International Journal of Internet Protocol Technology (24 papers in press)

Regular Issues

• Application of improved Particle Swarm Optimisation in Intelligent Prediction of Construction Cost  
  by Liangqiong Chen, Hao Zhang, Fang Wang 
  Abstract: In order to understand the application of intelligent prediction of construction project cost, the author proposed an improved particle swarm optimization algorithm in the application of intelligent prediction of construction project cost. With the proposal of the core task of the State Grid Corporation to implement the strategic goal of "three types and two networks and world-class", the traditional cost management mode of the power transmission and transformation projects will be promoted to the information, intelligent and professional cost management mode. The author first introduces the significance of construction project cost prediction, the construction of construction project cost prediction index system, and data pre-processing methods. Secondly, the particle swarm optimisation model with local search is selected to simulate the case project. The results show that the optimal cost of the project is 1224600 yuan, which is consistent with the actual cost results obtained by traditional methods.
  Keywords: Particle swarm; architectural engineering; Intelligent prediction; construction cost.
  DOI: 10.1504/IJIPT.2027.10075422
   
  
• Connected Graph with WiFi Network in Distance  
  by Selvaraj Thangaraj, Subramanian S 
  Abstract: This research investigates the optimisation of Wi-Fi networks using graph theory, where access points are modelled as nodes and communication links as edges. By analysing real-world Wi-Fi network data through various graph-theoretic metrics, we identify critical factors that influence network behaviour and performance, revealing significant correlations between these metrics and key performance indicators, such as latency, throughput, and packet loss. The study proposes a comprehensive graph-based framework for adapting Wi-Fi network settings, aiming to enhance network efficiency and user satisfaction. These findings provide valuable insights for the design, management, and optimisation of wireless networks, addressing the complexities inherent in modern digital communication environments and contributing to more reliable and effective connectivity solutions.
  Keywords: Wi-Fi networks; graph theory; performance evaluation; optimisation; access points; user satisfaction; network dynamics; distance metrics; graph metrics; efficiency.
  DOI: 10.1504/IJIPT.2026.10075921
   
  
• Intelligent Support and Content Delivery for ESP Terminology Learning: Design and Validation of NeuroXL-CRFNet  
  by Zhu Lingyi 
  Abstract: This study proposes a teaching assistance model for the cultivation of building informatisation talents based on a distributed architecture. Through the integration of microservice architecture and cloud computing technology, the dynamic scheduling and efficient management of teaching resources are achieved. This model takes modular design as the core, combines key technologies such as service registration discovery and API gateway, and supports high concurrent access and elastic scalability. Experiments showed that the model significantly improved the prediction accuracy by introducing preference factors and attendance behaviour analysis in the performance prediction algorithm. Meanwhile, the system could still maintain a stable response speed in high-concurrency scenarios, and the delay was controlled within an acceptable range. The proposed distributed teaching assistance model is scalable, flexible, and highly available, providing theoretical and practical references for the cultivation of talents in building informatisation.
  Keywords: Edge Computing; Protocol-Aware Networks; Neuro-Symbolic Graph Convolution; Multilingual Content Dissemination; Intelligent Routing; Feature Fusion; Adaptive QoS; Distributed Inference;ESP Terminology.
  DOI: 10.1504/IJIPT.2027.10075943
   
