Volume 17, Issue 4, September 2026

Middleware and Gateway Architectures for IoT Integration: A Comprehensive Review

Maliha Shafique

Institute of Data Science Department, University of Engineering and Technology

Abstract - The Internet of Things (IoT) encompasses billions of interconnected devices spanning consumer electronics, industrial machinery, smart city infrastructure, and healthcare systems. The fundamental challenge underlying all IoT deployments is heterogeneity: devices differ in hardware capability, communication protocol, data encoding, and power budget, creating significant barriers to interoperability, scalability, and security. Middleware and gateway architectures address this challenge by providing protocol translation, data aggregation, and orchestration at edge and fog layers, enabling seamless integration between constrained devices and cloud platforms. Despite substantial progress, the field lacks standardized solutions across vendor ecosystems, and security gaps remain critically unaddressed in production deployments. This paper presents a systematic review of recent advances in IoT middleware and gateway architectures across five principal focus areas: brokered middleware and protocol translation, multi-protocol gateway design and cross-vendor interoperability, edge and fog orchestration, microservices-based scalable architectures, and security and performance evaluation. The review finds meaningful progress in distributed broker scalability, modular microservices decomposition, and semantic ontology-based protocol reconciliation, with reported gains of up to 37.5% in end-to-end latency and 66.7% in throughput under containerized middleware deployments. Empirical security analysis reveals that 99.84% of real-world MQTT deployments lack transport-layer encryption, exposing a critical gap between proposed solutions and production practice. Significant research gaps are identified in standardized benchmarking, unified northbound abstraction APIs, Quality of Service enforcement, and native security integration at the middleware layer. Five future research directions are proposed to guide the development of robust, scalable, and secure IoT ecosystems.

Index Terms - IoT, Middleware, Gateway, Edge Computing, Fog Orchestration, Microservices, Multi- Protocol Interoperability, Semantic Interoperability, Security, Quality of Service and Benchmarking

A Proposed Unified Framework for Internet Monitoring Techniques and Procedures in IoT

Pages: 8-22 (15) | [Full Text] PDF (743K)

Abdullah Salik

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

Abstract - Billions of ordinary objects now transmit data continuously across the internet, yet the field still lacks a single, coherent framework for watching over them end-to-end. This paper is my attempt to fill that gap. Working from seven peer-reviewed, open-access studies published in 2024 and 2025, I trace the full lifecycle of IoT monitoring from the moment a sensor fires to the instant an operator receives an alert, examining the architectural choices, protocol trade-offs, and security controls that determine whether a deployment succeeds or fails. The synthesis reveals that no existing study integrates all four monitoring stages, namely collection, transmission, processing, and alerting, inside one security-aware design. To address this, I put forward the Unified IoT Monitoring Framework, abbreviated UIMF, which embeds TLS-based verification at every inter-layer handoff and organises the pipeline into four discrete, auditable layers. Because UIMF has not yet been built and tested on physical hardware, every performance figure in Section V is a projection drawn from the benchmark results of the reviewed studies, and I label each estimate explicitly as such.

Index Terms - Internet of Things, IoT Monitoring, Edge Computing, Communication Protocols, MQTT, TLS, Network Security, Real-Time Alerting, Sensor Networks and UIMF

A Secure and Adaptive Conversational Intelligence Framework Using Blockchain and Dynamic Ontologies

Pages: 23-28 (6) | [Full Text] PDF (392K)

Chinara Adebayo, Babatunde Damilola

Department of IT, University of Science and Technology, Nigeria

Abstract - Ontology-based conversational agents can maintain contextual understanding by identifying relationships among concepts during interactions and leveraging previously acquired knowledge. However, their reliance on centralized knowledge structures and exposure to hostile or uncertain environments may lead to anomalous behavior, compromised knowledge integrity, suboptimal action planning, and ineffective adaptation. To address these challenges, this study proposes a blockchain-enabled dynamic ontology-based conversational model that integrates decentralized and tamper-resistant knowledge management with adaptive conversational intelligence. The proposed model utilizes prior experiences and a structured dynamic knowledge base to select appropriate responses, thereby improving contextual understanding, action planning, and adaptability in adverse environments. Blockchain technology further provides a secure and trustworthy environment for knowledge sharing, integrity preservation, and interaction management, reducing the risks associated with unauthorized modification and manipulation of conversational knowledge.

