The Internet of Things (IoT) ecosystem is rapidly expanding. It is driven by continuous innovation but accompanied by increasingly sophisticated cybersecurity threats. Protecting IoT devices from these emerging vulnerabilities has become a critical priority. This study addresses the limitations of existing IoT threat detection methods, which often struggle with the dynamic nature of IoT environments and the growing complexity of cyberattacks. To overcome these challenges, a novel hybrid architecture combining Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory (BiLSTM), and Deep Neural Networks (DNN) is proposed for accurate and efficient IoT threat detection. The model’s performance is evaluated using the IoT-23 and Edge-IIoTset datasets, which encompass over ten distinct attack types. The proposed framework achieves a remarkable 99% accuracy on both datasets, outperforming existing state-of-the-art IoT cybersecurity solutions. Advanced optimization techniques, including model pruning and quantization, are applied to enhance deployment efficiency in resource-constrained IoT environments. The results highlight the model’s robustness and its adaptability to diverse IoT scenarios, which address key limitations of prior approaches. This research provides a robust and efficient solution for IoT threat detection, establishing a foundation for advancing IoT security and addressing the evolving landscape of cyber threats while driving future innovations in the field.

Mobile health (mHealth) systems are sipping into more and more healthcare functions with self-management being the foremost modus operandi. However, there has been challenges. This study explores challenges with mHealth self-management of diabetes and hypertension, two of the most comorbid chronic diseases. Existing literature present the challenges in fragments, certain subsets of the challenges at a time. Nevertheless, feedback from patient/users in extant literature depict very variegated concerns that are also interdependent. This work pursues provision of an encyclopedic, but not redundant, view of the challenges with mHealth systems for self-management of diabetes and hypertension.Furthermore, the work identifies machine learning (ML) and self-management approaches as potential drivers of potency of diabetes and hypertension mobile health systems. The nexus between ML and diabetes and hypertension mHealth systems was found to be under-explored. For ML contributions to management of diabetes, we found that machine learning has been applied most to diabetes prediction followed by diagnosis, with therapy in distant third. For diabetes therapy research, only physical and dietary therapy were emphasized in reviewed literature. The four most considered performance metrics were accuracy, ROC-AUC, sensitivity, and specificity. Random forest was the best performing algorithm across all metrics, for all purposes covered in the literature. For hypertension, in descending order, hypertension prediction, prediction of risk factors, and prediction of prehypertension were most considered areas of hypertension management witnessing application of machine learning. SVM averaged best ML algorithm in accuracy and sensitivity, while random forest averaged best performing in specificity and ROC-AUC.

Skepticism about the security of mobile payment applications has plagued user adoption of such platforms in some countries. Software developers have generally de-emphasized core principles guiding delivering safe mobile applications since for mobile payment applications, movement of monetary value is their priority. We find in surveyed literature that this situation is prevalent in low economy/low financial inclusion countries. Fifty fintech and traditional banks m-payment applications in both high and lower economic and technological advancement (high E&T apps and lower E&T apps, respectively) were selected in countries in Africa. This work may have significance in finance or economy, but it is mainly to unravel cybersecurity concerns. The analyses (static and dynamic) of the applications targeted the top 10 vulnerabilities on the 2023 Common Weakness Enumeration (CWE) and Open Worldwide Application Security Project (OWASP) lists. The study employed Mobile Security Framework (MobSF) as the primary tool for both Android and iOS applications, while the Automated Security Risk Assessment (AUSERA) tool was used to validate the vulnerabilities reported by MobSF. Results show that traditional m-payment apps were generally more secure than fintech m-payment apps. In the latter category, vulnerabilities under the information leakage and cryptography category were the most prevalent. On the average, no marked difference was observed in security performance between high E&T apps and lower E&T apps. Incorrect default permission, cleartext storage of sensitive information, use of risky cryptographic algorithm, use of insufficiently random values, and information exposure were the most prevalent vulnerabilities. Conversely, insecure implementation of Secure Socket Shell (SSL) and trusting all certificates or accepting self-signed certificates had the fewest occurrences. Poor code quality was the highest source of security vulnerabilities in the study. The declining statistics of SMS leakage in recent studies were confirmed in this work. The most implemented security measure was certificate pinning for preventing or detecting man-in-the-middle attacks. Received: 7 February 2024 | Revised: 28 May 2024 | Accepted: 31 May 2024 Conflicts of Interest The authors declare that they have no conflicts of interest to this work. Data Availability Statement Data are available from the corresponding author upon reasonable request. Author Contribution Statement Esther Edem Archibong: Methodology, Software, Investigation, Resources, Visualization. Bliss Utibe-Abasi Stephen: Validation, Data curation, Writing - original draft, Supervision, Project administration. Philip Asuquo: Conceptualization, Formal analysis, Writing - review & editing.

Aims: Mobile payment apps have become the most used mobile payment system, especially in underbanked regions. It has come be the most viable way of promoting access to and use of quality financial services, at affordable cost. As far as review works on use/adoption of mobile payment apps in low financial inclusion countries go, presentation has been from users’ perspective. Thus, under-emphasizing the challenges of other members of the m-payment app ecosystem. The work aims at describing such general challenges from the lenses of users, regulators and merchants of mobile payment apps. Also, pros and cons of some specific mobile payment apps in these regions are also provided in the work. Place and Duration of Study: The review covered literature covering m-payment adoption in low financial inclusion countries i.e. Asian and African countries, spanning 2015 to 2023. Methodology: In selecting the literature, we considered only journal publications considering specifically m-payment apps, and written in English. To achieve a measure of spread, major regions in the two continents were considered, and journal articles focusing on a particular country were limited to two. Conclusion: We observe that the challenges of m-payment apps from users’ perspective are mostly denser than other perspectives. Users’ and regulators’ concerns converge only so thinly. Only regulators’ task of enlightenment of users/customer of about demonetization policies mirrors one of users’ concerns, perceived trust. Users and merchants are both concerned with perceived risk and perceived ease of use concerns. The challenges of regulators as regards m-payment apps are quite dissimilar to those of merchants. In all, the observed disparate (or thin similarity of) concerns of the different m-payment app stakeholders strengthens the need for such an inclusive review done in this work.

The rapid expansion of cloud-native infrastructure and mobile computing has rendered traditional perimeter-based security obsolete, necessitating the development of resilient digital frameworks in alignment with United Nations Sustainable Development Goal 9 (Industry, Innovation, and Infrastructure). This review article investigates the convergence of two critical paradigms: Zero-Click Attacks, which exploit system components without user interaction, and Zero-Trust Architecture (ZTA), the necessary adaptive defence strategy. Adopting the PRISMA-based systematic review methodology, this study synthesises current literature to provide a critical evaluation of the evolution of zero-click vectors, focusing on critical vulnerabilities within Android system components and media processing libraries. The efficacy of current detection methodologies is evaluated, contrasting the limitations of static analysis with the promise of sandbox-based behavioural monitoring. Furthermore, the paper explores the transition from static ZTA policies to AI-driven enforcement, utilising the SmartTrust framework as a primary case study. This hybrid deep learning approach, integrating Convolutional Neural Networks, Long Short-Term Memory networks, and Reinforcement Learning, demonstrates how real-time, context-aware risk scoring can mitigate invisible threats. Finally, the review outlines open research challenges, specifically the scalability of behavioural analysis and the resilience of AI models against adversarial evasion, proposing a strategic roadmap for AI-native Zero-Trust in hyper-connected environments.