About Addis Ababa University Institutional Repository (AAU-ETD)

AAU-ETD is an electronic open access institutional repository of Addis Ababa University that makes available and digitally preserves the scholarly outputs produced at AAU. The repository contains both published and unpublished work including: theses and dissertations,preprint,staff and student publications.

Services provided by AAU Library repository specialist:

  • Create Colleges/Institutes and collections
  • Provide depositing services
  • Train and facilitate community members to add materials
  • Review and add additional descriptive information (metadata) to each record.
  • Maintain open access and ensure preservation of materials
  • Maintain the software and hardware required for continuous service
  • Research copyright and seek permissions

All faculty are invited to submit their research to the AAU-ETD which is operated and maintained by Addis Ababa University Library. For further information please contact us at ________


Recent Submissions

Comparative Assessment of Excavation Supporting Systems in Addis Ababa
(Addis Ababa University, 2024-01) Kidist Afewerk; Henok Fikre (PhD)
Addis Ababa, Ethiopia's growing capital city, has witnessed rapid urban development, necessitating the construction of towering structures. As the need for deep excavation arises, Engineers face the challenge of maintaining stability while digging into the earth. This necessitates using temporary earth retaining structures known as excavation supports or shoring to prevent soil collapse and ensure precise excavation. Various methods are available for excavation support, including soldier beams and lagging, sheet piling, bored pile walls, soil nailing walls, and slurry (diaphragm walls). In Addis Ababa, specialized firms dedicated to foundations construct excavation support systems. The first foundation specialist company in the country, BAUER MIDROC Foundation Specialist Plc., was established on June 12, 1998. Addis Ababa has varying soil types, but the protection system used is uniform, causing specific problems such as costly temporary shoring systems and time-consuming installation. Local companies compete by compromising safety, leading to the collapse of excavation support systems with major damage, including loss of human life, additionally, machinery has a short lifespan due to difficult strata placement. The study aims to access different shoring methods in Addis Ababa and compare conventional contiguous pile walls with soil-nailing walls for various sites and soil conditions that are practical, reliable, appropriate, and adaptable solutions for local firms. To achieve this, first gather information about three different sites, including soil strength, excavation depths, surcharge load, and available space for the supporting system structure. Numerical methods such as finite element analysis and limit equilibrium analysis will be used to create models of the deep excavation stabilizing methods for each site and method. After that compare two methods of construction costs and construction period based on the obtained designs. In conclusion, this research aims to provide valuable insights into the most effective and suitable excavation supporting systems for specific site conditions and soil types in Addis Ababa. The particular approach employed in this study, along with considering various factors, ensures that the findings are reliable and applicable to the context at hand. The results of this study can be valuable for contractors and Engineers in selecting the appropriate excavation supporting system in Addis Ababa.
Performance Comparison of Multi-Mode Modulation Techniques for SDR Using FPGA
(Addis Ababa University, 2023-11) Sisay Bogale; Yihenew Wondie (PhD)
Radio devices that were previously built in hardware have been replaced in recent years by reconfigurable software defined radio (SDR) systems. Conventional hardware-based radios have restricted multi-functionality and are physically changeable only. This leads to an increase in production expenses and a reduction in the number of waveform standards that can be supported. A rapid and affordable answer to this issue is provided by software-defined radio technology, which enables software upgrades for multi-mode, multi-band, and multi-functional wireless devices. In SDR, different modulation techniques are used to achieve efficient communication over a radio channel. Multi-mode modulation is an approach that allows the use of multiple modulation schemes in a single system, which can enhance the flexibility and resilience of communication systems. This paper presented a design and implementation of multi-mode modulation techniques for SDR using FPGA and analyze the performance based on the FPGA resource utilization. It combines six modulation schemes: QASK, QPSK, QAM, AM, PM and FM to create multi-mode modulation system. The performance of this multi-mode modulation system is evaluated in terms of FPGA resource utilization such as total computational power, total number of Look Table (LUT) or memory used, Flip Flops (FF) and Input/Output (IO) port usage. Xilinx Vivado system generator for DSP with MATLAB/Simulink is used to design, simulate and verify the multi-mode modulator, which would then be implemented on a Xilinx Zedboard FPGA hardware. A total of 0.225W power, 844 number of LUT and 1 IO port is utilized by the implemented design. The biggest thing we achieved in this research is that we saved computational power. 1.572W and 1.134W amount of power is saved by our design as compared to previous two studies.
Design and Optimization of Geodetic Network: A Case of Ethiopia
(Addis Ababa University, 2023-09) Haileslassie Muluken; Andinet Ashagre (PhD)