  
• An Internet Protocol-Based IoT and Machine Learning Framework for Enhancing Context-Aware English Translation Services  
  by Yan Li 
  Abstract: This paper proposes an Internet Protocol-based IoT and machine learning framework to enhance context-aware English translation and personalized language learning Traditional English teaching methods often fail to address individual learning differences, leading to reduced student engagement To address this, an IoT-supported learning system is designed to collect real-time learner data including behaviour, emotion and progress across multiple networked devices. A fuzzy logic-based evaluation model is developed to quantify student engagement across four dimensions: learning attitude, learning process, learning effectiveness and emotional experience. The system employs machine learning algorithms to analyse this contextual data, enabling adaptive translation services and targeted engagement strategies. Using a case study involving first-year English majors, the framework demonstrates a significant increase in engagement scores from 71.37 to 88.49 after implementing the proposed improvement mechanism. Results validate the effectiveness of integrating IoT data and intelligent protocols to support adaptive English learning environments and enhance translation accuracy based on real-world context.
  Keywords: Internet of Things (IoT); Internet Protocols Fuzzy Mathematical Model; Student Engagement; Personalized English Learning.
  DOI: 10.1504/IJIPT.2027.10076240
   
  
• Protocol-Sensitive B5G-IoT Learning Architecture for Real-Time Civic-Political Education in College English Writing Streams  
  by Shali Zhou 
  Abstract: The way in which distributed learning frameworks provide real-time communication is being redefined by the fast integration of intelligent Internet of Things (IoT) ecosystems with Beyond 5G (B5G) networks. This study suggests a Protocol-Sensitive B5G-IoT Learning Architecture (PS-BILA) that uses adaptive protocol intelligence to ensure the synchronisation of distributed IoT nodes with cloud-based learning platforms, control latency and boost bandwidth. Utilisation of resources, routing and session continuity may be optimised across dense user contexts with the help of the architecture's protocol-aware control layer. Dynamic transmission protocols that adapt QoS attributes in response to network performance and learner interactions are essential to the PS-BILA architecture, which also includes multi-source data collecting and attention detection based on the Internet of Things. Through the use of mathematical performance metrics including Signal-to-Noise Ratio (SNR), end-to-end latency, content integrity and engagement fidelity, we may evaluate the systems responsiveness and data transfer efficiency. After testing on fifty IoT nodes made possible by B5G technology, the suggested design achieved a total protocol efficiency score of 0.95 with latency of 30 ms, audio SNR of 30 dB and content integrity of 99%.
  Keywords: Protocol-Sensitive Communication; B5G-IoT Framework; Intelligent Learning Systems; Real-Time Education; Civic-Political Learning; College English Writing; Adaptive Streaming.
  DOI: 10.1504/IJIPT.2027.10076295
   
  
• An Optimising English Online Teaching Mode based on Learning Process Data Mining  
  by Fang Wang 
  Abstract: In order to adapt to the development needs of English language ability in the context of globalisation and shorten the gap with current education mode optimisation methods, an optimising English online teaching via learning process data mining using K-prototype clustering and neural evaluation is proposed. Using the K-prototype algorithm to mine process data of English online learning, determine the evaluation index data of English online teaching quality and input the evaluation index data into a fully connected neural network to obtain the teaching quality evaluation results. Optimising English online teaching mode from four dimensions: technology integration, instructional design, teacher-student interaction and resource integration. The experimental results show that the proposed method has a minimum relative error rate of 4.18% in data mining, a maximum improvement rate of 13.19% in exam scores and a maximum knowledge retention rate of 76.37%, which can promote further innovation in the field of education.
  Keywords: Learning process data mining; English online teaching; Mode; Optimization; K-prototype algorithm; Fully connected neural network.
  DOI: 10.1504/IJIPT.2026.10076299
   
  
• Automated Sensor Node Failure Detection and Recovery Procedure in Mobile Wireless Sensor Networks  
  by Supratik Banerjee, Sanjay Kumar Biswash 
  Abstract: Sensor node failure detection and recovery is one of the active research areas. The current literature lacks focus on automated systems for detecting and replacing faulty sensor nodes. To address this gap, we propose two schemes for automatic identification and replacement of faulty sensor nodes deployed in inhospitable terrains. In the first scheme, we propose a methodology to deploy some backup Mobile Sensor Nodes (MSN). In case of MSN failure, a backup MSN is relocated to the area of interest. In the next scheme, we propose to deploy multiple sensor nodes on a single mobile vehicle. One of these sensor nodes is in an active state. The remaining backup sensor nodes stay in sleep mode. As node relocation is not necessary, it reduces the reaction time after sensor node failure. Results show that Scheme 1 requires a longer recovery period compared to Scheme 2. On the other hand, Scheme 2 incurs more deployment costs.
  Keywords: Mobile Wireless Sensor Networks; Named Data Networking; Reliability; Weibull Distribution; Zigbee.
  DOI: 10.1504/IJIPT.2026.10076317
   
