Browse
Recent Submissions
Item A Study on bistatic RCS simulations, measurements and calibration(2018) Potgieter, MoniqueBistatic RCS improves detection, characterisation and identification and enhances target information. Its stealthy targets have a low monostatic return and shaping of larger bistatic scattering.Item Bistatic RCS Calculations of Complex Realistic Targets Using Asymptotic Methods(2018) Potgieter, MoniqueBistatic radar cross section (RCS) can improve target detection and identification due to increased target information. Calculating the bistatic scattering of electrically large targets by using asymptotic computational electromagnetic methods is important as these methods provide a suitable alternative to measurements and full-wave simulations. Measurements and full wave simulations can be impractical due to high cost associated with maintenance of the measurement facilities, long lead times and high computational requirements. The accuracy of physical optics and geometrical optics to calculate the bistatic RCS of two electrically large realistic aircraft targets is investigated in this paper. These two methods are compared to the full-wave multi-level fast multipole method.Item Comparison of the Mutual Information Content between the Monostatic and Bistatic Measured RCS Data of a 1:25 Boeing 707 Model(2018) Cilliers, JE; Potgieter, M; Blaauw, C; Odendaal, JW; Joubert, J; Woodbridge, K; Baker, CMany studies have suggested that bistatic radar can improve the classification and recognition of airborne targets due to enhanced target information. This paper makes use of the information theoretic concept of mutual information (MI) and investigates the comparison of the MI content between monostatic and bistatic radar cross section measurements of a 1:25 Boeing 707 scale model. It aims to quantify the increases in the MI content, and hence recognition performance, due to the additional reflected signals available in a bistatic scenario.Item Scatterer extraction and clustering for a DRFM HRR/ISAR resynthesis mode(2022) Potgieter, Monique; Nel, WAJThis presentation reports on (amongst others) the challenge of DRFM HRR profile resynthesis, the process to generate discrete scattering points for the DRFM, a SigmaHat overview, scatterer extraction (3D), and various clustering examples and tools.Item XR-training: A mixed reality platform for accelerated industrial equipment training(2025-07) Moodley, Jayandren; Van Eden, Beatrice; Mphephu, Mutali; Dire, Patrick OMExtended Reality (XR) technologies, including Augmented Reality (AR) and Mixed Reality (MR), are transforming industrial training by enabling immersive, scalable skill development. This paper presents a deployed framework integrating a Content Management System (CMS) for 3D/AR asset sustainability and Azure Communication Services (ACS) for remote collaboration, tailored to address South Africa’s digital resilience challenges. The platform leverages hierarchical asset categorization in the CMS to reduce redundancy, achieving a 90% reduction in duplicate 3D assets, while ACS-enabled remote guidance ensures 95% AR tracking accuracy during equipment training sessions. Evaluation metrics demonstrate a 60% decrease in on-site technical visits, lowering CO2 emissions, and a 40% improvement in trainee competency scores (measured via pre-/post-assessments). While leveraging commercial tools (Microsoft Teams, SharePoint), the framework’s novelty lies in its adaptation to infrastructure constraints (e.g., low-bandwidth optimization) and localized use cases (e.g., digital twin integration for legacy machinery). Deployment outcomes from a pilot with 150 trainees highlight scalability, with 85% reporting reduced dependency on physical trainers. This work aligns with 4IR goals by combining technological feasibility (validated ACS/CMS performance) with socio-economic impact, positioning XR as a catalyst for equitable digital economies.Item Thermocouple signal conditioning using augmented device tables and table look-up neural networks, with validation in J-Thermocouples(2022-01) Maseko, Moses L; Agee, JT; Davidson, IThe relatively high accuracy, large measurement range, and durability of thermocouple devices make these devices to probably be the most-widely used temperature measuring devices in industrial applications. The ability of thermocouples to sense temperature is derived from the generation of thermoelectric voltages arising due to temperature differences between the hot and cold junctions of the thermocouple. Thermocouple temperature measurement processes suffer from inaccuracies arising from both the unwanted or undetected variations in the cold junction temperature of the thermocouple, and nonlinearities in the generated thermoelectric voltage. This paper presents an enhancement of thermocouple temperature measurement using a combination of augmented thermocouple tables generated from thermocouple polynomial functions, look-up MLP neural networks trained to accept the thermocouple output voltage, and the