International Journal of Sensor Networks (42 papers in press)

Regular Issues

• Optimizing Power Management in Wireless Sensor Networks Using Machine Learning: An Experimental Study on Energy Efficiency  
  by Mohammed Amine Zafrane, Ahmed Ramzi Houalef, Miloud Benchehima 
  Abstract: Wireless sensor networks (WSNs) have emerged as essential components across various fields. Comprising small, self-sustaining devices known as "Nodes," they play a critical role in data collection and analysis. However, ensuring optimal longevity without compromising data collection timeliness is a fundamental challenge. Regular data aggregation tasks, while essential, consume substantial energy resources. Furthermore, constraints in computation power, storage capacity, and energy supply pose significant design challenges within the Wireless Sensor Network domain. In pursuit of optimizing energy efficiency and extending the operational lifetime of nodes through artificial intelligence, we have developed a prototype for data collection to create a comprehensive dataset. Our approach leverages both current and precedent measurements, triggering data transmission only in the presence of significant changes. This intelligent strategy minimizes unnecessary communication and conserves energy resources. Based on the
  Keywords: Artificial intelligent; WSN; Power optimization; data acquisition.
  DOI: 10.1504/IJSNET.2024.10068162
   
  
• A Path Optimization Method Based on Dynamic Clustering Strategy and Nondominated Sorting Genetic Algorithm II in Wireless Sensor Networks  
  by Liying Zhao, Jin Zhu, Chao Liu, Yu Wang, Sinan Shi, Chao Lu, Qi Luan 
  Abstract: The traditional wireless sensor network data transmission path selection often considers only single-objective optimization, such as energy consumption or transmission delay, which leads to problems such as unbalanced node load and insufficient path reliability. Therefore, this study proposed a wireless sensor network routing optimization method that integrates a dynamic clustering strategy with the nondominated sorting genetic algorithm II. First, the network nodes of the wireless sensor network were divided into clusters of varying scales using the MiniBatchKMeans method. Then, the residual energy of nodes, their distance to the cluster center, and historical load were comprehensively evaluated to elect a cluster head for each cluster. Subsequently, the nondominated sorting genetic algorithm II algorithm was employed to generate Pareto-optimal paths, with the objective functions encompassing minimization of energy consumption, reduction of transmission delay, and maximization of signal strength (Received Signal Strength Indication).
  Keywords: Clustering strategy, nondominated sorting genetic algorithm II, path optimization, wireless sensor network.
  
  
• A Survey on the recent methodologies of Secure Social Internet of Things (SIoT)  
  by Divya S, Tanuja R, Manjula S. H. 
  Abstract: The integration of social networks with internet of things has led to the emergence of the Social Internet of Things (SIoT), where intelligent objects interact socially to discover and share services. However, unreliable nodes can disrupt operations by spreading malicious information, compromising service quality and trust. This survey provides a comprehensive review of SIoT approaches focusing on trust management, security, and privacy. It outlines a generic SIoT framework, categorises existing research, and highlights key challenges and gaps to guide future developments. The analysis reveals that most studies employ trust management techniques with network simulators and social network datasets for evaluation, using trust score and accuracy as primary performance metrics.
  Keywords: Social Internet of Things (SIoT); trust management; privacy preservation; Deep Learning (DL); Attack detection.
  DOI: 10.1504/IJSNET.2025.10074737
   
  
• Optimizing the Adaptive Design of Wireless Sensor Networks Using an Enhanced Particle Swarm Optimisation Algorithm  
  by Mina Mirhosseini, Fereshteh Forouzesh, H. Nezamabadi-pour, Ali Darijani 
  Abstract: One of the key challenges in wireless sensor networks (WSNs) is reducing energy consumption to enhance lifespan and performance. In this paper, an improved Particle Swarm Optimization algorithm, called NBPSO, is adapted for optimal WSN design under both static and dynamic optimal design frameworks. The algorithm reduces energy consumption, extends network lifespan, and satisfies communication constraints and application requirements. Each sensor operates in one of four states: cluster head, active with long range, active with short range, or inactive. Simulation results show that NBPSO outperforms Genetic Algorithm, Particle Swarm Optimization, and Quantum Gravitational Search Algorithm, achieving significantly better network lifetime. To improve scalability, a multi-sink deployment strategy is introduced and evaluated, demonstrating NBPSO's practicality for larger networks. In addition, a brief sensitivity analysis of the NBPSO parameters has been conducted to examine their influence on performance.
  Keywords: Particle Swarm Optimization; Wireless sensor networks; energy optimization; adaptive design.
  DOI: 10.1504/IJSNET.2025.10075663
   
  
• Experimental Demonstrations of Visible Light Communication Using Quantum Noise for High Security  
  by Ning Xiao, Sihui Chen, Feifei Chen, Shuai Shi 
  Abstract: This paper reports on a symmetric-key direct-data encryption technique that ensures the security of optical signals at the physical layer in a visible light transmission system. Security is achieved by masking the signal with quantum noise. Encryption with quantum noise masking is realized by converting the data into a multilevel optical signal of intensity modulation. The effect of the quantum noise masking is easily superimposed on the ciphertext signal, which also fully uses the inevitable characteristics of quantum noise. We have demonstrated experimentally a 100Mbit/s visible light cipher system with quantum noise masking. The visible light transmission system implements high security with adequate quantum noise masking.
  Keywords: quantum noise; physical layer security; encryption; intensity modulation; visible light communication.
  DOI: 10.1504/IJSNET.2025.10075734
   