  
• Protocol-Aware AI Framework for 5G-Enabled English Translation in University Education Using Intelligent IoT and Machine Learning  
  by Qun Liu 
  Abstract: The integration of fifth-generation mobile communication (5G), artificial intelligence (AI), and smart Internet of Things (IoT) devices is transforming university instruction by enabling protocol or situational-awareness, real-time, and crypto-cognitive learning contexts. This study proposes the use of a protocol-aware AI framework for the English translation of instructional materials in higher education, employing 5G-enabled connections, intelligent IoT devices, and machine-learning algorithms to improve contextual awareness, learner engagement, and educational personalisation. In addition, the protocol-aware AI framework takes advantage of 5G's high bandwidth, ultra-low latency, and massive device connectivity capabilities to enable real-time data collection, intelligent learning analytics, and engaging immersive virtual learning experiences while providing secure and responsive communication between distributed university education modals. This method combines AI language models with IoT- enabled student interaction monitoring and adaptive content delivery to ensure that delivery of translated content and subsequent teaching remains fluid and relevant to the learning context. The experimental results demonstrate significantly improved translation accuracy, learner engagement and content personalisation using the protocol-aware framework versus conventional methods.
  Keywords: Protocol-Aware Framework; AI-Driven Translation; 5G-Enabled Education; Intelligent IoT Systems; Machine Learning in Education; Real-Time Language Processing; Context-Aware Learning; Adaptive Education.
  DOI: 10.1504/IJIPT.2027.10076393
   
  
• AI-Enhanced Nonlinear Economic Modelling for Sports Systems: Linking Material Performance and Cybersecurity Risk in Insurance Cost-Benefit Analysis  
  by Tang Shijie 
  Abstract: This research paper puts forward an AI-Driven Nonlinear Economics Modelling Framework (AI-NLEM) via a hybrid CNN-LSTM architecture aimed at supporting cost-benefit analysis of insurance systems within accountability measures in sport. It achieves this by merging manufacturer-quality material-performance data obtained from SEM and TEM imaging modalities along with logs of cyber vulnerabilities. Preprocessing of images consists of several steps including: denoising; histogram equalisation; augmentation; utilisation of cyber vulnerability data through tokenisation, BERT embeddings, TF-IDF, and DTW. The extraction of features from the images using LBP, Gabor Filters, ResNet50, and CC-GBD models. The optimisation and prediction model was performed using a novel approach called Deep Q-Guided Genetic Evolution which achieved superior predictive performance relative to all baseline models (Accuracy = 0.9957 and AUC = 0.9957). Data provides evidence that accurate risk prediction motivates superior safety equipment; future research will focus on creating real-time, scalable, multimodal systems.
  Keywords: AI-driven economic modeling,Nonlinear optimization,Sports system performance,Material property evaluation,Cybersecurity risk assessment,Insurance cost-benefit analysis,Predictive analytics in sports.
  DOI: 10.1504/IJIPT.2027.10076440
   