cold or reference junction temperature measurements: to produce improved hot-junction temperature outputs. Experimental validation of the current approach for a J thermocouple, using data from augmented device tables, reproduced the measured temperature values with a worst-case error of 0.0094%.Item Retrieval-Augmented story generation for isiZulu: Enhancing literacy through AI in low-resource contexts(2025) Mkhwanazi, Sthembiso N; Naidoo, Privolin; Moodley, Avashlin; Khumalo, Sandile; Mnisi, Given; Mothomoholo, MamolapiThis study investigates the use of Retrieval-Augmented Story Generation (RASG) to produce culturally relevant and educational children's stories in isiZulu, a low-resource yet widely spoken South African language. To address the scarcity of high quality narrative data, we combined translation-based data augmentation with fine-tuning of multilingual large language models (LLMs), including GPT-4o-mini and LLaMA 3B. A retrieval mechanism was integrated using multilingual-e5-large embeddings, which, despite lacking explicit isiZulu support, enabled contextual story generation from Wikipedia-derived passages. Qualitative evaluations involving native isiZulu speakers revealed that fine-tuned models outperformed baseline systems in terms of grammatical accuracy, coherence, and cultural relevance, though challenges such as language mixing and prompt sensitivity remained. A comparative English baseline using a non-fine-tuned LLaMA 3B model highlighted the performance disparities between high- and low-resource language settings. Our findings underscore the importance of targeted fine-tuning, curated datasets, and embedding models that better represent African languages. This research contributes to the development of AI-driven literacy tools for underrepresented linguistic communities and highlights future directions for improving story generation in low-resource contexts.Item Is local level implementation of the Integrated Geospatial Information Framework (IGIF) in South Africa feasible?(2025) Cooper, Antony K; Makgale, F; Hattingh, MThe United Nations Committee of Experts on Global Geospatial Information Management has introduced the concept of the Integrated Geospatial Information Framework (IGIF) to help overcome problems with accessing and reusing data and integrating various data sets together to improve decision-making, particularly to achieve the Sustainable Development Goals and to address other global challenges. The IGIF aims to address societal and environmental problems from local to global levels, but it appears to be aimed primarily at national governments and regional and international organisations. We explore here how the IGIF could be implemented at the local (municipal) level in a developing country such as South Africa.Item Considerations for a simplified temperature calibration procedure for infrared cameras(2025-07) Chirindo, Mathews; Molekoa, Malinkeng MTemperature calibration of infrared cameras is vital to ensure precise and accurate measurements amidst natural drift due to ageing and other environmental factors. However, conventional calibration is simultaneously constrained by multiple camera settings, such as gain and integration time, in addition to other factors such as emissivity and distance of the object, which impact the pixel value on which the temperature measurement is based. Different sets of camera settings usually require specific calibration tables, leading to a cumbersome calibration process, high computational burden, and huge storage requirements. This paper proposes a simplified temperature calibration process for infrared cameras that allows accurate temperature measurements despite camera settings, using linear regression, interpolation, and curve fitting techniques. Experimental results are presented to demonstrate the successful calibration of two infrared detectors, despite camera settings at an accuracy of 1.9 ° C compared to a calibrated pyrometer. The sensitivity analysis of the two infrared detectors is provided.Item Injecting explicit cross-lingual embeddings into pre-trained multilingual models for code switching detection(2025-12) Sindane, T; Marivate, V; Moodley, AvashlinCode-switching has become the modus operandi of internet communi-cation in many communities, such as South Africans, who are domestically multi-lingual. This phenomenon has made processing textual data increasingly complex due to non-standard ways of writing, spontaneous word replacements, and other challenges. Pre-trained multilingual models have shown elevated text processing capabilities in various similar downstream tasks such as language identification, dialect detection, and language family discrimination. In this study, we exten-sively investigate the use of pre-trained multilingual models - AfroXLMR, and Serengeti for code-switching detection on five South African languages: Sesotho, Setswana, IsiZulu, IsiXhosa, and English, with English used interchangeably with the other four languages, including various transfer learning