Browse
Recent Submissions
Item Assessment of modern training innovations for supervisors and trainers in the South African Mining Sector(2025-06) Van Schoor, Abraham M; De Kock, N; Khan, Sumaya; Müller, R; Van Rensburg, R; Govindasamy, K; Botha, W; Maphalala, Busisiwe V; Mpofu, Mvikel; Pelders, Jodi L; Ramparsad, SAs the mining industry modernises, skills development needs to be a priority. The aim of this paper was to develop guidelines that consider the assessment of modern training solutions for supervisors and trainers in modern mining. A literature review was conducted on best-practice criteria for the evaluation of modern mining upskilling and reskilling solutions. A draft evaluation matrix was developed based on the literature review insights and incorporated 48 best-practice criteria for the assessment of training solutions. The assessment instrument was applied to training curriculums for supervisors and trainers of two participant entities. Data gathering included assessments of the training solutions, and an industry panel review process. Strengths, weaknesses, opportunities, threats, and gap analyses were undertaken. Key insights were revealed for the respective training solutions. Recommendations included continuous review and improvement of the curriculum for alignment to mining modernisation skills needs, including consideration of modern training methodologies and facilitation; revised content and assessments; skills and training required for modernisation; tracking of graduates and learner feedback; better alignment with modernisation objectives and industry skills needs, increased focus on safety and risk assessment and control; and more immersive learning experiences. While sample training innovations were selected for evaluation, the recommendations remain relevant for training entities looking to align to their curriculum to mining modernisation skills needs and industry skills demands for modernisation.Item Exploratory analysis of modified deep learning model for potholes data augmentation(2025-04) Adebiyi, RF; Bello-Salau, H; Onumanyi, Adeiza J; Adebiyi, BH; Adekale, AD; Bello-Salahuddeen, RA major factor contributing factor resulting to a large proportion of vehicular-related traffic accidents in developing nations is the poor condition of road networks, characterized by potholes, bumps, and other anomalies. Despite efforts by authorities to address these issues, they persist. A new approach involves equipping vehicles with sensors to detect road anomalies, enabling drivers to make informed decisions. Various models using road surface images to detect and classify these anomalies have been proposed, with recent methods leveraging deep learning. The effectiveness of these models depends on the presence of abundant and well-labelled training datasets. To address this need, a modified Deep Denoising Diffusion Probabilistic Model (mDDPM) was proposed, enhancing the U-Net backbone architecture to improve the original DDPM's performance in augmenting pothole images. The mDDPM generates more diverse augmented images, evaluated through subjective and objective assessments, including the Fréchet Inception Distance (FID) score. Experimental results showed that 98% of participants could not distinguish between real and synthetic images, classifying the augmented images as real. Additionally, an FID score of 0.52 indicated that the augmented images closely resemble real pothole images. This demonstrates the model's effectiveness in generating training data for deep learning models aimed at road anomaly detection and classification, contributing to the development of robust models for detecting and classifying potholes and other road anomalies.Item Requirements management of the Auckland City Rail Link Project - An update(2025-08) Dennehy, S; McMullan, R; Meyer, Isabella AThe Auckland City Rail Link (CRL) project was the first major transport infrastructure project in New Zealand to adopt a requirements-led approach. This paper reflects on lessons learned, with the aim of identifying learnings that may inform approaches in other complex, real-world transport infrastructure projects. The recommendations do not cover the full scope of requirements management or systems engineering, rather they are selected aspects that presented particular challenges or successes, including project planning, tool choice and configuration, establishing the requirements management database as the single, trusted source of truth, change control, stage gates, levels of abstraction, interface management, reliability/availability/maintainability/safety, and post-construction verification. This work is limited to the analysis of a single case, with recommendations made from the perspective of the requirements management team rather than from a wider range of disciplines. Notwithstanding these limitations, the insights and recommendations should be considered for future infrastructure projects, in conjunction with experience and lessons learned from other projects, and with appropriate tailoring for project context – which is always unique.Item Correlating defects in electroluminescence images to photovoltaic module power loss using DeepLabV3 for semantic segmentation(2025-07) Pratt, Lawrence E; May, Siyasanga I; Mkasi, Hlaluku WThis article investigates a multi-class semantic segmentation model for correlating defects in electroluminescence (EL) images and PV module power degradation in fielded modules. The case study uses machine learning and statistical methods to identify a potential root cause for power degradation in 680 PV modules sampled from a multi-MW PV plant in South Africa. The EL image and the electrical performance were measured and recorded