Index Terms - Blockchain Enabled Technology, Integrity Preservation, Challenges, Dynamic Ontologies and Conversational Intelligence

A Comprehensive Review of IoT-Based Classrooms: Sensors to Monitor Attendance, Engagement, and Environmental Conditions

Pages: 29-36 (8) | [Full Text] PDF (438K)

Hafiz Muhammad Moeez

Institute of Data Science, University of Engineering and Technology, Lahore

Abstract - The fast development of the Internet of Things (IoT) has resulted in significant changes in educational settings and the concept of the “Smart Classroom”. This paper deals with sensor technologies used for three crucial dimensions i.e. student attendance monitoring, engagement detection and environmental condition assessment and provides a comprehensive analysis of IoT based smart classroom systems. In this paper we review the literature published between 2015 and 2024 and study the sensor designs, communication protocols, data processing frameworks and machine learning algorithms used in different deployments of a smart classroom around the world. We identify key technology developments, highlight current issues such as data security, infrastructure costs and privacy concerns, and provide a coherent framework for future smart classroom implementations. Our analysis indicates that there has been a great deal of progress.

Index Terms - Internet of Things (IoT), Smart Classroom, Attendance Monitoring, Engagement Detection, Environmental Sensors, RFID, Computer Vision, Machine Learning and Edge Computing

Autonomous AI Agents for Facilitating Cross-Chain Interoperability in Heterogeneous Blockchain-Based Iot Ecosystems

Pages: 37-46 (10) | [Full Text] PDF (580K)

Shamroz Ali

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

Abstract - Due to the fast growth of the Internet of Things (IoT), there is an emergence of many connected devices that are communicating and exchanging information in real-time. Sensors, drones, autonomous cars, health devices, and other machines are becoming increasingly popular in today’s digital infrastructures. While the development of these systems continues, blockchain technology has become recognized as an appropriate way of providing security in IoT systems due to decentralized communication and trust management. Although blockchain offers significant benefits, interoperability remains a major challenge within IoT based on blockchain. Ethereum, Polkadot, Binance Smart Chain, and Hyperledger are examples of independent blockchain systems that are built on different architectures and protocols. This makes service coordination across several blockchain networks difficult. Current interoperability solutions such as blockchain bridges, relays, and atomic swaps lack flexibility and come up with their own security problems, scalability concerns, high transaction fees, and manual configurations. However, the aforementioned deficiencies are even more problematic in IoT networks where communication among machines is constant and real-time decision-making is required. This study proposes an innovative framework using Autonomous AI Agents (AAAs) for facilitating the secure and cost-effective interoperability of heterogeneous blockchain-based IoT networks. The proposed architecture would enable autonomous AI agents to automatically assess the conditions of the blockchain, choose proper blockchain networks for performing transactions, and perform machine-to-machine communications. Such a framework could ensure a reduction of delays, optimization of transaction fees, improvement of scalability, and support for intelligent Web3-based infrastructures. Simulation-based evaluation against Random, Round-Robin, and Static Rule-Based baselines showed that the proposed AAA-RL framework reduced the mean cross-chain transaction execution time by approximately 83% (to 312 ms) and the normalized transaction cost by approximately 63%, while achieving a 94.8% fault-recovery rate and near-linear scalability up to 1,000 IoT devices.

Index Terms - Autonomous AI-Agents, Blockchain Interoperability, Cross-Chain Communication and Decentralized IoT

Explainable AI for IoT Based Healthcare Systems: A Systematic Review

Pages: 47-56 (10) | [Full Text] PDF (628K)

Sehrish Ghouri, Muhammad Shahzad, M. Junaid Arshad

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

Abstract - The Internet of Things (IoT) has been a game-changer in the field of healthcare, with its ability to enable real-time tracking of patients, remote diagnostics, and intelligent clinical decision support. The utilization of Artificial Intelligence (AI) with the data created by IoT has enhanced the precision of disease forecasting, risk assessment, and individual treatment planning. The use of machine learning and deep learning models in healthcare, however, has come with the challenge of transparency, interpretability, and trustworthiness. The vast majority of AI systems used in healthcare are black box models, which means that it is hard for healthcare professionals to understand why and how the automated predictions are made. Explainable Artificial Intelligence (XAI) has come up as a potential remedy for these drawbacks by offering explanatory and transparent explanations for the outcomes of AI. The aim of this systematic review is to explore the latest trends in Explainable AI (XAI) for Internet of Things (IoT) based healthcare systems, covering research published between 2022 and 2026. Based on PRISMA, relevant literature was selected, screened, and evaluated to explore the use of XAI techniques in the context of healthcare applications including disease diagnosis, remote patient monitoring, heart disease prediction, epileptic seizure detection, elderly healthcare, and intelligent clinical decision support systems. The review emphasizes on the explainability methods commonly used such as SHapley Additive exPlanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), Gradient-weighted Class Activation Mapping (Grad-CAM), attention mechanisms, and federated approaches to explainability. The results show that SHAP and LIME are the most popular XAI methods because they offer both local and global explanations. Additionally, emerging technologies like blockchain, federated learning, and edge computing further ensure security, privacy, and scalability of healthcare ecosystems powered by IoT. While there is a great deal of progress, there are still multiple issues around computational complexity, real-time explainability, privacy preservation, clinical validation and the absence of standardized evaluation frameworks. This review highlights the current research gaps and outlines future research directions to build trustworthy, transparent and human-centered intelligent healthcare systems.