settings. Addition-ally, we explore the modeling of known switching pairs within a dataset through explicit cross-lingual embeddings extracted using projection models: VecMap, Muse, and Canonical Correlation Analyses (CCA). The resulting cross-lingual embeddings are used to replace the embedding layer of a pre-trained multilingual model without additional training. Concretely, our results show that performance gains can be realized (from 59.1% monolingual to 74.1% cross-lingual, and to 90.8% multi-lingual) by closing the representational gap between the languages of the code-switched dataset with known codes, using cross-lingual representations. Moreover, expanding code-switched datasets with datasets of closely related lan-guages improves code-switching classification, especially in cases with minimal training examples.Item Using the isiZulu GF Resource Grammar for morphological annotation(2025-11) Marais, Laurette; Pretorius, LThe isiZulu GF Resource Grammar (ZRG) enables syntactic parsing using the GF runtime system. In order to perform this task, the ZRG implicitly encodes rich morphosyntactic information about isiZulu. In this paper we show how such information can be made explicit by adapting the way the grammar linearises GF abstract syntax trees. The result is annotated text, which can be utilised in various ways for supporting natural language processing of an under-resourced, morphologically complex language like isiZulu.Item A survey of logistics skills in South Africa: Identifying logisticians of the future(2025-10) Dondofema, Richmore ASouth Africa faces major structural challenges, including an unemployment rate of 27%, which rises to 55% among youth. Although Africa's most advanced economy, it has underperformed compared to other middle-income countries, leading to a critical skills shortage, particularly in logistics. Results of this study show that while there are adequate traditional logistics skills, there is an increasing need for technologically adept logisticians due to rapid advancements, requiring expertise in data analytics, telecommunications, and environmental management. Future logisticians must understand value creation, identify market opportunities, and possess both technical and soft skills to succeed in the evolving logistics landscape.Item Intersection safety: A Tshwane case study(2025-07) Miyambu, M; Marole, Busisiwe C; Malope, Retang K; Kgoa, Lerato; Mashaba, Hasane P; Malima, Thendo; Venter, KarienThe Safe System Approach (SSA) advocates for a forgiving and self-explaining roads and covers aspects such as safer vehicles, safer speeds, safer roads, post-crash care and safer people. The principles that SSA incorporates are zero tolerance for death and serious injuries, human error is inevitable, human vulnerability, shared responsibility, proactive interventions for safety, and redundancy of the transport system. This case study considers road safety aspects at a specific intersection in the City of Tshwane. The study made use of observations and traffic conflict modelling tools to assess intersection safety. Effective planning of intersections is essential in transportation and road safety engineering. Intersections are critical junctions in the road network where various traffic flows converge and where the potential for traffic conflicts is higher. The intersection design significantly influences safety, traffic movement, and the overall efficiency of the road system. It has been observed that when vehicles approach the full access intersection (T-junction) to make a right turn, drivers experience restricted visibility of oncoming traffic from the right due to parked vehicles along the roadside. This configuration forces drivers to advance cautiously into the oncoming traffic lanes to gain a clearer line of sight, leading to unsafe turning manoeuvres and traffic conflicts. Additionally, these vehicles are required to queue within the intersection between the traffic islands, while waiting for a safe gap from the oncoming traffic from the left to complete their turn and merge onto the lane after making the turn. This operational challenge raises concerns regarding the potential need for a traffic signal or a general intersection upgrade at the intersection, or similar intersections with comparable safety and flow issues. This paper concludes with recommendations to address safety concerns at this intersection type aligned with the Safe System Approach.Item Bridging digital health literacy gaps in South Africa’s community health centers: A scoping review on the role and capacity of community health workers(2025-11) Mocuminyana, Lerato L; Maremi, Keneilwe JDigital health technologies are crucial for improving healthcare access, patient engagement, and health outcomes. However, their success depends on the digital health literacy of users, which remains a significant barrier in South Africa. Community Health Workers (CHWs), as part of Ward-Based Primary Healthcare Outreach Teams (WBPHCOTs), play a crucial role in bridging these