for each module at the CSIR PV Module Quality and Reliability Lab. A deep-learning model previously trained for multi-class defect detection in EL images of solar cells was re-trained to include 24 samples of a new ‘brightening’ defect. The updated model was used to classify each pixel on 48,960 solar cells into one of 29 classes. Based on an exponential decay model, the degree of ‘brightening’ defect averaged across all cells in a module correlated to the module fill factor (r2 = 0.42), indicating that 42% of the variability in Fill Factor (FF) measured for this set of modules could be explained by the brightening defect. While the strength of the correlation is moderate, it may provide some insight into the root cause of the degradation in FF. The authors speculate that the brightening defect is a signature of solar cells with non-uniform current distribution resulting from ageing. Updates to the multi-class prediction model were relatively simple and fast, enabling a similar analysis for new defects in PV modules as they emerge.Item Exploring the role of dew computing in enhancing cybersecurity and digital forensics: A systematic literature review(2025-06) Nelufule, NthatheniThe Fourth Industrial Revolution (4IR) has precipitated the emergence of sophisticated cyberthreats, which evolve with time. The evolution of these threats demands innovative solutions applied in cybersecurity and digital forensics to improve data security and incidence response. This paper presents a systematic literature survey conducted on dew computing architectures, which integrates local computing capabilities with cloud resources to create a hybrid framework that enhances data accessibility, data security, and digital forensic analysis. By allowing devices to operate independently of constant internet connectivity, dew computing addresses significant challenges faced by traditional cloud computing architectures, such as latency and data vulnerability during transmission. The main objective of this paper was to study how dew computing architectures can enhance real-time data processing, data protection, data integrity, and enhance digital evidence acquisition. Furthermore, this paper discusses the challenges and limitations associated with implementing dew computing, including technical barriers and privacy concerns. The research survey findings suggest that dew computing offers a promising approach to mitigating cybersecurity risks and provides digital forensic investigators with the tools necessary for effective investigations.Item Towards a circular economy in mobile communications technology: A systematic review(2025) Ebrahim, Rozeena; Burger, Chris R; Masonta, Moshe T; Sikrenya, Siyanda KS; Hlatshwayo, Ronald OThis paper presents the findings of a systematic literature review examining the relationship between the circular economy (CE) and terrestrial mobile networks. This review identifies key approaches to address a CE from 14 highquality academic papers and classifies them according to four categories: infrastructure reduction and hardware optimisation; reduced energy demand; reduced reliance on fossil fuel-based power grids; and recycling. These approaches were also mapped to the “3R” framework which reveals a strong emphasis on reducing resource consumption and reusing resources, with minimal attention to recycling. The study emphasises the need for a holistic CE approach addressing all stages of the product lifecycle and calls for expanded research and development efforts, particularly in developing economies. Collaborative initiatives between academia, industry, and policymakers are recommended to promote sustainable and comprehensive CE practices in mobile telecommunications.Item Power Consumption Analysis of a 5G NR Base Transceiver Station Using a Custom Measurement Setup(2025-07) Ebrahim, Rozeena; Vilakazi, Mlamuli C; Mabunda, Ntsako D; Makaleng, Koketso F; Sikrenya, Siyanda KS; Lysko, Albert A; Mfupe, Luzango PAs Fifth Generation (5G) and future mobile networks continue to expand, understanding the power consumption of the base transceiver station (BTS) is necessary for improving the energy efficiency of the Radio Access Network (RAN). The power consumption of the BTS is influenced by factors such as its operating mode, transmission power and traffic levels. This work has explored the power consumption of an outdoor commercial 5G New Radio (NR) BTS using an inexpensive and custom-built power measurement setup. Indoor testing was done at a reduced transmit power level to analyse the power consumption under different operation modes. The measurements presented a clear correlation between activity levels and energy usage. Insights have been provided into the power consumption requirements of outdoor BTSs compared to 5G setups deployed on an in-house, indoor mobile network testbed, including the trade-off between energy efficiency and operational stability. Further, these findings have highlighted areas for future research, such as the impact of deployment environments, transmit power levels and the number of connected devices on energy efficiency. Understanding these factors is considered essential for designing, building and deploying more sustainable mobile networks for the future.Item A satellite-based decision support service for the marine fisheries and aquaculture industry of southern Africa(2025-09) Smith, Marié E; Molapo, Nkadimeng R; Ngulube, Mabuela; Sibolla, Bolelang, H; Vhengani, Lufuno MMarine Aquaculture in South Africa includes abalone, mussels, oysters, and finfish. Depending on the cultured organisms these could include land-based pump-ashore operations in-water cages or rafts. Each method of operation has different environmental