Joyce Mwangama

Prof. Joyce Mwangama

orcid

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 August 2025 in Journal of Robotic Surgery
Verified
Journal Article
Open access
PDF file
MCMaurine Chepkoech
Maurine Chepkoech
BMBessie Malila
Bessie Malila
JMJoyce Mwangama
Joyce Mwangama
Telementoring in surgical training enables expert surgeons to provide real-time remote guidance to trainees. This technique is increasingly adopted to address surgical training gaps in low- and middle-income countries (LMICs), i.e., nations with a gross national income per capita of $13,845 or less. However, existing systems are constrained by conventional communication systems, which experience high latency, limited bandwidth, and poor video resolution. These challenges hinder the integration of immersive and interactive technologies such as virtual reality (VR), augmented reality (AR), and haptic feedback, which would enhance learning outcomes, especially in resource-constrained environments. AR overlays digital images onto the real-world environment, whereas VR immerses users in a fully computer-generated environment, enabling trainees to interact with surgical components as if they are in the real world. In addition, the advent of the fifth generation of mobile networks (5G), which delivers ultra-low latency, high bandwidth, and support for network slicing, offers a promising foundation for scaling high-fidelity telementoring systems. Moreover, the integration of artificial intelligence (AI), i.e., equipping computer systems with the ability to perform tasks that would typically require human intelligence, can enable real-time performance analytics and skill assessment of the trainees. In regions lacking reliable network backbone infrastructure, such as fiber, hybrid approaches that combine low-cost 5G deployments with satellite communication can be leveraged to achieve reliable end-to-end connectivity. Therefore, this systematic literature review evaluates current surgical telementoring systems, their enabling technologies, and associated challenges, with emphasis on LMIC contexts, where such systems will have the greatest benefits. Through a structured Population, Intervention, Comparison, and Outcome (PICO)-based synthesis, we address five key research questions that span key aspects such as current telementoring systems and technologies, functional and technical requirements, educational outcome effectiveness, and the gaps addressed by modern technology and connection solutions, i.e., 5G, AI, VR, and AR. In addition, we identify recurrent challenges and propose a design framework that can be adopted in low-resource settings. Finally, we outline future development directions, including AI-driven evaluation models, scalable system architectures, and policy frameworks to guide the development of secure, cost-effective, and equitable telementoring platforms to advance global surgical education.

1 January 2025 in Journal of Communications Software and Systems
Verified
Journal Article
Open access
PDF file
EMElena-Ramona Modroiu
Elena-Ramona Modroiu
JMJoyce Mwangama
Joyce Mwangama
MCMarius Corici
Marius Corici
TMThomas Magedanz
Thomas Magedanz
The Fifth Generation (5G) mobile network was developed to provide greater capacity, lower latency, and higher throughput, thereby enhancing multimedia service delivery. In the context of private 5G networks, open-source software is a key enabler of innovation and widespread adoption. This study introduces the first open-source prototype implementation of Voice over New Radio (VoNR) and evaluates its performance in two 5G Standalone (SA) private network deployment models: distributed and all-in-one (”in-a-box”). Leveraging an experimental setup composed entirely of open-source tools, the paper offers a comparative analysis of VoNR and Over-the-Top (OTT) voice services, examining their respective advantages and limitations. The findings contribute to the optimization of voice and multimedia services in next-generation (6G) networks and underscore the role of open-source platforms in achieving scalable, high-quality voice integration.

4 September 2026 in EPJ Research Infrastructures
Verified
Journal Article
Open access
PDF file
Raffaele BrunoRaffaele Bruno
Raffaele Bruno
MNMichela Natilli
Michela Natilli
SFSerge Fdida
Serge Fdida
DSDamien Saucez
Damien Saucez
NMNikos Makris
Nikos Makris
ABAdam Belloum
Adam Belloum
Digital research infrastructures underpin data-intensive science, reproducible experimentation, and collaborative innovation. In Africa, their development remains hindered by fragmentation, limited computing and connectivity resources, shortages of technical expertise, and sustainability constraints. This paper presents the vision, methodology, and deployment roadmap of DIGITAfrica, a Horizon Europe initiative preparing a pan-African Research Infrastructure in digital sciences. Building on a multi-country demand analysis and lessons from the ESFRI research infrastructures SLICES and SoBigData, we introduce a blueprint-based methodology for specifying interoperable, incrementally deployable, and sustainable digital infrastructures. We describe two representative blueprints—edge-interoperable AI/ML and heterogeneous networking—and the shared service architecture that operationalise them. Finally, we outline the stakeholder engagement process intended to support adoption, capacity building, and long-term governance. Overall, DIGITAfrica proposes a federated framework to strengthen digital science capabilities in Africa through shared infrastructure, reusable experimental environments, and coordinated community development.

1 November 2025 in arXiv
Verified
Preprint
Open access
PDF file
TBTravis Barrett
Travis Barrett
AMAmit Kumar Mishra
Amit Kumar Mishra
JMJoyce Mwangama
Joyce Mwangama
In this paper we present a new implementation of a Variational Autoencoder (VAE) for the calibration of sensors. We propose that the VAE can be used to calibrate sensor data by training the latent space as a calibration output. We discuss this new approach and show a proof-of-concept using an existing multi-sensor gas dataset. We show the performance of the proposed calibration VAE and found that it was capable of performing as calibration model while performing as an autoencoder simultaneously. Additionally, these models have shown that they are capable of creating statistically similar outputs from both the calibration output as well as the reconstruction output to their respective truth data. We then discuss the methods of future testing and planned expansion of this work.

17 November 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
SEStephen O. Ekwe
Stephen O. Ekwe
JMJoyce Mwangama
Joyce Mwangama
In dense New Radio (NR) sidelink Internet-of-Things (IoT) networks, weakly connected devices can reach the base station (gNB) through relays. However, a relay with a strong channel may still drop packets because it is selfish, low on battery, or malicious, or because its behaviour changes. Existing schemes assume that relays always forward, rank relays by trust, keep trust fixed, or require mutual trust, so they admit unreliable relays, miss behaviour changes, or exclude useful relays. We propose Trust-Feasible Relay Selection (TFRS), a closed-loop scheme in which trust decides which relays may be used. Trust starts from a ridge-regression prior on behavioural features and is updated from acknowledgements of relayed packets. Only pairs above a threshold are admitted, and a multi-objective evolutionary algorithm (MOEA/D) selects relays and transmit powers. A device keeps its relay only if this saves gNB resource blocks without degrading its service, or clearly improves it. Against trust-ranking and mutual-trust coalition schemes with 100–300 devices, TFRS lowers the outage under heavy gNB load by 3–5 percentage points. Paired Wilcoxon signed-rank tests with Holm correction confirm this reduction at p < 0.001. After a quarter of the altruistic relays turn compromised, it removes 77–92% of the added unreliable relaying within four epochs while keeping 81–85% of its relayed devices, whereas 44% of the trust-ranking scheme’s relayed devices stay on unreliable relays.