  
• Energy-Aware Nonlinear User Behaviour Modelling for Adaptive Power Management in Edge Computing Systems  
  by Yu Bai, Yu Qin 
  Abstract: An Energy-Aware Behavioural Modelling Network (EABM-Net) framework for adaptive power management of complex user behaviours in edge computing is presented. Using BiLSTM for temporal analysis of user behaviour, and CNN for spatial analysis of device state and energy consumption, this model also uses a self-attention mechanism to help identify the most relevant user behaviours and device-level factors that affect energy consumption. The data were pre-processed by smoothing, normalising, and enhancing features before applying hybrid feature extraction and optimisation techniques for use in energy consumption forecasting and anomaly detection. Our experimental results show that EABM-Net demonstrated superior performance compared to existing approaches, achieving accuracy values of 0.9587, F1 scores of 0.9746, and R2 values of 0.987, proving the applicability of EABM-Net to develop energy-efficient systems for edge environments.
  Keywords: AI-Driven Modelling; User Behavior; Edge Computing; Material Degradation; Deep Learning; Anomaly Detection.
  DOI: 10.1504/IJIPT.2027.10077231
   
  
• Enhancing IOT -Image Processing using Adoptive Filtering Techniques  
  by Sangeeta Mangesh, Seema Shukla 
  Abstract: Image pre-processing is crucial in any advanced image processing based smart IoT system. Through this work, we suggest a critical preprocessing stage, using adaptive filtering technique for impulse and Gaussian noise removal performed on images captured by personal camera. We are presenting a comparative analysis of various filters commonly used in image pre-processing, such as mean, median, hybrid median, order or min-max filter, Wiener filter, and adaptive bilateral filter (ABF). Our study involved adjusting filter parameters to assess image quality and adaptability. The results showed that by cascading filters and optimizing the algorithm, we achieved superior image quality. Key performance metrics, including Signal to Noise Ratio (SNR) and Structural Similarity Index (SSIM), supported these findings. We demonstrate how adaptable filtering techniques benefit image processing in IoT applications.
  Keywords: Image Processing; IoT- Internet of Things; SNR-Signal-to-Noise Ratio; HMF-Hybrid Median filter; ABF- adaptive bilateral filter.
  DOI: 10.1504/IJIPT.2026.10077343
   
  
• A Study on Internet Protocol-Driven Digital Transformation of Higher Education using Artificial Intelligence artificial intelligence  
  by Shaoyun Yang, Yubo Liu 
  Abstract: Recent advances in internet protocol (IP) networks and artificial intelligence (AI) are impacting the infrastructure of higher education to a great extent. The purpose of this research will be to investigate what IP-based, AI-enabled service architectures exist, and how network performance, scalability and QoS can be optimised. The proposed multi-layer Smart Campus architecture will consist of access, edge and cloud layers, and the simulation results will demonstrate that QoS-based IP routing via edge-enabled AI will have lower latency and more stable networks than traditional, centralised systems. The use of adaptive and intelligent services also poses challenges to scalability and resiliency due to issues such as congestion, security, and privacy; thus, protocol-level solutions must be implemented for educational networks to become scalable and resilient.
  Keywords: Internet Protocol Technologies; Digital Transformation; Artificial Intelligence in Education; Higher Education Systems; Network Architecture.
  DOI: 10.1504/IJIPT.2027.10077559
   
  
• Internet of Things and Deep Learning Applications for Ocean Tourism Economy: a Smart Growth Model for Ocean Hub Cities  
  by Yu Chen, Lu Ke, Yibo Wang 
  Abstract: With an emphasis on coastal hub towns, this research explores how deep learning and the Internet of Things (IoT) may propel sustainable development in the maritime tourism industry. It looks at collaborative urban development strategies that use network linkage, dual-core interconnectivity, and core-edge synergy to facilitate resource integration, cross-sector cooperation, and market complementarity. Using big data analytics, deep learning-based prediction models, and IoT-enabled real-time data collecting, a multi-layered smart architecture is suggested to maximise the distribution and use of maritime tourist resources. Through automated decision-making and behavioural analysis, the intelligent system enables targeted marketing, predicted demand management, and personalised travel experiences. In addition to improving visitor engagement and operational efficiency, the results show that integrating IoT and deep learning creates a technological basis for intelligent marine tourism ecosystems, establishing coastal cities as innovation-driven centres in the nascent digital blue economy.
  Keywords: Internet of Things (IoT); Deep Learning; Smart Tourism; Ocean Tourism Economy; Coastal Hub Cities; Intelligent Systems; Big Data Analytics; Smart Infrastructure; Digital Blue Economy.
  DOI: 10.1504/IJIPT.2027.10077880
   