risk.Item Future trends in AI for cybersecurity and digital forensics: A systematic literature survey(2025-06) Nelufule, Nthatheni NThe Fourth Industrial Revolution has brought many opportunities, including the integration of artificial intelligence technologies into cybersecurity and digital forensics. These integrations represent a transformative change in how organizations protect their digital assets and investigate their cybersecurity incidents. As cyber threats become increasingly sophisticated, traditional cybersecurity measures often fail, necessitating the adoption of advanced AIdriven solutions. This paper presents a systematic literature survey that explored future trends in artificial intelligence applications in these critical domains, focusing on their potential to improve threat detection, automate incident response, and improve the efficiency of forensic investigations. The survey has identified some key challenges associated with the deployment of Artificial Intelligence technologies, including ethical considerations, data privacy issues, and the complexities of integration into existing systems. The findings from this survey paper have revealed a growing reliance on artificial intelligence for real-time threat detection and response, highlighting its effectiveness in identifying anomalies and predicting potential breaches before they escalate into significant incidents.Item Competency-based training for mine emergency response(2025-06) Lange, Pieter; Bergh, Adriaan V; Pelders, Jodi L; Khan, SumayaMine workers are exposed to hazards that can cause injuries or fatalities, including fires, underground explosions, irrespirable, falls of ground and mobile machinery. Emergency preparedness is important for improved safety outcomes and includes the deployment of self-contained self-rescuer (SCSR devices, effective escape routes, and adequately located refuge bays. The need has been identified for improved training solutions or mine worker escape situations, which should provide some exposure to the stressors that would be experienced. The CSIR mining Cluster has developed innovative multimodal competency-based training modules to improve the emergency response of mine employess. The modules include interactive e-learning, virtual reality training, SCSR donning and breathing simulation, and competency-based assessments. A pilot study was successfully completed with participation from a major coal mining operation. The competency-based approach improves the overall efficiency of the training and is especially well-suited to training for high-consequence, low-frequency scenarios.Item Understanding the risk: Mapping deepfake cyberattacks to a temporal attack model(2025-07) Pieterse, HeloiseThe advancement of artificial intelligence (AI) technologies has become a trending topic in the cybersecurity domain. These technologies, however, present cybersecurity with a double-edged sword as AI offers enhanced threat detection and protection, but also enables cybercriminals to craft sophisticated cyberattacks. Deepfakes, which are a form of digitally manipulated synthesised media created using deep learning techniques, have garnered widespread attention due to the use of deepfakes in cyberattacks to cause influence, spread disinformation, or conduct fraudulent activities. While extensive research efforts have been undertaken to develop defences against deepfakes, the solutions are technical and not easily accessible. Innovative strategies are required to equip personnel from government, academia, and the business sector with the fundamental knowledge to detect and defend against cyberattacks employing deepfake technology. This paper evaluates the most significant events involving deepfakes since the emergence of the technology in November 2017. Key trends and characteristics are identified and mapped to a temporal attack model to separate the different stages of a cyberattack involving deepfakes. The outcome is a Deepfake Attack Framework that offers valuable insights essential to understanding the risks associated with deepfakes. The Deepfake Attack Framework presents a theoretical solution that can be applied practically to minimise risk and enable personnel to be better prepared to defend against deepfake-driven cyberattacks.Item A case for high capacity coal trucks to reduce costs and emissions at Eskom(2021-07) De Saxe, Christopher; Van Eeden, J; Steenkamp, A; Mokone, OlwethuSouth Africa’s national power utility, Eskom, is under heavy strain to maintain an undisrupted electricity supply and contain costs, while at the same time reducing its environmental impact. In 2018/19, Eskom acquired 118 Mt of coal, at a purchase cost of approximately R 47 billion, of which around R7 billion (15%) can be attributed to the transport of coal via conveyor, rail and road. Eskom has been unable to meet its road-torail modal shift targets, and so road haulage still accounts for around 30% of coal deliveries. The “Smart Truck” or “PBS” demonstration project in South Africa has shown how an innovative approach to truck design and regulation can drastically improve the efficiency of road haulage, reducing the cost per tonne-km, while reducing emissions and improving safety. An existing Smart Truck trial in coal transport has demonstrated a 15% reduction in fuel use and associated carbon emissions per tonne-km, which translates into an approximate 6% reduction in total road transport costs. This was achieved through the introduction of innovative 74-tonne tridem interlink truck combinations, which has resulted in fewer truck trips and reduced costs for the same haulage task. At the same time, the trucks are more road friendly due to additional axles