gaps by serving as intermediaries between formal health systems and communities. This paper presents a scoping review to examine the role of CHWs in improving digital health literacy and supporting the effective implementation of digital health interventions within Community Health Centers (CHCs). Findings reveal that CHWs face challenges, including limited digital skills, poor connectivity, system complexity, and resistance to digital adoption. Despite these barriers, they actively mitigate challenges through peer mentoring, patient education, and adaptive workflows. Drawing on international case studies, the review highlights strategies such as continuous digital literacy training, culturally adapted tools, mentorship programs, and sustained technical support to enhance CHW effectiveness. Strengthening CHW capacity can close the digital divide, improve data quality, and ensure equitable healthcare delivery. This paper recommends targeted investments in CHWs and digital infrastructure to optimise South Africa’s digital health ecosystem and improve primary healthcare outcomes.Item A web scraping approach towards cryptocurrency investigations(2025-06) Mawhayi, B; Botha, Johannes G; Leenen, LThe investigation of cryptocurrency crimes is still in its infancy with no standardised process or methodology to follow. This paper describes research that forms part of a broader project led by the second author (Botha, et al., 2025). The broader project’s aim is to develop a methodology to follow when conducting cryptocurrency crime investigations. One of the steps in the proposed methodology is web scraping. The authors of this paper present a detailed exploration of web scraping techniques within the broader context of the proposed investigation methodology. In this paper, the focus is on developing a well-structured methodology for scraping social media platforms and online forums to gather data related to fraudulent activities; the goal is to find posts that include references to the wallet address of interest. This exploration uses an iterative approach; for every new cryptocurrency wallet address discovered or revealed through on-chain analysis, a parallel path is followed by scraping the Internet. If a mention of the cryptocurrency address should be discovered it is considered to be a key finding, creating a pivot point in the investigation. From a pivot point, further open-source intelligence (OSINT) techniques will be applied, though this aspect falls beyond the scope of this paper. If no relevant information or link is found, the scraping path will not be pursued, and the investigation proceeds with on-chain analysis to identify additional wallet addresses. Additionally, challenges encountered in web scraping, such as handling platform restrictions, ensuring data accuracy, and managing large volumes of data, are addressed. The goal of the proposed methodology is to enhance data extraction and analysis efficiency contributing to the proposed methodology for investigating cryptocurrency scams.Item Enhancing digital wallet security: A systematic comparison of passwordless and risk-based authentication approaches(2025-12) Mthethwa, Sthembile N; Ndhlovu, Nomalisa; Myaka, Zanele S; Ntshangase, Sthembile N; Shadung, Lesiba D; Singano, ZothileThe dynamic nature of the digital landscape necessitates robust security measures for the use of digital wallets, with authentication being pivotal in ensuring both user safety and system integrity. Traditionally, password-based authentication has been the predominant method employed. However, it is also the primary target for cyber attackers, with numerous successful breaches resulting from compromised credentials. Despite the availability of alternative methods, passwords continue to be the preferred choice. This paper examines various authentication techniques—such as passwordless, behavioral, continuous, and adaptive authentication—emphasizing their respective advantages and disadvantages. Additionally, it discusses the challenges associated with the implementation of these methods and outlines key considerations for organisations prior to adoption. Specifically for digital wallets, passwordless and risk-based authentication methods are identified as the most appropriate. Future research will focus on testing and comparing the effectiveness of these two methods by simulating attacks to determine which provides greater security.Item Evaluation of guidance, navigation, and control algorithms for hydrogen-powered multi-aircraft systems(2025-09) Makhubo, Mamokete; Ndebele, Bright B; Ragimana, PhumudzoThis paper presents a modular digital-twin framework to compare inner-loop attitude controllers: geometric PD (“PID”), discrete Linear-Quadratic Regulator (LQR), and a movepenalized linear Model Predictive Controller (MPC), for a heavy-lift T30-class quadcopter intended for hydrogen propulsion study. The twin couples 6-DoF rigid-body dynamics, actuator mixing, motor/ESC lag, a bus-level electrical model, and stochastic wind (OrnsteinUhlenbeck) with