Kampala International University

Artificial intelligence (AI) is increasingly applied to photovoltaic (PV) systems for power forecasting, maximum power point tracking, converter and storage coordination, fault diagnosis, and operational decision-making. However, reported gains are difficult to compare because studies differ in datasets, baselines, climates, system scales, metrics, and validation environments. This systematic review evaluates not only whether AI methods perform well, but also where, against which comparators, under what conditions, and with what deployment burden. Searches conducted from January to July 2026 across eight scholarly databases identified 1284 records; 1012 unique records were screened, 286 full texts assessed, and 71 studies included. Evidence was coded by task, AI family, application level, data source, metric, validation fidelity, computational reporting, and limitation. No AI family was universally superior. Machine-learning and fuzzy methods suit resource-constrained settings; deep learning performs strongly with high-dimensional data but compute-intensive; evolutionary methods support offline optimization; and reinforcement learning is promising for sequential control but needs stronger safety and field validation. Key gaps include external validation, explainability, computational reporting, cybersecurity, and long-term field evidence.

This study examines the reliability and maintenance of solar photovoltaic (PV) systems in Lango, Uganda, using a descriptive survey of 280 respondents, site observations, Failure Modes and Effects Analysis (FMEA), and exponential reliability modelling. Aggregate reported-event counts were 460 for solar panels, 965 for batteries, 690 for inverters, and 310 for charge controllers. Using a nominal exposure of 102,200 component-days per class, estimated event rates were 0.004501, 0.009442, 0.006751 and 0.003033 day⁻¹, respectively, with corresponding 95% Poisson/chi-square confidence intervals of 0.004099–0.004932, 0.008856–0.010057, 0.006257–0.007275 and 0.002705–0.003390 day⁻¹. Under an independent exponential series model, calculated system reliability was 1.40% at 180 days and 0.01733% at 365 days. These estimates are conditional on nominal exposure and unverified classification of reported events as functional failures. An analytical cost and availability model compares corrective, age-based, and grouped maintenance policies. The constant-hazard model predicts no preventive-maintenance reliability gain; an optimal interval requires intervention costs, outage durations, and measured maintenance effects. The findings identify priorities for validating operational records and developing component-specific maintenance policies.

The increasing digitization of academic environments has made campus networks essential for learning, research, and administration; however, these networks are highly vulnerable to evolving cyber threats. Among these threats, Distributed Denial of Service (DDoS) and Address Resolution Protocol (ARP) spoofing attacks pose significant risks to network availability, integrity, and confidentiality. Traditional rule-based and signature-based intrusion detection systems are limited by their static nature and poor adaptability to the dynamic traffic patterns typical of campus networks. This review systematically examines Machine Learning (ML) techniques for detecting and preventing DDoS and ARP attacks. Guided by the PRISMA framework, a total of 110 studies were selected from 206 peer-reviewed papers published from 2020 to 2025. Supervised, unsupervised, and hybrid ML models utilizing algorithms such as Random Forests, Support Vector Machines, Convolutional Neural Networks, Long Short-Term Memory networks, and CNN-LSTM ensembles demonstrate high detection accuracy and robust anomaly modeling under real-time, resource-constrained conditions. Integration with Software-Defined Networking (SDN), federated learning, and edge computing enhances scalability, privacy preservation, and low-latency mitigation. Despite these advances, challenges remain in dataset representativeness, model generalization, interpretability, computational overhead, and the development of unified frameworks capable of simultaneously mitigating volumetric DDoS, application-layer attacks, and Layer-2 ARP spoofing. Emerging approaches, such as reinforcement learning and explainable AI, offer promise for adaptive, zero-day, and adversarial threat mitigation. The review highlights the need for modular, context-aware, and scalable ML frameworks validated on real campus traffic to enable resilient and operationally feasible intrusion detection and prevention. These insights provide a foundation for designing secure, adaptive, and intelligent network defense strategies in modern campus environments.

Persistent toxic substances (PTS), including heavy metals, persistent organic pollutants (POPs), and persistent, mobile, and toxic/very persistent and very mobile (PMT/vPvM) substances present an increasing menace to soil health, alimentary systems, atmospheric cleanliness as well as human health. Despite the large amount of literature on each of the individual groups of contaminants, there is still no unified model that connects the dynamics of the soil-atmosphere environment, bioaccumulation in the food chain, new detection techniques, and policy measures. This review presents an interdisciplinary synthesis of dynamics in the PTS in the agricultural environment, explicitly incorporating (i) historic contaminants and emerging PMT/vPvM chemicals, (ii) soil-crop-livestock-human transfer pathways, and (iii) the state-of-the-art remediation and monitoring technologies into a single management framework. We critically evaluated conventional remediation methods alongside next-generation methods, such as engineered consortia of microorganisms, synergistic phytotransformation of plants and microbes, biochar-assisted immobilization, nanosensor-based detection, IoT-based soil sensing, precision agriculture, machine-learning-driven risk prediction, and blockchain-based traceability. Contrary to the previous reviews, which only take into account the remediation, detection, and policy separately, this study presents a systems-based approach, which integrates technological innovation, sustainable agronomic practices, and multilayered governance tools (such as the Stockholm Convention, REACH, and national soil action plans). We highlight the fact that the combination of smart agricultural technology and regenerative land management will help reduce the accumulation of PTS and maintain productivity, especially in resource-scarcity settings. The review outlines the research gaps, including contaminant-microbiome interactions, longitudinal deterioration of ecosystem services, and socioeconomic barriers to technology adoption. We propose a transdisciplinary roadmap that aligns environmental toxicology, soil science, public health, and policy innovation to mitigate PTS and safeguard food security. This integrative approach provides a strategic framework for advancing sustainable management of persistent toxic substances in agricultural systems. This study looks at persistent toxic substances (PTS), harmful chemicals like some pesticides, industrial pollutants, and heavy metals that do not easily break down in the environment. Because they linger in soil, water, air, and food, they can move through the food chain and affect both ecosystems and people. In this study, the authors reviewed recent research and real-world cases to explain where PTS come from (e.g., farming chemicals, industrial waste, plastics), how they spread (air, water, and soil), and what health problems they can cause (such as hormone disruption, breathing issues, nerve damage, and cancer). They also examined solutions, from traditional cleanup methods to newer, nature-based options. Created in https://BioRender.com

Glycated hemoglobin (HbA1c) is a well-established biomarker reflecting chronic glycemic control in diabetes. It accumulates through non-enzymatic glycation of hemoglobin under sustained hyperglycemia and serves as a surrogate of metabolic memory. Emerging in parallel, glycosylated RNA (glycoRNA), small noncoding RNAs bearing covalently attached N-linked glycans, has revealed unexpected roles in immune signaling and glycoimmunomodulation. While glycation and glycosylation represent distinct biochemical processes, both are modulated by glucose availability and cellular stress. This opinion paper aimed to explore the conceptual parallels between HbA1c and glycoRNA, proposing that hyperglycemia-induced metabolic changes may simultaneously influence both processes. In light of these, glycoRNA represents an emerging biomarker as a functional effector in metabolic disease, mirroring the hyperglycemia-driven immunological dimensions of HbA1c with a further advantage of a defined metabolic sequence, which can deepen diagnostics towards disease progression patterns. Though direct experimental evidence is currently limited, we outline plausible mechanistic intersections and suggest methodological frameworks for future research. Therefore, this perspective aims to stimulate interdisciplinary investigation into glycoRNA biology within the broader context of glycemic dysregulation and immune modulation in diabetes.