  
• An edge-computing-enabled internet protocol framework for personalised learning systems: a music composition use case  
  by Hua Bai 
  Abstract: This paper proposes an edge-integrated Internet Protocol framework (IP-ECF) to address latency and scalability issues in personalised learning applications. By shifting computation and orchestration to edge nodes, the framework enables traffic prioritisation, adaptive resource management, and local processing, thereby reducing backbone congestion and round-trip delay. It integrates the HTNKT model to estimate learners real-time states and employs a reinforcement learning-based recommendation engine formulated as a partially observable Markov decision process (POMDP) for adaptive content delivery. Evaluation using a music composition learning platform shows that the proposed approach significantly reduces response latency, improves throughput, lowers jitter, and enhances prediction accuracy. Results demonstrate that IP-ECF provides a reliable and efficient architecture for supporting adaptive, latency-sensitive applications over modern Internet infrastructures.
  Keywords: edge computing; knowledge tracing; hypergraph neural network; self-attention mechanism; internet protocol; network protocol design.
  DOI: 10.1504/IJIPT.2027.10078023
   
  
• Edge-IoT Architecture for Real-Time Cross-Border Safety Monitoring Using Light-Based Tracking in High-Altitude Tower Construction  
  by Xianhua Liu, Yongsheng Huang, Zhe Zhang, JinYan Zhang 
  Abstract: Building towers at a high altitude poses several difficulties in terms of worker safety, mobility, and real-time monitoring. In this context, the present study advocates the use of an Edge-IoT IP-based LightTrack system using wearable sensors, light trackers and edge computing nodes to achieve low latency safety monitoring. This would allow workers to be tracked in real time, to assess their movement and to detect incidents, while also decreasing the load on the IP backbone. When compared with conventional cloud based IoT monitoring systems, the proposed system can improve latency; reliably send packets; provide more bandwidth efficiency and increase the scalability of monitoring capabilities. The evidence provided supports the assumption that the proposed LightTrack system is suitable for use in safety critical industrial IoT applications at dispersed high altitude construction sites.
  Keywords: Edge Computing; Internet of Things (IoT); Internet Protocol Technology; Real-Time Monitoring; Cross-Border Communication; LightTrack System; IP-Based Networking; Low-Latency Communication.
  DOI: 10.1504/IJIPT.2027.10078455
   
  
• An Ensemble XGBoost-Based Framework for IoT Botnet Detection Using SMOTE and Recursive Feature Elimination  
  by Hiren Raithatha, Pratik Patel, Amin Shaikh, Pathan Bilalkhan Raufkhan 
  Abstract: The exponential growth of Internet of Things (IoT) devices has led to increased exposure to botnet-driven cyberattacks, underscoring the critical need for effective detection strategies. Despite various advancements, current detection techniques frequently exhibit limitations in classification accuracy, computational efficiency, and generalisation to real-world environments. A robust framework is essential to bridge these gaps and enable scalable, real-time threat detection. This study introduces an optimised ensemble model based on Extreme Gradient Boosting (XGBoost) for accurate and efficient detection of IoT botnet attacks. The proposed framework incorporates Recursive Feature Elimination (RFE) to select the most informative features and employs the Synthetic Minority Over-sampling Technique (SMOTE) to resolve class imbalance, thereby enhancing overall model robustness. Evaluation on the widely recognised Bot-IoT dataset reveals outstanding performance, with accuracy, precision, recall, and F1-score each approaching 99%. The model also demonstrates reduced computational overhead during both training and inference. These results highlight the models potential for real-world IoT security applications and offer a substantial improvement over traditional detection method.
  Keywords: IoT security; Botnet attacks; Extreme Gradient Boosting; Classification; Machine learning; Evaluation.
  DOI: 10.1504/IJIPT.2027.10078497
   