look-ahead guidance on a sharp lawnmower survey. To isolate controller effects, task difficulty is equalized by autotuning a single scalar so that the achieved cross-track Root-Mean-Squared Error (RMSE) lies in a 2.6 ± 0.25 m band. The tuned controllers then run identical 600 s simulations under the same wind seed and retimed speeds. On the equalized run, PID and LQR achieve 2.45 m and 2.55 m RMSE, respectively, while MPC settles at 3.13 m due to its move penalty and finite horizon. All three deliver survey-class performance with mean bus power ≈ 4.8 kW and peaks in the 7–8 kW range, but MPC reduces energy per meter by approximately 2.3% at the cost of relaxed lateral accuracy. A 10-seed Monte Carlo confirms this trade: PID/LQR remain in-band for 90%/80% of seeds, while MPC consistently lowers energy per meter with similar mean power but gentler peaks. For hydrogen-electric UAVs, these metrics map directly to propulsion co-design, where energy per meter informs hydrogen mass and range, peak power sets stack/buffer sizing, and actuator smoothness affects balanceof-plant transients. The results show that controller selection is not only a matter of tracking accuracy but also an energy-management lever: PID/LQR suit survey tolerance, while MPCstyle penalization favours endurance and balance-of-plant stability.Item Top crypto scams in 2022-2024: Analysing trends, tactics, and regulatory responses(2025-05) Botha, Johannes G; Luoma-Aho, V , V Leenen; Leenen, LThe digital society has proven to be vulnerable in different ways, and many of these vulnerabilities manifest only once something intangible or tangible is misused or destroyed. One such vulnerable area is the financial markets, especially in the realm of cryptocurrencies. These rapidly growing cryptocurrency markets have in recent years not only attracted a wave of new investors but have also created fertile ground for fraudulent activities [19]. Many individuals in this new terrain find both valid and false information on cryptocurrencies, making it challenging to verify facts. Further, as generative Artificial Intelligence continues to improve, microtargeting and personalisation of digital content become easily accessible to the masses. As digital assets gain mainstream acceptance, the allure of high returns has drawn many individuals into a complex and often opaque financial landscape. Unfortunately, this environment has also given rise to an alarming increase in cryptocurrency scams, which exploit investor naivety and market volatility.Item Integrating Sentinel-1 and Sentinel-2 derived parameters for assessing wetland carbon in the Grassland Biome of South Africa(2025-09) Ngebe, S; Naidoo, L; Van Deventer, Heidi; Tsele, PAccurate estimation of above-ground biomass (AGB) and teal carbon stocks in wetlands remains a challenge due to seasonal variability and the complex interactions between vegetation structure, soil moisture, and hydrological dynamics. Previous studies have demonstrated the value of Sentinel-1 and -2 predictors for seasonal modelling of AGB in various wetland types but remains poorly quantified in African palustrine systems. This study assessed the potential of combining Sentinel-1 and -2 data with Random Forest (RF) modelling to estimate AGB and associated carbon stocks in a palustrine wetland in South Africa. Five modelling scenarios which included the fusion of optical and radar derived parameters were constructed for both summer and winter using RF %IncMSE-based important variable selection. Results demonstrated that combined Sentinel-1 and -2 images and RF models achieved consistently high predictive performance, with R2 > 0.9; where summer (R2 = 0.941, Root Mean Square Error [RMSE] = 18.29 g m-2, relative Root mean square error [relRMSE] = 14.95%) and winter (R2 = 0.928, RMSE = 29.79 g m-2, relRMSE = 16.42%). These findings highlight the robustness of multi-sensor approaches for seasonal wetland AGB estimation and provide a valuable foundation for monitoring their seasonal dynamics for improved carbon accounting strategies.Item Development of a quantum-safe public key infrastructure for South Africa using qualitative method(2025-07) Ntshangase, Cynthia S; Baruni, Kedimotse P; Lefophane, SamuelThis research analyses public key infrastructure frameworks from various countries to identify primary components and specifications, aiding South Africa in establishing a secure infrastructure. It outlines best practices and security considerations to be adapted, influenced by technological advancements. PKI framework development for the top seven countries with good cybersecurity posture was reviewed, namely, United States, United Kingdom, Saudi Arabia, Estonia, South Korea, Spain, and Singapore. The outcome of the study indicates that the main components of the PKI to be considered before, during and after development are policies, people involved, procedures, software, hardware, and services. The study improves knowledge for officials and policymakers managing national infrastructure.