Index Terms - Explainable Artificial Intelligence, XAI, Internet of Things, IoMT, Smart Healthcare, SHAP, LIME, Disease Prediction, Remote Patient Monitoring, Healthcare Analytics and Federated Learning

A Systematic Review of IoT-Based Smart Home Automation: Architectures, Communication Protocols, Security Challenges, and AI-Driven Future Directions

Pages: 57-66 (10) | [Full Text] PDF (438K)

Muhammad Samar Azhar, Muhammad Usama Naeem, Muhammad Junaid Arshad

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

Abstract - Residential environments have been profoundly reshaped by the steady diffusion of Internet of Things (IoT) technology, and smart home automation now ranks among the most consequential application verticals in the broader IoT ecosystem. Modern deployments stitch together embedded hardware, a mixture of short-range and wide-area wireless links, cloud and edge compute tiers, and increasingly capable machinelearning pipelines in order to deliver measurable gains in comfort, energy efficiency, and physical safety. This article offers a systematic, critical review of the peerreviewed scholarship on IoT-based home automation, drawing on twenty-eight studies published between 2020 and 2025. We examine the layered hardware and software architectures that anchor contemporary systems, perform a structured comparison of the principal wireless and messaging protocols—MQTT, CoAP, Zigbee, Z-Wave, Bluetooth Low Energy, Wi-Fi 6, Thread, and the Matter interoperability standard—and assess the recurrent adoption obstacles of cybersecurity exposure, cross-vendor fragmentation, and scalability. Particular attention is paid to the ongoing fusion of IoT with supervised learning, deep learning, reinforcement learning, federated intelligence, and on-device inference, with empirical evidence from the reviewed corpus pointing to substantive gains in energy use, occupant welfare, and behavioural automation. Five forward- looking research directions are identified and unpacked: highdensity connectivity through 5G, privacy-preserving federated learning, digital twin simulation, blockchainbased decentralised trust, and standardised ambient assisted living frameworks for elderly and differently abled users.

Index Terms - Internet of Things, Smart Home Automation, Communication Protocols, MQTT, Machine Learning, Edge Computing, Cybersecurity, Federated Learning, Energy Management and Ambient Assisted Living

Quantum Machine Learning for Predicting Battery Drain Attacks in Drone Swarm Networks

Pages: 67-75 (9) | [Full Text] PDF (513K)

Sajid Iqbal, Muhammad Junaid Arshad

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

Abstract - Drone swarm networks are increasingly being applied in the field of surveillance, logistics, emergency response, industrial inspection and other mission-critical cyber-physical systems. Their distributed coordination, wireless connectivity, and small batteries leave them vulnerable to the attack of depletion-of-battery in which an adversary repeatedly forces them to communicate, process, retransmit or move to deplete their battery. This paper introduces a research study on quantum machine learning (QML) to predict the battery drain attack in drone swarm networks. The UAVIDS-2025 data sets serve as the basis of this network telemetry, with an energy simulation layer to convert intrusion traffic into battery-drain indicators to create a total of 32 features and binary class labels of Drain/Non-drain. Three models of quantum nature are taken into consideration only: optimized quantum support vector machine (QSVM), pure quantum variational quantum classifier (VQC), and quantum transfer learning (QTL) model. The optimized QSVM achieved 0.9933 accuracy, 0.9927 F1-score, 0.9998 ROC-AUC, and 0.9856 MCC. The pure VQC achieved 0.9760 accuracy, 0.9740 F1-score, 0.9926 ROC-AUC, and 0.9479 MCC. The QTL model was found to have an accuracy of 0.9800 and an F1 score of 0.9782, an ROC-AUC of 0.9983, and an MCC score of 0.9565 with edgefeasible inference latency of 2.41 ms. The results indicate that the early-warning ability of the QML-based prediction can be high-discrimination against UAV battery drain attacks, and both QSVM and QTL exhibit the best classification score and practical latency respectively.

Index Terms - Battery Drain Attack, Drone Swarm Network, Intrusion Detection, Quantum Machine Learning, Quantum Support Vector Machine, UAV Security and Variational Quantum Classifier

Volume 17, Issue 4, September 2026

IJCST Managemet

Current Issue

Recent Issues

Author Guidelines

IJCST RSS Feeds

News & Updates