  
• Aerial Muling As a Service For Terrestrial Wireless Sensor Networks  
  by Soumaya B.E.L. HadaJ Youssef, Slim Rekhis 
  Abstract: In this paper, we propose a system architecture based on the use of multi-broker public cloud of unmanned aerial vehicles (UAVs) to offer a muling as a service (MuaaS) to largely deployed terrestrial wireless sensor networks (WSNs). The proposed system is designed to coordinate the UAVs arrival and manage their mobility during the collection of sensed data from terrestrial data collectors. To address this, we describe the muling data problem and present an algorithm designed to achieve minimal data delivery delay. In addition, we address a grid-based WSN deployment, enabling the computation of model parameters. Therefore, a mathematical expression for minimising data delivery delay is provided. Furthermore, we present a scheme for selecting the best UAV broker offer based on a decision-making criteria that consider minimisation of data delivery delay, UAV energy consumption, and cost of delivery service. Finally, a simulation is conducted to evaluate the performance of the solution.
  Keywords: Wireless sensor network; Unmanned Aerial Vehicle Network; cloud computing; Muling As A Service; data delivery delay minimisation.
  DOI: 10.1504/IJSNET.2025.10076118
   
  
• SGSCP: Subnetwork-based Grid Search Cooperative Positioning for Irregular Wireless Sensor Networks  
  by Xingsheng Xia, Yilin Xia, Xiaoyong Yan, Zhi-E Lou, Jiajia Yan 
  Abstract: Wireless sensor networks are widely deployed in complex and unstructured environments, where accurate node location information is fundamental for network operation and data interpretation. In practical scenarios, obstacles, signal attenuation, and uneven node distributions often lead to irregular network topologies, under which conventional hop-based positioning methods suffer from degraded performance due to their reliance on homogeneous assumptions. To address this issue, a subnetwork-based grid search cooperative positioning (SGSCP) algorithm is proposed. The method allows beacon nodes to exchange hop information using the Bellman-Ford protocol and independently estimate local hop-distance characteristics. Each unknown node then constructs a constrained subnetwork composed of its four nearest beacons and performs a bounded grid search guided by hop similarity. Extensive simulations demonstrate that SGSCP achieves improved localisation accuracy, enhanced robustness to beacon density variations, and stable performance across diverse irregular topologies, outperforming existing methods in terms of median error and error dispersion.
  Keywords: Grid Search; Cooperative Positioning; Irregular Wireless Sensor Networks.
  DOI: 10.1504/IJSNET.2025.10076201
   
  
• NTOS: Nature-inspired Task Offloading Strategies in IoT-Fog Environment for Improved Response Time  
  by Ritarani Sahu, Aniket Ganvir, Suchismita Chinara 
  Abstract: The rapid growth of internet of things devices and the increasing demand for low-latency communication and high-performance applications have brought forth the emergence of fog computing as a viable solution. Fog computing takes advantage of the nearness of fog nodes to the edge of the network to offload compute-intensive tasks from resource-limited devices to fog nodes with the aim of optimising the response time and balancing the workload of fog nodes. The task offloading problem has been already proven to be NP-hard and it is very challenging to address this issue. The current work aims to handle this offloading problem by proposing two nature-inspired meta-heuristic optimisation algorithms: offloading algorithm-ant colony optimisation and offloading algorithm-artificial bee colony. Simulations has been performed considering various scenarios and it is observed that both the proposed algorithms perform wonderfully better as compared to Round Robin, grey wolf optimiser, particle swarm optimisation, and sparrow search algorithm, to achieve minimum average response time. Also simulations confirm that both algorithms are very good at maintaining the workload balance among fog nodes and quality of task offloading.
  Keywords: Internet of Things; compute-intensive; resource-limited; task offloading.
  DOI: 10.1504/IJSNET.2025.10076236
   
  
• Optimized Monocular Vision Ranging via Parallel-Line Feature Rectification  
  by Jingwen Qian, Jin Zhang, Kangwei Wang, Jie Sheng, Cheng Wu 
  Abstract: To address the degradation of distance estimation accuracy caused by perspective distortion in long-range imaging scenarios, this paper proposes a monocular vision ranging method via parallel-line feature rectification, termed PLFR. The method leverages the geometric regularity of parallel lines under perspective projection, using their pixel widths at different distances as stable reference scales to mitigate distortion-induced errors. Specifically, PLFR first employs an improved YOLOv5 to extract the position and size information of targets, and then integrates the geometric attributes of parallel lines to dynamically calibrate the model parameters via recursive least squares (RLS), enhancing accuracy and robustness in long-range scenarios. This strategy integrates object detection outputs with scene geometry information to effectively suppress distance estimation errors at extended ranges. Experimental results demonstrate that PLFR maintains an average ranging error within 5% over a 200-metre range. Notably, in the long-distance range of 150200 metres, the average error can be reduced to 2.68 m, significantly outperforming representative existing methods
  Keywords: long-range; object detection; monocular ranging; parallel elements.
  DOI: 10.1504/IJSNET.2025.10076395
   
  
• CIC-LQR Power Control Strategy for Reliable Wireless Communication in Cable Tunnel Inspection Robots  
  by Xue Liu 
  Abstract: With the expansion of urban power grids and the rapid development of underground infrastructure, the reliability of wireless communication for cable tunnel inspection robots faces major challenges. The tunnel environment, characterised by severe obstructions, strong electromagnetic interference, and strict energy constraints, makes it difficult for traditional power control methods to ensure both link stability and energy efficiency. To address these issues, this paper proposes a confidence interval compensation linear quadratic regulator (CIC-LQR) power control strategy. A state-space model is established based on the logarithmic path loss model, and a skew-normal distribution is employed to capture the skewness and heavy-tailed characteristics of received signal strength indicator (RSSI) samples. A confidence interval compensation mechanism is then introduced to dynamically adjust the reference RSSI, enhancing robustness against channel disturbances. Simulation and real-world tunnel experiments show that CIC-LQR significantly outperforms conventional MPC, LQR, LQG, and CI-MPC methods. Specifically, it reduces the proportion of samples falling below the required RSSI threshold to 2.8% (compared with 12.8% for LQR and 20.2% for MPC) and achieves a clear reduction in average transmit power. These results demonstrate that CIC-LQR enables reliable and energy-efficient wireless communication in harsh tunnel environments.
  Keywords: Cable Tunnel; Inspection Robot; Wireless Communication Reliability; RSSI; Skew-Normal Distribution;Power Control.
  DOI: 10.1504/IJSNET.2025.10076523
   