  
• A Network-Aware Virtual Reality Art Intervention Framework for Real-Time Mental Health Support Using Internet Protocol-Based Systems  
  by Ke Li 
  Abstract: As mental health concerns among university students increase, scalable technologysupported interventions are needed. This study implemented the Network-Aware Virtual Reality Art Intervention Framework (VRAI-P), a network-enabled VR art therapy system for university counselling settings. The framework integrates immersive VR sessions delivered through headmounted displays with Internet Protocol-based real-time data transmission and monitoring, enabling adaptive support during intervention. Participants with mild to moderate psychological distress completed structured VR-based art sessions, and stress, anxiety, and depression outcomes were assessed using established psychological scales at two time points. Statistical analyses showed improved psychological well-being after the intervention, particularly reduced stress and anxiety. Compared with stand-alone VR approaches, VRAI-P offers greater scalability, remote accessibility, and real-time responsiveness. The findings suggest that network-enabled VR art therapy may be a promising approach for supporting student mental health in higher education.
  Keywords: Virtual Reality Art Intervention (VRAI-P); Internet Protocol-Based Systems; Network-Aware Healthcare Systems; Real-Time Mental Health Support; Virtual Reality Therapy; University Student Mental Health.
  DOI: 10.1504/IJIPT.2027.10079173
   
  
• AIoT-Driven Real-Time Detection of Mitral Valve Prolapse  
  by Meng-Han Chen, Guanling Lee 
  Abstract: The incorporation of Artificial Intelligence of Things (AIoT) into health care makes scalable, real-time diagnosis possible with smart sensors embedded and lightweight communication protocols. In this study, an AIoT-based early warning system for Mitral Valve Prolapse (MVP) from electrocardiogram (ECG) signals is proposed. Heart rhythm information is continuously harvested using IoT sensors and transported using the Message Queuing Telemetry Transport (MQTT) protocol to the central processor for computation. After denoising, significant features like P-wave duration, T-wave duration, R-R interval, and S-T interval are extracted and subjected to seven AI-based detectors. The highest accuracy (0.965) was achieved by the Long Short-Term Memory (LSTM) model. Feature importance analysis revealed T-wave duration to be the key predictor. The system combines low-latency data transport, advanced signal processing, and explainable AI to facilitate timely and accurate MVP screening. The result confirms the feasibility of AIoT-based cardiovascular monitoring for clinical and telehealth applications.
  Keywords: Artificial Intelligence of Things (AIoT); Internet of Things (IoT); Machine Learning Algorithm; Mitral Valve Prolapse; Message Queuing Telemetry Transport (MQTT) Protocol.
  DOI: 10.1504/IJIPT.2027.10079642
   
  
• Improving stock price forecasting accuracy on the Shanghai Stock Exchange utilising hybrid LSTM and BRO algorithm  
  by Mingbo Zhang 
  Abstract: The stock market is one of the most reliable indicators of the condition of a countrys economy, but it has been a challenging endeavour to accurately forecast stock prices since the nature of market volatility and complex trends made it so. This study attempts to bridge the gap in stock price prediction models by proposing a new hybrid model of forecasting that combines the application of Long Short-Term Memory (LSTM) networks with optimisation techniques. Specifically, the present research explores predictive accuracy optimisation through the application of biological-based optimisation (BBO), antlion optimisation (ALO), and Battle Royale Optimisation (BRO) algorithms. The efficiency of the suggested method is demonstrated within this research through historical data of the Shanghai Stock Exchange (SSE) from 2015 to 2023. The results indicate that the hybrid BRO-LSTM model achieves a coefficient of determination (R2) of 0.993, achieving better accuracy prediction compared to other models. The study shows the potential of optimisation-based LSTM models in enhancing stock price forecast accuracy, creating a more precise instrument for financial analysts and investors. The study adds to the construction of forecast techniques and emphasises the necessity of hybrid models to tackle complicated prediction problems in the financial market.
  Keywords: Shanghai Stock Exchange; Financial market; Optimization Algorithms; long short-term memory (LSTM); BRO; Investment Risk and Return; Economic Indicators; Stock anticipation.
  DOI: 10.1504/IJIPT.2027.10080365
   
