet-B5 blocks. Our proposed method passed the extracted features to ResNet-18 for the classification of human facial emotions of thermal images into five emotional expressions such as happy, sad, angry, natural, and surprise. The proposed method used comprehensive facial thermal datasets for training and testing to compare its performance on baseline work. We employed the Fréchet Inception Distance to evaluate the realism and quality of the generated images. Accuracy, precision and F1-score metrics are utilized to assess the performance of the model. Experimental results demonstrated a significant improvement in recognition performance, achieving an accuracy of 98.81%. These findings highlight the effectiveness of the proposed pipeline for thermal facial emotion recognition and suggest promising potential for deployment in real-world conditions.
Multimodal Understanding Amharic Video Question Answering using Bidirectional Cross Modal AttentionThesis
(ASTU, 2025-11) Helina Tefera
Amharic Video multi modal Understanding for Amharic Video Question Answering using Bidirectional Cross Modal Attention is a novel deep learning approach designed to enhance the comprehension of Amharic video content through a fusion of visual and textual modalities. One of the primary challenges in video question answering is the heterogeneous nature of visual and textual data, especially in low resource languages like Amharic. Conventional approaches often rely on randomly sample video frames, did not consider sematic relation between object ,and all of them are for English. To overcome these limitations, this study introduces a Bidirectional Cross Modal Attention mechanism with CLIP based best frame selection, which models fine grained interactions between video representations CLIP features, temporal embeddings, object features and the question encoding using BERT. Previous models either aggregate all visual features at once or treat the question as a global embedding, which results in loss of word level alignment and spatial temporal correspondence. In contrast, the Bidirectional Cross Modal Attention model allows both visual and textual tokens to attend to each other iteratively, improving semantic alignment between questions and relevant visual content. To further enhance understanding, multiple visual cues such as CLS tokens, CLIP embeddings, object detections from FastRCNN, and temporal spatial features are integrated. An bidirectional cores modal attention based fusion layer selectively combines these features. The Bidirectional Cross Modal Attention Bidirectional Cross Modal Attention VQA model not only introduces the first ever benchmark for Amharic Video Question Answering (Amharic VQA) but also achieves significant improvements over current state of the art methods on the English MSVD QA dataset. The Amharic Video Question Answering model achieved 48.21% accuracy, the English model using English MSVD QA reached 58.71%, showing a notable improvement compered with Yu et al. 2024 with final accuracy with 48.2% on MSVD QA and Tang et al. 2024 with final result 39.1%. These results highlight the effectiveness of fine grained, bidirectional attention in enhancing semantic fusion between video content and questions, improved Video Question Answering performance, particularly in English