  
• Wavelet Entropy-Based Adaptive Compression of Transient Waveform Data in Distribution Networks  
  by Chenglong Ye, Wei Sun, Xinyu Qiu, Zhiqiang Zhang, Yue Li, Zhen Zhang 
  Abstract: Integration of distributed photovoltaic in distribution networks increases abnormal conditions, requiring efficient data compression for fault monitoring and early warning under limited communication bandwidth. In this case, it is necessary to strengthen the monitoring of distributed photovoltaic systems, which greatly increases the amount of data. However, the resource limitations of the distribution network require efficient compression of fault monitoring and early warning data within constrained communication bandwidth. Therefore, efficient compression of fault monitoring and early warning data is essential under limited communication bandwidth. Existing compression techniques treat transient waveform data as regular data, neglecting targeted compression. They fail to account for the energy characteristics of waveform data across different frequency bands critical for fault detection. To tackle this challenge, this study introduces an adaptive data compression and reconstruction method based on wavelet entropy. Initially, different types of transient waveform data are adaptively selected for wavelet transform using appropriate wavelet basis functions, enhancing the transform's adaptability to various data types.
  Keywords: Wavelet transform; Data compression; Photovoltaic.
  DOI: 10.1504/IJSNET.2025.10076639
   
  
• Collaborative Adaptive Pricing (CAP) Framework: a Dynamic Approach to AI Model Trading  
  by Milon Biswas, Yifan Guo, Weixian Liao, Wei Yu 
  Abstract: Artificial Intelligence (AI) model trading has become an emerging paradigm, enabling organizations and researchers to access high-quality pre-trained models. Traditional static pricing mechanisms often fail to align with dynamic market conditions, leading to inefficiencies in affordability and revenue generation. This paper introduces the Collaborative Adaptive Pricing (CAP) Framework, an innovative pricing strategy that incorporates adaptive noise injection, real-time pricing adjustments, and auction-based mechanisms to optimize revenue while ensuring accessibility. We present a mathematical formulation, experimental evaluation, and a scalability analysis demonstrating CAP's effectiveness over fixed and tiered pricing strategies. Our results show that CAP maximises revenue while maintaining buyer engagement and preventing market arbitrage. The framework is adaptable to various AI trading ecosystems, including federated learning and decentralised AI markets.
  Keywords: Collaborative Adaptive Pricing (CAP) Framework; AI Model Trading; Algorithm Design.
  DOI: 10.1504/IJSNET.2025.10076693
   
  
• A Differential Game-Based Security Mechanism for Protecting Data Elements Circulation  
  by Lei Wang, Mengmeng Li, Haitao Xu, Haoyu Shi, Xianwei Zhou 
  Abstract: In intelligent connected vehicles (ICVs), driver-users transmit sensitive information to roadside units (RSUs) to obtain value-added services. Attackers may compromise RSUs to exfiltrate sensitive data, thereby posing severe privacy threats to vehicular networks. Most existing defence mechanisms rely on passive protection strategies, which exhibit limited adaptability against dynamic attacks. Recent studies have applied differential game models to describe attacker-defender dynamics in vehicular networks. However, these works often assume static payoffs, adopt passive defence strategies, and lack explicit mechanisms linking defence strength to real-time packet behaviour, limiting their practicality in complex environments. To address this, this paper proposes a data-element circulation security mechanism based on differential game theory, where the RSUs health level is modelled as a contested resource. The proposed framework introduces packet-marking strength (PMS) as an active defence variable within a feedback differential-game formulation, enabling the defender to dynamically adjust PMS in response to variations in attack frequency. By deriving the feedback Nash equilibrium (NE), the defender can remove malicious packets, enhance RSUs health, and protect data privacy. Simulation results confirm that the proposed mechanism achieves greater adaptability and faster convergence than existing passive defence models, validating its theoretical soundness and practical effectiveness.
  Keywords: Intelligent connected vehicle; privacy protection; strategy optimization; data packet marking technology; differential game.
  DOI: 10.1504/IJSNET.2025.10076817
   
  
• A Univariate Time-series Anomaly Detection Method for Sensor Network based on Conditional Variational AutoEncoder  
  by Ping He, Zhifeng Wu, Sirui Hao, Hao Liu, Jiangchuan Chen, Xuefeng Song, Yipeng Liu, Junjiang He 
  Abstract: With the rapid advancement of Industry 4.0, anomaly detection secures critical infrastructure by monitoring and analysing sensor time-series data. However, existing anomaly detection methods are often suffering from incomplete temporal feature extraction and delayed detection responses. To overcome these issues, we introduce conditional variational AutoEncoder-based anomaly detection (CVAD). CVAD enhances temporal pattern learning by utilising an improved CVAE that jointly processes raw time-series data and their corresponding frequency-domain representations. To maintain robustness under concept drift, we implement a self-adjusting threshold mechanism that dynamically adjusts detection criteria, eliminating the overhead of model retraining. Experimental results on the SensorScope dataset, NASA-SMAP dataset, and NASA-MSL dataset demonstrate that the proposed model outperforms seven baseline methods in terms of the F1 score: compared with the second-best baseline TimesNet (89.45% on SensorScope, 82.99% on NASA-SMAP, 89.79% on NASA-MSL), CVAD achieves 93.35%, 93.45%, and 89.84%, respectively representing improvements of 3.9%, 10.46%, and 0.05% in F1 score.
  Keywords: Sensor Networks; Univariate Time-series Data; Anomaly Detection; Conditional Variational AutoEncoder.
  DOI: 10.1504/IJSNET.2025.10076827
   