  
• A Survey of Protocol-Aware IoT Architectures in Healthcare: Comparative Analysis for Immunological Decision Support  
  by Meng-Han Chen, Guanling Lee, Lun-Ping Hung 
  Abstract: Immunological decision support depends on continuous physiological monitoring, but the role of the underlying communication protocol in shaping Machine-Learning (ML) validity is still rarely discussed. We argue that protocol behavior handshake cost, retransmission strategy, and Quality of Service (QoS) semantics directly affects data fidelity and, ultimately, clinical safety. We introduce a protocol-aware reference architecture that links sensing modalities to clinical back-end services through Internet-native messaging. We then examine how end-to-end traffic evolves across heterogeneous access networks, emphasizing how transport dynamics and congestion control influence the timing of medical data streams. Using Message Queuing Telemetry Transport (MQTT), Constrained Application Protocol (CoAP), and HTTP/REST as representative choices, we compare latency, reliability, and energy under both nominal conditions and stressed links with 10% packet loss. Finally, we provide workload-oriented recommendations and a reporting checklist for protocol-aware evaluation to support trustworthy, Internet-native healthcare Internet of Things (IoT) deployments.
  Keywords: Healthcare IoT; Internet protocols; immunological decision support; MQTT; CoAP; Quality of Service (QoS); protocol-aware machine learning.
  DOI: 10.1504/IJIPT.2027.10080626
   
  
• WPWaO_QDCNN: Water Plant Walruses optimisation-based Quantum dilated Neural Network for Intrusion Detection in Internet of Things  
  by Shwetambari Chiwhane, Srinivas Ambala, Kapil Netaji Vhatkar, Geeta S. Navale, Vidula V. Meshram 
  Abstract: This work proposes a hybrid framework named WPWaO_QDCNN for efficient routing and intrusion detection in IoT environments. Initially, an IoT network is simulated, and routing is optimised using the Weighted Particle Warbler Optimization (WPWaO) algorithm based on a multi-objective fitness function. At the base station, Minmax normalisation is applied to the collected data. Then, feature selection is done using Ruzicka similarity and data augmentation via Borderline Synthetic Minority Over-Sampling Technique (SMOTE). The optimised Quantum Dilated Convolutional Neural Network (QDCNN), trained using WPWaO, is employed for intrusion classification. Experimental evaluation is conducted using the NSL-KDD and BoT-IoT datasets. The proposed WPWaO achieves a routing distance of 36.57m, residual energy of 0.295J, and a detection rate of 94.45%. The WPWaO_QDCNN further attains an accuracy of 92.34%, a True Positive Rate (TPR) of 90.72%, and a True Negative Rate (TNR) of 91.46% for intrusion detection.
  Keywords: Internet of Things; Routing; Intrusion detection; Quantum dilated Convolutional Neural Network; Walruses Optimisation Algorithm.
  DOI: 10.1504/IJIPT.2027.10080915
   