and fewer truck trips, and the trucks are designed to be inherently safer than the conventional coal interlinks currently in use. In this paper, we benchmark the costs and emissions of Eskom’s current road haulage coal supply operations in South Africa, and calculate the potential savings from migrating to 74tonne interlink PBS truck combinations. We demonstrate potential savings of R 120 million and 35 000 tonnes of CO2 per year, while removing 300 000 truck trips from the roads.Item Classification of trucks using camera data(2021-07) Mokone, Olwethu; De Saxe, ChristopherUnderstanding the precise movements of different commodities on South African roads can help in not only describing the logistics sector more accurately, but also in the planning of road infrastructure maintenance and investment. Truck combinations can be classified into several classes broadly associated with different commodity groups, including tautliners, tankers, flatbeds (general freight) and flatbed (containerised freight). Current truck classification systems in South Africa can classify trucks by number of axles and vehicle mass but are unable to determine the combination type and hence commodity group. Video data allows for truck combinations to be classified in more detail using image-based classifiers. The latest developments in deep learning algorithms have made it possible for accurate classification of vehicle types using camera data. A CCTV camera feed of a section of the N3 was provided by the South African National Roads Agency Limited (SANRAL) and was used as a case study to develop a proof-of-concept classifier for tautliner and tanker truck combinations, using a transfer learning approach and the pretrained ResNet50 classifier. The results indicate good accuracy based on relatively small datasets. Future work will focus on further optimisation and investigating the training dataset requirements in more detail.Item Accelerated Pavement Testing (APT) of a test section surfaced with an asphalt wearing coarse containing plastic waste incorporated using the ‘wet method’(2025-07) Smit, Michelle A; Rust, FC; Mturi, Goerges; Mokoena, Refiloe; Ntombela, R; Marais, HermanThe incorporation of plastic waste in road pavement materials presents a promising opportunity for sustainable infrastructure development. In South Africa, introducing any innovation requires compliance with national performance criteria and guidance from mechanistic-empirical design methods. This study evaluated the rutting resistance performance of a road pavement section surfaced with plastic waste modified asphalt (PWMA) produced via the wet method – where plastic waste is integrated into the bituminous binder before mixing. An Accelerated Pavement Testing (APT) program was adopted for the permanent deformation testing of a coarse continuously-graded asphalt wearing course modified with plastic waste. The PWMA was produced with post-consumer recycled plastic waste and also incorporated a Reactive Elastomeric Terpolymer (RET). Test sections were constructed in Gauteng, South Africa, comprising a reference asphalt (based on a standard unmodified bitumen used in South Africa) structure and a PWMA layer. Both sections were subjected to simulated traffic loading using a Heavy Vehicle Simulator (HVS) at speeds of 12km/h, varying wheel loads (40, 60 and 80 kN dual wheel load), and controlled temperatures reflective of local pavement conditions. Performance monitoring involved surface and embedded measurement tools, including Road Surface Deflectometer (RSD), Multi Depth Deflectometer (MDD), standard straight edge, thermocouples and temperature buttons. After 2.9 million equivalent standard axles (ESALs) of HVS loading, the PWMA section demonstrated enhanced rutting resistance, with an average rut depth of 7.2 mm, compared to 10.4 mm for the reference section, which reached a maximum rut of 12mm.These results align with laboratory findings, confirming that the addition of plastic waste increases the structural integrity of asphalt layers by enhancing resistance to permanent deformation. This study supports the potential for adopting PWMA in South African road infrastructure to meet national performance standards and sustainability goals.Item Application of 2D and 3D SAR backprojection techniques for intrawall and Throughwall Object Detection(2025-05) Dass, Reevelen; Nel, Willem AJ; Paine, SA radar system has been developed to image the interior structures within walls, revealing details such as conduits, piping, and other substructures. This is achieved by moving an antenna system across the wall surface to generate synthetic aperture radar (SAR) images using frequency-domain backprojection. The materials imaged included a copper pipe, a PVC pipe, a wooden beam, and a calibration target. Results showed that all targets, except for the PVC pipe, were clearly visible to the untrained eye. To enhance target isolation from the wall background, background subtraction and differential SAR (DSAR) techniques were applied, both of which effectively highlighted the targets against the wall background. Image contrast was then evaluated as a quantifiable metric of image quality, which is critical for optimising post-processing techniques such as autofocus. Finally, a 3D image of the wall’s interior was generated, demonstrating the feasibility of a SAR-based wall scanner and its potential in non-destructive inspection and security applications.Item Sticking (with) the Landing - A modern case for Knights Landing in resource-constrained environments(2025-07) Johnston, Bryan J; Crosby, Charles P; Reynolds, QG; Schopf, JM; Whalen, CJThe HPC Ecosystems Project has repurposed decommissioned tier1 HPC systems into