  
• An Edge-Driven Based Sensor Selection and Sensory Data Prediction Model for Minimizing Energy Consumption in WSNs  
  by Vipin Maurya, Sumit Kumar, Ruchir Gupta 
  Abstract: Owing to the limited storage and battery power, wireless sensor nodes frequently encounter challenges in achieving long-term energy sustainability. This problem is exacerbated when all sensors are activated simultaneously, resulting in unnecessary energy consumption and a reduced network lifetime. Sensor selection can be a promising solution because it activates only a subset of sensors while keeping others in sleep mode, thereby conserving energy. Consequently, sleeping sensors result in missing parameter data, which can degrade sensing coverage. To align energy efficiency with adequate coverage, this paper proposes a sensor selection approach that selects the optimal sensors based on the available energy of sensor nodes and the normalised mutual information among sensors for sensing environmental parameters. The proposed model is theoretically analysed, and it is observed that the selected sensor set is stable and Pareto optimal. The sleep sensor parameters are predicted at the edge node using a temporal convolutional network-bidirectional long short-term memory (TCN-BiLSTM)-attention-based deep learning prediction model. Simulations are performed to validate the effectiveness of the proposed model. It is observed that the proposed approach reduces the sensing energy of the sensor node by 54%, and the prediction model reduces the error by 21% in the prediction of sleep sensor parameters.
  Keywords: Bidirectional-LSTM; Edge Computing; GPR; Normalized Mutual Information; Pareto Optimal; Sensor; Sensing Energy; Shapley Value; Temporal Convolutional Network (TCN).
  DOI: 10.1504/IJSNET.2026.10076996
   
  
• Enhancing Contact Tracing using Smartphone Sensors  
  by Paramasiven Appavoo, Aatish Chiniah, Roushdat Elaheebocus, Shehzad Jaunbuccus, Kavi Kumar Khedo, Anuja Meetoo-Appavoo, Raj Moloo, Avinash Utam Mungur, Leckraj Nagowah, Nitish Chooramun 
  Abstract: Contact tracing is used to identify and monitor individuals who may have potentially been exposed to a contagious disease. Several mobile applications exist to support contact tracers. The widely adopted underlying method is based on radio frequency, namely Bluetooth. However, this method does not tell anything about whether they were in the same room (possibly in contact) or in different rooms (not in contact). In this paper, the potentials of smartphones common sensors are investigated and leveraged to generate a privacy-preserving room signature. The latter allows the contact tracing system to effectively eliminate the otherwise false positives inherent in radio frequency technology for this purpose. The proposed system showed a high recall and precision when tested under multiple scenarios by project assistants. Further investigations and evaluations were performed and the outcomes were considered in the final prototype. With the latter, neither false positive nor false negative is expected.
  Keywords: contact tracing; sensors; smartphones; localization; privacy.
  DOI: 10.1504/IJSNET.2025.10077032
   
  
• Fault Diagnosis Technique for the Reel Bearing of a Cigarette Manufacturing Machine utilising Multi-Level Feature Fusion  
  by Xinyan Chi, Shanliang Liu, Zhao Li, Xuan Xie 
  Abstract: The reel bearings operational state in a cigarette-making machine is crucial for both machine stability and product quality. To accurately assess its health, this paper introduces a multi-sensor monitoring system. It utilises various sensors, including vibration, acoustic emission, and temperature sensors, to collect real-time data on the reel bearings five operational states: normal, inner ring fault, outer ring fault, rolling element fault, and compound fault. To tackle noise interference in multi-source signals and address heterogeneous feature distribution under complex conditions, wavelet decomposition is used. This technique extracts multi-scale time-frequency features to enhance the sensor datas multi-level features and eliminate redundancy. At the modelling level, this approach facilitates the development of a multi-channel feature modelling structure that integrates convolutional neural networks, long short-term memory networks, and transformers. This structure captures local patterns, global temporal dependencies, and long-range interactions, thereby improving the expression and discrimination of fault features in sensor data. Experimental results demonstrate that the multi-level feature fusion-based bearing fault diagnosis method achieves an accuracy of up to 99.73%, approximately 1.5 percentage points higher on average than other typical algorithms. This method shows significant potential for diagnostic accuracy and operational reliability in complex sensor network environments.
  Keywords: cigarette manufacturing machine; reel bearings; Multi-source sensors; Wavelet decomposition; Multi-level feature fusion; Bearing fault diagnosis.
  DOI: 10.1504/IJSNET.2025.10077062
   
  
• Fault Prediction and Diagnosis of Programmable Logic Controller Control Units via Multi-Sensor Data  
  by Lixia Nan, Xinni Hao 
  Abstract: Ensuring the reliable operation of programmable logic controller control units is essential for industrial system safety. Traditional diagnostic approaches, which rely on manual inspection and delayed response, prove inadequate for the predictive maintenance requirements of smart manufacturing. This study tackles challenges in multi-sensor data fusion and model adaptability by developing an intelligent framework that combines digital twin technology with meta-learning. The approach integrates vibration and current signals through an attention-based fusion network, enabling rapid adaptation to new equipment with minimal data samples. Experimental results demonstrate exceptional performance: a Matthews correlation coefficient of 0.965 for fault classification and a root mean square error of 0.048 for remaining useful life prediction. These results significantly surpass current state-of-the-art methods, improving diagnostic accuracy by over 3.4% and prediction precision by more than 17% compared to the best existing baseline.
  Keywords: programmable logic controllers; fault prediction; digital twins; meta-learning; multi-sensor fusion.
  DOI: 10.1504/IJSNET.2025.10077193
   
