School of Electrical & Information Engineering (ETDs)

Permanent URI for this communityhttps://hdl.handle.net/10539/37968

Browse

Subject: search.filters.subject.SDG-17: Partnerships for the goals

Search Results

Now showing 1 - 3 of 3
  • No Thumbnail Available
    Item
    Navigating the Underground: Assessing Vision-Based SLAM Methods in Simulated Subterranean Scenarios
    (University of the Witwatersrand, Johannesburg, 2024) Steenkamp, Dani¨el Johannes; Celik, Turgay
    This dissertation explores the viability of vision-based localization methods in subterranean environments, employing a variety of feature extraction techniques including traditional methods and advanced deep learning approaches. A unique dataset was generated using an autonomous exploration UAV within a simulated subterranean environment. This dataset served as the testing ground for evaluating various feature extraction methods. The ORB-SLAM3 was modified to integrate these methods, adapting its feature extraction module to accommodate alternative approaches while retaining its core pose optimization and backend components. The study includes detailed experiments and analyses of different sensor configurations and feature extraction methods, providing insights into their applicability and performance in subterranean settings.
  • No Thumbnail Available
    Item
    Analysing Test-Driven Development Adherence in Open-Source Projects Using Test-to-Code Traceability Links
    (University of the Witwatersrand, Johannesburg, 2024) Kirui, Gerald Kipruto; Levitt, Stephen
    Several studies have been conducted to determine the impact of Test-Driven Develop- ment (TDD) on software quality. Many studies utilise test-to-code traceability tools and strategies to detect TDD adherence as part of their methodology. However, most test-to-code traceability tools rely on filename-based matching algorithms, which suffer from low recall; therefore, most TDD detection methods are not accurate. This study aims to assess the effectiveness of a statement-based matching algorithm over a filename-based one and whether it can be used to detect the TDD adherence of a software project. The filename-based and statement-based matching algorithms are implemented in Python. To evaluate these algorithms, 500 tests from sixteen Java projects (encompassing frameworks, libraries, and tools for data processing, testing, and web services) are used. These projects range in size from 5,115 lines of code (LoC) to 378,167 LoC. This evaluation helps in understanding the performance of the algorithms through their weighted F1-scores. A mathematical function is created from first principles to correlate method coverage and time deltas with the test-with-development (TWD) adherence of a project. Thereafter, 100 Java projects are used to demonstrate the utility of this function. The results show that the statement-based matching algorithm, with a weighted F1-score of 0.771 and a 95% confidence interval of [0.741, 0.786], is more accurate than the filename-based matching algorithm, with a weighted F1-score of 0.218 and a 95% confidence interval of [0.193, 0.246]. Additionally, the results show that the relative TWD adherence of eleven projects is found to be highly correlated with TDD scores from surveys (rs = 0.723 and p-value = 0.012), and the method coverage of eighteen projects is found to be highly correlated with code coverage obtained from Codecov (rs = 0.744 and p-value = 0.0003). The literature review reveals that no studies have explored the correlation between TDD adherence and the development step size in software projects. An investigation is conducted using a statement-based test-to-code traceability tool and relative TWD adherence values to determine if TDD projects have smaller development steps compared to non-TDD projects. The study finds no significant correlation between median production method churn and relative TWD adherence of projects in all but one out of eight cases.
  • No Thumbnail Available
    Item
    Physical layer of Security with NOMA in mixed RF/FSO systems
    (University of the Witwatersrand, Johannesburg, 2024) Osman, Amani Alged Fetaha; Takawira, Fambirai; Moualeu, Jules
    The ever-increasing demand for spectrum resources, driven by the surge in high data tra c poses a signi cant challenge to the e ciency of wireless communication systems. In response, non-orthogonal multiple access (NOMA) has been proposed as a pivotal technique designed to address the spectral challenges inherent to tra- ditional multiple access schemes. NOMA, through its ability to allow multiple users to share the same time-frequency resources non-orthogonally, represents a promising avenue for improving spectral e ciency (SE). To further improve connectivity and enhance e ciency, NOMA can seamlessly be integrated with multiple-input-multiple-output (MIMO) systems. MIMO, a cor- nerstone technology in modern wireless communication, utilizes multiple antennas at both the transmitter and receiver to exploit spatial diversity, thereby achieving substantial gains in terms of capacity and reliability. The integration of MIMO with NOMA not only addresses the escalating demand for connectivity but also unlocks the potential for massive connectivity and enhanced spectral e ciency in wireless networks. However, the distribution of radio resources within the MIMO- NOMA framework brings forth security challenges owing to the intrinsic charac- teristics of wireless transmissions. In this regards, this thesis tackles the security aspect of MIMO-NOMA systems from an information theoretic viewpoint. The information theoretic security or physical layer security (PLS) e ectively mitigates vulnerabilities stemming from eavesdropping and interception, thereby fortifying the overall security of wireless communication in the present case of MIMO-NOMA systems. Another challenge that requires some attention is ensuring reliability over long distances. Although the radio frequency (RF) spectrum band is recognized for its reliability in diverse weather conditions, its inherent nature renders it sus- ceptible to security concerns. To address the imperative for secure long-distance communications, it is essential to explore innovative solutions that harness the inherent reliability of the RF spectrum while concurrently mitigating its vulner- abilities through enhanced security measures. Integrating technologies such as free-space optical (FSO) communication, acknowledged for its secure transmission capabilities, along with the implementation of advanced security protocols stands as a promising solution to achieving reliability and security in long-distance com- munications. In this work, PLS of a MIMO-NOMA based system in a mixed RF-FSO com- munications is investigated. Additionally, this thesis delves into the injection of vi arti cial noise (AN) in an e ort to explore other features of the proposed sys- tem from a PLS perspective. The last part of the work investigates the secrecy performance of the proposed system in the presence of multiple colluding and non- colluding eavesdroppers. To this end, the statistical properties are examined by deriving the statistical characteristics of the underlying system model, which will subsequently be used to establish both the analytical and asymptotic frameworks for evaluating secrecy performance measures. Finally, Monte Carlo simulations are adopted to validate the accuracy of our proposed mathematical derivations.