Volume 17, Issue 3, July 2026

Agri Prediction Using Computational Techniques

Faiza Asghar, M. Junaid Arshad

Data Science Department, University of Engineering and Technology, Lahore, Pakistan

Abstract- In agricultural crop prediction, IoT [1] sensing is rapidly expanding the body of literature by integrating it with machine learning (ML) [2] and deep learning (DL) [3]. Nonetheless, the domain remains disaggregated: despite work advancing much technical progress with thousands of studies focused on narrow task-specific silos, using a single recirculated benchmark dataset, and seldom addressing practical aspects of real-world systems in terms of scalability, privacy, or adaptability. Using a targeted problematization approach, this paper critically reviews ten empirical studies published in 2024-2025 which were retrieved through Google Scholar using the structured query ("IoT") AND ("agri*") AND ("crop") AND ("prediction"). The review was restricted to open-access studies published between 2024 and 2025 within the agriculture domain. Not only is the merit of a study assessed (justifiably) based on the performance metrics it reports, but also in relation to the rigor and completeness of its data acquisition approach, methodological decisions, plausibility of reported results, as well as relevance of tabled limitations. In this chapter, seven systemic gaps are identified as a result of the critical synthesis. In this context, a novel six-layer conceptual architecture called the Multi-Layer Intelligent Agricultural Prediction Framework (MIAPF) is proposed that accurately closes each gap by cohesively designing and directly integrating all system aspects e.g., real-time IoT sensing, federation-based privacy preserving learning, multi-task prediction and continuous adaptive feedback into a cohesive deployable design. In the paper, they talk about how algorithmic improvements are not enough, and that there must be a from-the-ground-up rethink of agricultural intelligence systems architecture for advancement in this space.

Keywords- Internet of Things, Crop Prediction, Precision Agriculture, Machine Learning, Deep Learning, Federated Learning, Critical Review and MIAPF

Download full paper PDF format (Page: 1-12)

Assessment of Some Nigeria Gypsum Minerals for Application in Manufacturing, and Allied Industries

Jimoh S. O., Samuel O. Igbudu, Irabor P.S.A., Ediale Ameya Lucky

Department of Materials and Metallurgical Engineering, Faculty of Engineering and Technology, Ambrose Alli University, Ekpoma
Department of Mechanical Engineering, Faculty of Engineering and Technology, Ambrose Alli University, Ekpoma

Abstract- The present work investigates Assessment of some local gypsum minerals and to develop a process technology for their exploitation, refining and utilization in Nigeria. A number of gypsum deposit have been found in Nigeria; three varieties were assessed in this work. The gypsum samples from Igbokotor and Ibeshe villages in Ogun State were observed via a manual pitting method; while the third variety was procured from Potiskum in Bornu State. The raw gypsum was beneficiated to remove obvious physical impurities and air-dried. In this work, the raw gypsum was assessed to determine their chemical constituents using the conventional wet silicate technique. The six major significant constituents, carbon dioxide (CO2), Calcium oxide (CaO), magnesium oxide (MgO), sulphur trioxide (S03), ferrous oxide (Fe203) and combined matter were determine. An electrical kiln with digital control, a calcinations sequence of 1650C – 350C (temperature) range against 60 minutes – 300minutes (time) was used during the heat treatment procedure. The results of the assessment showed that the optimum water –plaster ratio 4:3 while the setting/hardening time was between 4.0 – 9.0 minutes. Other physical properties such as the density, colour, and particle size were found to be in agreement with literature. Consequent upon the assessment reported here an adaptive refining process technology for Nigeria Gypsum Mineral has been assessed and the process assessment and description are presented.

Keywords- Calcination, Dehydration, Rehydration, Setting and Refining

Download full paper PDF format (Page: 13-22)

Self-Management in IoT-Enabled Mobile Ad-Hoc Networks Using Machine Learning

Saleha

Department of Computer Science, UET (Main Campus)

Abstract- Mobile Ad Hoc Networks (MANETs) based on IoT are actively supported in the environment when a coherent communication system is missing or in its infancy or when it is impaired like emergency management, military communication, transportation, smart healthcare, and environmental monitoring. However, providing IoT-MANETs routing may be difficult, as the mobility of nodes, low battery, unstable wireless connections, congestion and dynamic topology attenuate the performance of classical routing options. Traditional standards, such as AODV and DSR, tend to be connection-based and rely on route exploration, and do not require any persistent contextual information regarding the remaining energy, mobility, queue occupancy, congestion, packet loss and link stability. The current paper proposes the machine learning-based self-management structure of adaptive routing in MANETs with the IoT. The design integrates monitoring, analysis, decision-making, adaptation and learning into a closed loop architecture. The assessment using the uploaded dataset of simulations is obviously a simulation based rather than a field-data based assessment (real time). The dataset is 180 simulation observations with low, moderate and high-mobility conditions, three routing strategies and re-executions. Exploratory data analysis shows RL-SelfManaged gives the best routing performance in the sense that it enhances the ratio of packets delivery by 9.07, reduces delay increment by 18.12, 19.05 in energy consumption and network lifetime respectively with AODV. The findings confirm the notion that learning-based self-management can supplement the dynamic control of IoT-enabled MANET settings to improve the routing assurance, energy efficiency, and sustainability of this type of network.