  
• A review of beaconing schemes for routing protocols in flying ad-hoc networks  
  by Vikramjit Singh, Krishna Pal Sharma, Harsh Kumar Verma 
  Abstract: Flying ad-hoc networks (FANETs) play a pivotal role in a wide range of military and civilian applications. The effectiveness of communication in these networks is largely determined by routing protocols, which rely heavily on beaconing schemes for neighborhood discovery. Beaconing involves the periodic exchange of beacons or hello messages to maintain an up-to-date view of neighboring nodes. Inefficient beaconing can result in outdated or inaccurate neighbor information, leading to route failures, increased latency, and significant data loss. Despite its significance, beaconing has received limited focused attention in prior reviews, which have primarily concentrated on routing protocols as a whole. This paper addresses this gap by presenting a comprehensive review of existing beaconing schemes specifically designed for FANETs. We propose a new taxonomy based on operational strategies and conduct a detailed comparative analysis using various evaluation criteria. Finally, we outline key open research challenges and future directions to guide the development of reliable and efficient beaconing systems for FANETs.
  Keywords: UAV; FANET; Beaconing; Routing; Hello message.
  DOI: 10.1504/IJSNET.2025.10077432
   
  
• Deep Reinforcement Learning-based Secure and Reliable Opportunistic Routing for VANETs  
  by Huibin Xu 
  Abstract: The dynamic nature of vehicular ad hoc networks (VANETs), characterised by rapid topological changes and adversarial vulnerabilities, presents critical challenges in ensuring reliable and secure packet transmission. This work introduces an adaptive secure and reliable relay selection (ASRS) framework designed to optimise routing performance through a multi-criteria decision-making paradigm. By integrating mobility factor, packet forwarding ratio, and link duration, ASRS constructs a robust candidate relay node (CRN) set to maximise packet delivery ratio (PDR) while minimising end-to-end delay (EED). The relay selection process is formalised as a Markov decision process (MDP) and solved via a double deep Q-network (DDQN) architecture, enabling adaptive learning of optimal routing policies through continuous environmental interaction. The framework employs historical behavioral analysis to dynamically isolate malicious nodes executing black hole (BHA) and grey hole (GHA) attacks, while optimising exploration-exploitation tradeoffs through cyclic learning rate scheduling (OneCycleLR) and -decay mechanisms. Comprehensive simulations demonstrate ASRSs superiority over state-of-the-art benchmarks. These results validate the frameworks capacity to balance security, reliability, and efficiency in highly dynamic vehicular environments.
  Keywords: Vehicular Ad Hoc Networks; opportunistic routing; deep reinforcement learning; black hole attack; gray hole attack.
  DOI: 10.1504/IJSNET.2025.10077459
   
  
• Machine Learning-Based Sensor Drift Detection for Precision Agriculture IoT applications  
  by Munawar Hussain, Muhammad Ibrahim, Rab Nawaz Bashir, Rehan Ashraf, Uzair Khan, Muhammad Huzaifa Bin Nasir, Tanzila Saba 
  Abstract: Sensor drift are variations in sensor values that affect the performance of internet of things applications. Early detection and correction of drift are essential to avoid performance degradation. The influence of the sensing environment still requires investigation. This study examines the effect of soil characteristics on sensor drift and explores machine learning-based methods for its detection and correction. The proposed approach is implemented on capacitive soil moisture sensors under varying soil electrical conductivity (EC), pH, and temperature conditions. Results indicate that drift increases with higher EC and temperature and decreases at neutral pH. These observations are used to train artificial neural network, long short-term memory, support vector regressor, and gradient boosting regressor (GBR) models. Among them, GBR achieved the best performance with an R2 of 0.83, MSE of 2.68, RMSE of 1.63, and MAE of 1.27. SHAP analysis identifies EC as the most influential factor, followed by temperature and pH.
  Keywords: Sensor drift; Internet of Things (IoT) applications; Capacitive soil moisture sensors; Soil Electrical Conductivity (EC); Potential Hydrogen (pH); Soil temperature,.
  DOI: 10.1504/IJSNET.2025.10077563
   
  
• Synchronized Multimodal Physiological Sensor Networks for Emotional State Mapping in Interrogation Environments  
  by Ning Luo 
  Abstract: This study confronts two principal obstacles in achieving objective psychological assessment within interrogation settings. The first is the substantial disparity between controlled laboratory models and real-world application contexts, and the second is the critical synchronization challenges inherent in wireless, heterogeneous physiological sensor networks. To address these issues, this study introduces an integrated technical framework. A novel synchronization protocol for multi-parameter physiological sensing networks is proposed to achieve high-precision temporal alignment across heterogeneous data streams, attaining a mean synchronization error of 1.2 milliseconds. Furthermore, a hierarchical emotional-state mapping model is developed to interpret complex, interrogation-like stress states from synchronized multimodal physiological signals. Rigorous evaluation on a reconstructed benchmark dataset confirms the framework’s efficacy, with the proposed model achieving a state recognition accuracy of 92.3%. This result significantly surpasses the performance of current state-of-the-art baseline models.
  Keywords: interrogation environment; data synchronization; emotional state mapping; wireless body area network; stress recognition.
  DOI: 10.1504/IJSNET.2026.10077628
   