The optimization of a geodetic network is to enhance precision and efficiency in surveying practices. Precision involves in controlling the quality of a geodetic network. The research objective is to strategically position control points and minimize errors to improve the overall geodetic network. Finding the optimal design of geodetic network of Ethiopia is the main objective of this thesis by solving the zero order design and first order design problems by applying one of the classical methods that is the trial and error technique using a MATLAB programing language. Zero order design problem was applied to a case study network consists of 30 points and 70 designed distances with a priori deviation equal to 5mm, to determine the best points in the network to consider as control points. The results showed that P18 and P19 having the minimum ellipse of error and considered as control points. These points are therefore chosen as the control points since they have an area of 0.094 and 0.101, respectively, making them the best points. First order design problem was applied on a selected network to be analysed using the objective function, with selected range of movement of 100m to each point in each direction. This first order design problem optimization is done by the trial and error method. By taking P18 and P19 as control points the optimal design of the geodetic network with high precision is developed
Correlating CBR values with basic soil parameters (by using Neuroxl Predictor)
(Addis Ababa University, 2024-01) Airmeyas Aychew; Henok Fikre (PhD)
The California bearing ratio (CBR) is an essential design parameter for soils and an indirect measure of soil strength. This is broadly used for the design of sub-grade, sub-base and base course materials for road, railway, and airfield projects. It is also used as a direct relationship to determine the response of the base or sub-base soil. This research studies the relationship between CBR values and other parameters of soil properties, although the samples include coarse and fine-grained soils, using advanced neural network programs to help us obtain an accurate predicted value. Most of the previous studies were conducted on fine-grained soil types and also used conventional multiple linear regression analysis methods. To satisfy the objective of this study, one hundred and ninety-eight soil sample test results were collected. I participated as a material engineer in the testing and reporting process. Laboratory testing was performed in accordance with AASHTO standard test methods. Modified compaction, soak three-point modified CBR, wet sieve analysis, and Atterberg limit tests were performed on a total of one hundred and ninety-eight soil samples. Statistical analyses were performed to validate the new model using 30 percent of the total sample size. Two types of analysis programs—Microsoft Excel software (ANOVA) for multiple non-linear regression relationships and the advanced NeuroXL predictive neural network program—were used to predict CBR values. Independent soil property parameters were liquid limit, plastic limit, plastic index, amount of particle size less than 0.075 mm, amount of particle size less than 0.425 mm, amount of particle size less than 2.00 mm, amount of particle size less than 4.75 mm, optimum moisture content, and maximum dry density. This study provided two alternative models. The first alternative model included compaction test parameters (OMC and MDD), particle size distribution parameters (4.75 mm PP, 2 mm PP, 0.425 mm PP, and 0.075 mm PP), and plasticity parameters (LL, PL, and PI). They were taken as independent parameters. The second alternative model excludes the compaction test parameters (OMC and MDD) as independent parameters when compared to the alternative one. This study used two alternative analysis techniques: the first group of analysis techniques developed model equations for each classified data set (sub-grade, sub-base, and sub-base), and the second technique developed model equations for the unclassified data set group. The predicted CBR values of both the NeuroXL prediction and multiple nonlinear ANOVA regression models were compared with the actual CBR values, which confirmed that there was an acceptable difference between the actual and predicted CBR values between both analysis methods.
Training Stability of Multi-modal Unsupervised Image-to-Image Translation for Low Image Resolution Quality
(Addis Ababa University, 2023-05) Yonas Desta; Bisrat Derbesa (PhD)
The ultimate objective of the unsupervised image-to-image translation is to find the relationship between two distinct visual domains. The major drawback of this task is several alternative outputs from a single input image. In a Multi-modal unsupervised image-to-image translation model, There exist common latent space representations across images from many domains. The model showed one-to-many mapping and its ability to produce several outputs from a particular image source. One of the challenges with the Multi-modal Unsupervised Image-to-Image Translation model is training instability, which occurs when the model is training using a data set with low-quality images, such as 128x128. During the training instability, the generator loss reduces slowly because the generator is too hard trying to find a new equilibrium. To address this limitation, We propose spectral normalization as a method for weight normalization, which would limit the fitting ability of the network to stabilize the training of the discriminator in networks. The Lipschitz constant was a single hyperparameter that was adjusted. Our experiments used two different datasets. The first dataset contains 5000 images, and we conducted two separate experiments using data set with 5 and 10 epochs. In 5 epochs, our proposed method has achieved overall training loss generator losses reduced by 5.049 % on average and discriminator losses reduced by 2.882 % on average. In addition, in 10 epochs, total training loss generator losses of 5.032% and discriminator losses of 2.864% decreased on average. The second data-set contains 20000 images, and we used datasets with 5 and 10 epochs in two different experiments. Over 5 epochs, our proposed method reduced overall training loss generator losses by 4.745 % on average and discriminator losses by 2.787 % on average. Furthermore, in 10 epochs, the average total training loss was reduced, with generator losses of 3.092 % and discriminator losses of 2.497%. In addition, During the transition, our approach produces output images that are more realistic than multi modal unsupervised imageto- image translation.