entry-level clusters across Africa for over a decade. Stampede2 Knights Landing (KNL) systems are available for global distribution through the Texas Advanced Computing Center’s Legacy Computing Program, and to HPC Ecosystems Project partners. To ensure the novel KNL architecture is fit for purpose for sites contemplating adopting the legacy systems, this publication provides a brief performance reference guide for prospective adopters. Benchmark tests were conducted to evaluate Stampede2’s KNL processors on modern workloads to help inform prospective adoption decisions. It was concluded that the Stampede2 KNL processors remain particularly suitable for applications that benefit from good memory bandwidth, but that multi-node use is only feasible if high-performance networking is also available.Item Evaluating offline LoRaWAN for surface-level maritime communication(2025-07) Manga, Amisha; Giesler, Achmed; Govindsamy, Reesen; Mowzer, Mohammed YBThis research investigates the feasibility of using LoRaWAN (Long Range Wide Area Network) for sea surface- level maritime communication. Sea-based field experiments were conducted across False Bay, South Africa, with the intention to evaluate line-of-sight communication. A LoRaWAN gateway was configured as a private network server to process incoming messages that contain geolocation information from commer cial off-the-shelf (COTS) LoRa (Long Range) devices. Custom software was developed to extract, record and display key performance metrics such as the Signal to Noise Ratio (SNR), Received Signal Strength Indicator (RSSI), frequency bands and spreading factor. The devices were deployed on the sea surface and achieved a stable 2 km communication range in rough sea conditions with 2 m swells. Communication was achieved at 5 km when the devices were lifted 0.5 to 1 m above the sea surface, indicating improved performance with elevation. The results support LoRaWAN’s potential for maritime communication with applications in emergency response and defence, while highlighting the challenges and key insights for future research.Item BlockBaRT: Towards a Blockchain-based battery passport with selective disclosure of verifiable data(2025-07) Kanjere, Julian S; Ford, Merryl; Louw, Jakobus MThe market for electric vehicles (EVs) is surging as automakers move towards decarbonisation, and consumer preferences towards environmentally friendly vehicles. In addition, technological advances are making EV production and operation cheaper, and EV performance comparable with that of traditional internal combustion engines. One of the core components of an EV is the lithium-ion battery, which has a multi-stakeholder and often siloed value chain. Given the nature of the EV battery supply chain, visibility into the end-to-end lifecycle of a battery is opaque. To address this limited visibility and the automotive decarbonisation agenda, governments are introducing battery passport legislation for the collection and sharing of EV battery-related data, throughout its lifetime, among supply chain actors.Item Evaluating trust models for the IoT-enabled peer to-peer energy market(2025-07) Leotlela, B; Ledwaba, Lehlogonolo PI; Coetzee, MThe decentralised nature of peer-to-peer (P2P) energy markets creates an environment where trust is difficult to establish. Supported by Internet of Things (IoT) devices, trust and security challenges arise, due to the absence of central oversight and the risk of uncooperative participant behaviour. A comparative analysis of existing trust management models and schemes is conducted in this study comparing how they are designed to encourage cooperation and ensure reliable interactions in decentralized energy markets. Emphasis is placed on how these models integrate security mechanisms to build trust and on their computational efficiency for IoT constrained environments. The analysis highlights the trade offs between trust management and IoT device performance, identifying limitations in scalability and latency as participation scales. Results indicate that while several models effectively build trust and promote cooperation, many impose significant resource demands, underscoring the need for balance between trust assurance and operational efficiency. This work provides a comprehensive evaluation of trust mechanisms in transactive energy systems and offers insights into their practical viability in resource-constrained, decentralized environments.Item Comparative evolution of production ready programmable data plane hardware(2025-07) Mnyandu, Wandile T; Makondo, Ntshuxeko; Kobo, Hlabishi ISoftware Defined Networking (SDN) has revolutionised the architecture of the traditional network device by centralizing its control and management planes, leaving it with only the data plane for packet forwarding. While this architecture improves network programmability and management, it introduces latency and is rigid in packet processing. To address these limitations, the Programming Protocol-independent Packet Processors (P4) language allows for direct programming of the data plane, enabling the creation of custom packet processing pipelines (PPP) using programmable hardware like ASICs and FPGAs. This paper provides a comprehensive review and comparative analysis of production-ready P4 solutions and outlines the key milestones in the development of programmable data planes as research publications typically cover their origins in relation to SDN while they predate it. Furthermore, we discuss their practical applications and emerging trends in programmable data plane hardware. The findings aim to guide researchers and practitioners in evaluating suitable P4-enabled technologies for diverse network environments.