Keywords- Internet of Things, Mobile Ad Hoc Network, Machine Learning, Self-Management, Reinforcement Learning, Intelligent Routing and Simulation-Based Evaluation

Download full paper PDF format (Page: 23-32)

Wireless Data Transfer in Decentralized Systems: Technologies, Challenges, and Future Directions

Albert Okino, Olutosim Aminu

University of Science and Technology, The Technical University of Kenya, Nairobi, Kenya

Abstract- The increasing demand for reliable, scalable, and energy-efficient data communication has accelerated the adoption of decentralized systems across diverse application domains. Wireless communication technologies play a crucial role in enabling seamless data transfer within these systems. This review paper presents a comprehensive analysis of prominent wireless technologies, including Bluetooth, Wi-Fi, LoRaWAN, and ZigBee, in the context of decentralized architectures. A comparative discussion of centralized and decentralized systems is provided, highlighting their respective advantages, limitations, and architectural differences through illustrative diagrams. The review further examines existing literature on decentralized wireless communication and presents a detailed comparison of the selected technologies based on key performance parameters such as communication standards, operating frequency, transmission range, data rate, and energy consumption. Based on this comparative analysis, ZigBee and Bluetooth are identified as the most suitable technologies for decentralized applications due to their low deployment cost, high packet delivery performance, low power consumption, and support for large network capacities. The findings of this review provide valuable insights for researchers and practitioners in selecting appropriate wireless technologies for efficient and scalable decentralized communication systems.

Keywords- Wireless Technology, Energy Efficient, LoRaWAN, ZigBee, De-centralized and Power Consumption

Download full paper PDF format (Page: 33-38)

Designing Secure and Scalable Networks for Modern Enterprise Applications

Muhammad Asim, M. Junaid Arshad

Computer Science Department, University of Engineering and Technology, Lahore, Pakistan

Abstract- The paper reviews modern approaches for designing secure and scalable enterprise networks in response to the rapid growth of cloud-native applications, edge computing, and distributed architectures. It focuses on key architectural patterns such as microservices, container orchestration, and hybrid/multi-cloud systems, along with enabling technologies like software-defined networking (SDN) and network function virtualization (NFV) that allow flexible and programmable network control. The study also emphasizes security-by-design principles, particularly Zero Trust Architecture (ZTA), identity-based access control, multi-factor authentication, and end-to-end encryption to protect data across distributed environments. It further examines integrated security and performance mechanisms, including firewalls, intrusion detection and prevention systems, secure API gateways, service meshes, and observability tools for monitoring and anomaly detection. Scalability techniques such as auto-scaling, traffic engineering, and distributed storage (sharing and replication) are also discussed for ensuring high availability and fault tolerance. The paper concludes by identifying challenges like system complexity, interoperability issues, latency and cost trade-offs, and regulatory compliance, while highlighting future directions for building more adaptive, resilient, and secure enterprise network infrastructures.

Keywords- Enterprise Networks, Zero Trust Security, Hybrid and Multi-Cloud Systems, Cloud-Native Architecture, Software-Defined Networking (SDN) and Network Function Virtualization (NFV)

Download full paper PDF format (Page: 39-44)

Wireless Communication Networks for Internet of Things

Aqsa Bashir, Arzoo

CS Department, University of Engineering and Technology (UET), Lahore

Abstract- The rapid advancement of wireless communication technologies has significantly transformed the development of Internet of Things (IoT) systems in recent years. IoT enables communication between interconnected devices, sensors, machines, and smart applications through internetbased networks. Modern IoT systems are increasingly integrated into healthcare, agriculture, smart cities, industrial automation, transportation, and environmental monitoring. This review paper presents a detailed discussion of IoT communication technologies and their role in nextgeneration smart systems. Different wireless technologies including Bluetooth, ZigBee, LoRa, Wi-Fi, LTE, NB-IoT, and 5G are critically analyzed based on communication range, power consumption, reliability, and application requirements. The study also examines enabling technologies such as Edge Computing, Software Defined Networking (SDN), Artificial Intelligence (AI), and Network Function Virtualization (NFV). Additionally, the paper highlights important challenges including cyber-security risks, interoperability, latency, scalability, and energy management. The findings indicate that the integration of AI and 5G communication technologies will play an important role in the future development of intelligent and fully automated systems.

Keywords- AI, SDN, 5G, NFV, NB-IoT and Communication Technologies

Download full paper PDF format (Page: 45-50)

Review of Anomaly Detection Techniques in Internet of Things

Iram Fatima, Rimsha Tariq

Department of Computer Science, University of Engineering and Technology, Lahore, Pakistan

Abstract- With the Internet of Things (IoT) has come the deployment of billions of connected devices producing vast amounts of complex, heterogenous data in a variety of application areas. Despite the many benefits of IoT systems in terms of automation, efficiency, and real-time decision-making, they also come with a number of security and reliability challenges, especially when they operate in a distributed environment and are resource-constrained. In IoT systems, the identification of abnormal behaviors, cyber-attacks and system faults has become a central problem and anomaly detection has proved a viable approach to this problem. We provide a comprehensive systematic literature review (SLR) of the anomaly detection techniques in the IoT networks, including statistical approaches, machine learning, deep learning, hybrid approaches, and emerging approaches. A review of recent studies (2019-2026) gathered from the main scientific databases is carried out and existing techniques are classified according to their methodology, data, evaluation metrics and application areas. Different approaches are compared with regard to their strengths and weaknesses regarding accuracy, scalability, complexity, and real time applicability. This study also highlights several research challenges such as data heterogeneity, class imbalance, concept drift, computational limitations, and privacy issues. Lastly, it points to future research paths including federated learning, edge intelligence, self-supervised learning, and light weight deep learning models for future scalable and privacy-preserving IoT security systems. The results will inform the research and practice community in next generation IoT environments in the creation of efficient, robust and adaptive anomaly detection solutions.

Keywords- Internet of Things, Anomaly Detection, Machine Learning, Deep Learning, Intrusion Detection Systems (IDS), Federated Learning, Edge Computing and IoT Security

Download full paper PDF format (Page: 51-62)