  
• Privacy-Preserving Computation Offloading for Intelligent Connected Vehicles via a Joint-Objective Alternating Multiplier Method  
  by Lei Wang, Fei Cheng, Liang Li, Haitao Xu, Haoyu Shi, Yueqiang Xu, Xianwei Zhou 
  Abstract: The paper investigates a privacy-preserving computation offloading strategy in a vehicular edge cooperative (VEC) computing system. The system consists of multiple task vehicles (TaVs) and edge computing nodes (CNs) that collaboratively execute delay-sensitive tasks. To address the joint optimisation of task offloading, delay performance, and privacy protection, a joint-objective alternating multiplier method algorithm (JO-AMMA) is proposed. The model incorporates a differential privacy perturbation mechanism to resist inference attacks while maintaining service efficiency. The proposed algorithm constructs a weighted objective function that jointly considers latency and privacy costs. Through alternating optimisation and multiplier updates, it determines the optimal offloading ratio and privacy-preserving intensity under nonlinear system constraints. Simulation results demonstrate that the proposed approach achieves faster convergence and a lower total system cost than benchmark algorithms, effectively reducing latency while ensuring differential privacy. Overall, the proposed approach provides an efficient and scalable solution for secure computation offloading in intelligent connected vehicles (ICVs).
  Keywords: Computation offloading; Differential privacy (DP); Vehicular edge computing (VEC); Intelligent connected vehicles (ICVs); Joint-objective optimization.
  DOI: 10.1504/IJSNET.2026.10077707
   
  
• Fusing Multi-Source Sensor-Derived Microclimate Data for Thermal Comfort Mapping of Urban Blocks  
  by Hang Gao, Yan Chen 
  Abstract: Precise construction of thermal comfort maps is of significant importance for sustainable urban development. However, microclimate data from a single source struggles to capture the thermal conditions of urban blocks fully. To address this issue, this paper introduces sensor network technology by deploying various micro-sensor nodes across urban blocks to collect multi-source microclimate data in real time. Compressed observation is applied to the data sensed by each node, with data reconstruction performed at aggregation nodes to reduce node communication volume fundamentally. Building upon this foundation, a thermal comfort map for urban neighborhoods is constructed. The traditional graph attention network is extended to a multi-layer architecture. Experimental results demonstrate that the proposed method achieves at least a 25% improvement in data reconstruction success rate and a 3.2% increase in map construction accuracy, validating its effectiveness.
  Keywords: thermal comfort in urban block; graph construction; sensor network; multi-source microclimate data; graph attention network.
  DOI: 10.1504/IJSNET.2026.10077870
   
  
• Integrating Time-Series InSAR with Lightweight Sensor Networks for Subsidence Monitoring in Black Soil Wetlands  
  by Yiyang Zheng, Xiangxi Lv, Jinbao He, Chen Zhen 
  Abstract: Synthetic aperture radar interferometry can monitor subsidence in black soil wetlands, but its observation cycle limitations hinder the acquisition of critical environmental parameters. To address this, this paper leverages the real-time data-collection capabilities of wireless sensor networks to establish a subsidence early warning system for black soil wetlands that integrates time-series synthetic aperture radar interferometry with wireless sensor networks. The system employs an energy-balanced routing algorithm based on ant colony optimization to ensure stable operation of the sensor network in wetland environments. Building upon this foundation, a Kalman filter fusion algorithm deeply integrates real-time sensor data with deformation information. Combined with variational modal analysis and long short-term memory networks, this approach enables early warning of subsidence trends. Experimental results demonstrate that the proposed method achieves a packet transmission success rate of 98.9% and a mean absolute error of only 0.602 mm, significantly enhancing the accuracy of early warning.
  Keywords: time-series synthetic aperture radar interferometry; wireless sensor network; data fusion; black soil wetland subsidence; early warning.
  DOI: 10.1504/IJSNET.2026.10077871
   
  
• Diffusion-Based Reconstruction of Sensor-Collected Motion Data for Cinematic Production  
  by Ziwei Zang, Youdong Ding 
  Abstract: Film motion capture sensor networks face challenges like heterogeneous sensor noise, data inconsistency, and spatiotemporal misalignment, requiring precise, real-time restoration. Traditional methods struggle with dynamic, high-dimensional sensor data due to rigid noise modelling and insufficient consistency guarantees. To address low restoration accuracy, poor real-time performance, and weak anti-interference, this study proposes a sensor network-aware diffusion model. It integrates the distributed architecture of sensor networks with the probabilistic generation capability of diffusion models, enabling collaborative optimisation of local data restoration and global consistency through edge-node parallel iteration and cross-node feature fusion. Key innovations include a distributed diffusion iteration mechanism tailored to the sensor network topology, an adaptive noise modelling strategy that matches heterogeneous sensor characteristics, and a spatiotemporal consistency constraint module embedded in the reverse diffusion process. Our model reduces motion errors by about 30%, processes each frame within 35 ms, and remains robust under noisy and incomplete data.
  Keywords: sensor network; motion capture; diffusion model; data restoration; spatiotemporal consistency; film production.
  DOI: 10.1504/IJSNET.2026.10077872
   
  
• Spatio-Temporal Feature Extraction in Sensor Networks for Enhancing Wind Forecasting Accuracy  
  by Peiheng Duan, Jianyun Pei, Shaopeng Yang 
  Abstract: Wind energy plays a vital role in renewable energy integration, and the precision of its generation forecasts critically influences the efficiency of grid dispatch operations. While sensor networks provide the fundamental infrastructure for collecting wind energy data, effectively extracting spatiotemporal features from massive heterogeneous sensor data remains a key challenge in improving forecast accuracy. To address this issue, this paper first employs the K-means clustering algorithm to partition the wind turbine monitoring subnetwork. For each cluster, optimized head election mechanisms and intra-cluster node scheduling enable efficient data collection. Subsequently, a graph attention network extracts spatial features from sensor data, while attention mechanisms and causal convolutions capture temporal characteristics. Experimental results demonstrate that this approach reduces total network energy consumption by at least 10.88% and lowers root-mean-square error by at least 0.26, significantly improving prediction accuracy while maintaining high data-collection efficiency.
  Keywords: wireless sensor network; wind power forecasting; data acquisition; graph attention network; attention mechanism.
  DOI: 10.1504/IJSNET.2026.10077873
   