  
• Trustworthy 6G Networks: Blockchain-Based Federated Learning with Deep Spinal Forward Harmonic Net  
  by Sushma T. Shedole, Dharamendra Chouhan 
  Abstract: In this research, the DSFHNet is devised for Blockchain-based Trust Management for Trustworthy 6G Networks in FL. Here, the FL framework comprises different entities, such as 6G user, server, Blockchain and BSs. The end user generates the running data and synchronizes their data with the corresponding BSs. The BS is in charge of training the models and the training is performed employing DSFHNet. Moreover, the trust must be an integral part of the decision-making process. Hence, the trust establishment mechanism is modeled between the server and users by considering trust. Finally, the local updation and aggregation at the server is conducted based on the Average method. It is noted that the developed DSFHNet has accomplished an FPR of 0.091, MSE of 0.337, TPR of 90.88%, RMSE of 0.581, an accuracy of 91.00%, bandwidth of 13.280 GHz, communication overhead of 143.400 KB, and latency of 0.350 ms.
  Keywords: Federated Learning; Blockchain; trust management; Deep Neural Network; SpinalNet.
  DOI: 10.1504/IJIPT.2027.10080980
   
  
• Enhancing Student Dropout Prediction in Higher Education Through Supervised Machine Learning and Metaheuristic Optimisation  
  by Tao Yan 
  Abstract: Student dropout is a persistent challenge in higher education, leading to significant academic, economic, and institutional consequences. Despite growing efforts, many predictive systems lack the adaptability and accuracy needed to proactively identify at-risk students. This study addresses this gap by proposing two hybrid frameworks that integrate supervised machine learning models with advanced metaheuristic optimisation techniques to enhance dropout prediction performance. Specifically, Gradient Boosting Classification (GBC) and Support Vector Classification (SVC) were employed to classify students based on dropout risk. To optimise model hyperparameters and improve predictive capability, two nature-inspired algorithms Black Widow Optimisation Algorithm (BWOA) and Mountain Gazelle Optimiser (MGO) were applied, resulting in four hybrid configurations: HGBW, HGBC, SVM-BWOA, and SVM-MGO. The experimental evaluation was conducted on a real-world higher education dataset, and model performance was assessed using key classification metrics such as accuracy and precision. Among the models, the HGBC (Hybrid GBC with MGO) framework demonstrated superior performance, achieving an accuracy of 0.981 and a precision of 0.983. Comparative analysis confirmed the efficacy of the metaheuristic-tuned models over non-optimised counterparts. These findings underscore the potential of hybrid MLmetaheuristic approaches in developing intelligent earlywarning systems for student retention. By identifying students at risk with higher precision, educational institutions can implement timely, data-driven interventions, thereby enhancing student success and institutional outcomes.
  Keywords: Student performance; Dropout; Machine learining; Hybird model; Histogram gradient boosting; Support vector classification; Black widow optimization algorithm; Mountain gazelle optimizer.
  DOI: 10.1504/IJIPT.2027.10081068
   
  
• Federated Autonomous Cybersecurity Agents for Privacy-Preserving Threat Intelligence Sharing Across Internet Protocol Networks  
  by Yangyang Xiang, Debao Zhao 
  Abstract: As IP-based networks continue to expand, there is a growing requirement for collaborative cybersecurity, allowing detection of distributed threats to be carried out without compromising sensitive operation details. The current work presents a framework for privacypreserving federated cybersecurity agents that autonomously monitor local network traffic, extract relevant security information, perform anomaly detection, classification, risk assessment, and produce local threat intelligence. Rather than disclosing sensitive network packets or security logs, agents transfer model updates and anonymous threat indicators to the federated coordinating server which effectively brings them together and provides a global model update to all agents. The effectiveness of the framework is evaluated using metrics such as accuracy, precision, recall, F1-score, latency, communication overhead, scalability, and privacy protection. The outcomes indicate enhanced effectiveness of collaborative detection while also keeping control of local data, yet larger-scale and resource-limited validation of the framework is still needed.
  Keywords: Federated Learning; Autonomous Cybersecurity Agents; Threat Intelligence Sharing; Privacy Preservation; Internet Protocol Networks; IP-Based Communication.
  DOI: 10.1504/IJIPT.2027.10081455