  
• Real-time emotion recognition in visual art using mobile sensing terminals  
  by Ran Wei 
  Abstract: Visual art emotion recognition enables intelligent guiding services on mobile sensing devices in museums. Deploying complex recognition models on resource-limited sensing devices, however, remains challenging due to the trade-off between accuracy and efficiency. This paper proposes a lightweight dual-path neural network for efficient on-device processing. The architecture employs two parallel branches to separately analyse stylistic elements (e.g., colour, texture) and semantic content, effectively capturing artistic emotions. It integrates depthwise separable convolutions and channel attention to reduce computational cost. Evaluations on two public art datasets, ArtEmis and Metropolitan Art, show that a lightweight dual-path neural network achieves accuracies of 68.7% and 65.4%, surpassing the lightweight baseline MobileNetV3 by over 5% while containing only 0.79 million parameters. The model achieves an inference delay of 23.4 milliseconds per image on a mobile platform, demonstrating its suitability for real-time sensing applications. This work facilitates the deployment of sophisticated art analysis capabilities directly on sensor network nodes.
  Keywords: art emotion recognition; mobile sensing terminal; lightweight neural network; dual-path architecture; real-time inference.
  DOI: 10.1504/IJSNET.2026.10078021
   
  
• Trusted on-chain data collaboration for smart healthcare sensor networks enabled by federated learning  
  by Lang Liu, Siling Chen, Siyi Chen 
  Abstract: Smart healthcare sensor networks, where wearable and implantable devices act as distributed edge nodes, confront critical trust barriers in collaborative learning due to their resource constraints and openness, impeding the development of high-performance diagnostic models. To overcome these challenges, this paper proposes a novel framework that deeply integrates blockchain with federated learning to establish trustworthy and efficient cooperation. The framework incorporates an on-chain verifiable aggregation protocol to ensure auditability, a dynamic contribution assessment model based on temporal Shapley values to enable fair incentives, and a lightweight hybrid consensus mechanism to defend against malicious clients. Experiments conducted on a large-scale, real-world electrocardiogram dataset demonstrate that the framework achieves a macro-F1 score of 0.823, detects 98.5% of malicious clients with a 3.2% false-positive rate, and maintains a contribution-utility correlation of 0.94. These results significantly surpass existing state-of-the-art methods and confirm the frameworks effectiveness in real-world medical scenarios.
  Keywords: federated learning; blockchain; healthcare sensor networks; trustworthy collaboration; incentive mechanism.
  DOI: 10.1504/IJSNET.2026.10078122
   
  
• Distributed haptic sensor networks for interactive cyber-physical art systems  
  by Tianyuan Guo, Longguo Tian, Quan Ning 
  Abstract: This research addresses the critical challenge of enabling expressive, real-time tactile interaction in interactive art installations, where conventional approaches are limited by low-resolution sensing and computationally intensive models. This paper proposes an end-to-end system framework that integrates a high-density, flexible distributed haptic sensor network with a novel lightweight dual-stream spatiotemporal convolutional neural network. The system captures high-fidelity touch signals; our dedicated network architecture explicitly models spatial pressure distribution and temporal dynamics separately before adaptive fusion, enabling efficient and accurate gesture recognition. Experimental validation on the ArtTouch-12 dataset shows that our model achieves 95.7% accuracy and a low inference latency of 12.3 ms, significantly outperforming traditional and contemporary deep learning baselines. These results confirm the systems effectiveness in interpreting complex touch intentions and driving responsive multimedia content, thereby enhancing artistic expressivity and user immersion.
  Keywords: haptic sensor; interactive art installation; gesture recognition; spatiotemporal neural network; real-time system.
  DOI: 10.1504/IJSNET.2026.10078123
   
  
• A remaining useful life prediction method combining multi-bearing common features evaluation and ConvLSTM networks  
  by Zhidong Huang, Zhenrui Peng 
  Abstract: Effective sensor data processing is fundamental to predictive maintenance. While deep learning excels at automatic feature learning for remaining useful life (RUL) prediction, end-to-end application on raw data remains challenging. Current methods often extract features via signal processing, then screen them using metrics such as monotonicity and robustness before feeding them into deep networks. However, conventional evaluation typically relies on linear weighted aggregation of indicators, introducing subjectivity and uncertainty due to manual weight-setting. To address this, we propose a data-driven paradigm for assessing sensor data feature quality based on degradation consistency. The approach identifies common degradation features by computing the average Pearson correlation coefficient of each feature across multiple bearings. These features are used to predict RUL using a convolutional long short-term memory (ConvLSTM) network, which captures spatiotemporal dependencies. Experiments on bearing datasets demonstrate our methods superior accuracy and generalisation, providing a robust, data-driven framework for evaluating sensor data features.
  Keywords: bearings; feature evaluation; degradation consistency; remaining useful life; RUL; convolutional long short-term memory; ConvLSTM; predictive maintenance.
  DOI: 10.1504/IJSNET.2026.10078189
   
  
• Stacked Ensemble Deep Learning for Trust Evaluation in Wireless Sensor Network Nodes  
  by Li Chen 
  Abstract: Wireless sensor networks are fundamental to the Internet of Things, yet are highly vulnerable to insider attacks that traditional cryptographic defenses cannot mitigate. Existing trust evaluation models often rely on single-view learning or static fusion mechanisms, which lack adaptability and interpretability. To overcome the limitations of single-model approaches and static fusion mechanisms in wireless sensor network trust evaluation, this paper proposes a stacked ensemble deep learning framework. The framework coordinates three heterogeneous deep-learning base learnersa one-dimensional convolutional neural network, a long short-term memory network with attention, and a graph attention network to extract behavioural, temporal, and topological trust evidence. Evaluated on a public wireless sensor network security dataset, the proposed model achieves a root mean square error of 0.083, a weighted F1-score of 0.941, and a normalised discounted cumulative gain at rank 10 of 0.953, demonstrating statistically significant improvements over recent state-of-the-art baselines.-
  Keywords: wireless sensor networks; trust evaluation; stacked-ensemble learning; deep learning; network security.
  DOI: 10.1504/IJSNET.2026.10078295
   
  
• An Overheat Detection Method for Substation Switchgear based on Microbolometer Infrared Imaging  
  by Xiaoyong Bo, Zhaoyang Qu, Lei Wang, Yunchang Dong, Feng Liang 
  Abstract: While distributed infrared sensor networks are crucial for substation maintenance, their application is hindered by complex backgrounds, blurred thermal anomalies, and limited edge-node computing resources. This paper proposes a lightweight model for infrared overheating detection, specifically designed for deployment on resource-constrained edge nodes. Employing GhostNet as its backbone, the model incorporates multi-scale feature modelling and attention mechanisms to enhance overheating feature discrimination and suppress background interference. Additionally, a decoupled target-detection strategy optimises edge data processing by reducing the coupling between classification and positioning during inference. Experimental results show the proposed method meets real-time online inspection requirements and outperforms existing methods. Achieving an accuracy approximately 10 percentage points higher than the overall average of comparison models, this approach proves highly effective for detecting overheating in substation switchgear.
  Keywords: Microbolometer infrared imaging; Infrared overheating detection; Attention mechanism; Operation and maintenance of substation equipment; Substation switching equipment.
  DOI: 10.1504/IJSNET.2026.10078296
   
  
• An Improved Gray Wolf Optimiser for Wi-Fi Signal Strength-Based Localisation  
  by Hongwei Hu, Haoyuan Chen, Ming Li, Hanhong Shi, Wenyi Xia, Yang Jun 
  Abstract: Indoor wireless network positioning based on Wi-Fi signal strength is susceptible to significant positioning errors due to multi-path reflections and environmental noise within large structures and enclosed spaces. To address these challenges, this study proposes an enhanced grey wolf optimiser. By incorporating a nonlinear convergence factor, dynamic fitness weighting, and a neighbourhood search mechanism, it addresses the original algorithms premature convergence and insufficient local exploration, thereby enhancing convergence efficiency, robustness, and positioning accuracy. The algorithm was validated using the 2022 IEEE Congress on Evolutionary Computation (IEEE CEC 2022) single-objective numerical optimisation benchmark functions (eight functions). Convergence analysis confirmed its excellent global exploration capability, demonstrating outstanding performance in multi-obstacle environments. Positioning experiments demonstrate superior performance over existing methods across maximum, mean, and root-mean-square errors, achieving significantly improved positioning accuracy. The algorithm maintains stable operation even with limited access points, effectively mitigating signal interference caused by obstacles. In summary, this robust algorithm meets high-precision indoor positioning requirements in complex scenarios.
  Keywords: Improved gray wolf optimizer; indoor positioning; received signal strength; Wi-Fi.
  DOI: 10.1504/IJSNET.2026.10078462
   
  
• Sensor Data Fusion and Demand Forecasting Modelling in Smart Retail Inventory Management  
  by Cuihong Zhao, Lining Wang 
  Abstract: Wireless sensor networks solve the problem that traditional inventory management cannot perceive the status of goods in real time by deploying multiple source sensors. However, the generated multi-source heterogeneous data is redundant, and the temporal and spatial correlations with complex requirements are difficult to model. This paper constructs a full-chain solution from data collection and fusion to prediction. An improved clustering algorithm is proposed to optimize network energy consumption and reliability. An adaptive fusion method based on recurrent neural networks is designed to improve data quality. A prediction model with an attention network in the fusion time-series diagram is established to model the dynamic influence between goods explicitly. Experimental results show that the compression rate of sensing data is as low as 10%, the prediction accuracy is 96.08%, and all evaluation indicators have improved significantly, providing an effective approach for intelligent retail inventory management.
  Keywords: inventory management; wireless sensor network; clustering algorithm; sensor data fusion; demand forecasting.
  DOI: 10.1504/IJSNET.2026.10078582
   
  
• Coordinated Coverage Scheduling of Mobile Sensing Nodes for Population Health Monitoring Blind Spots  
  by Xing Yan, Dan Duan 
  Abstract: The accelerating global population aging and rising prevalence of chronic diseases have driven explosive growth in demand for population health monitoring. However, existing monitoring technologies rely on fixed cameras or wearable devices, which suffer from line-of-sight limitations and poor user compliance, leading to frequent monitoring blind spots and unmet needs. To address this, this paper proposes a collaborative coverage scheduling method for mobile sensor nodes. This system integrates multiple node types to collect physiological and environmental parameters. It employs spatiotemporal fusion of multi-source data and an enhanced You Only Look Once version 5s (YOLOv5s) model to identify blind spots accurately. Coverage scheduling is modeled as a multi-objective optimization problem, with a two-stage cooperative evolution algorithm dynamically dispatching nodes to maximize coverage while minimizing energy consumption. Experiments demonstrate that network coverage increases to 94.08%, while node mobility energy consumption decreases by 15%, achieving a balance between coverage effectiveness and energy efficiency.
  Keywords: health monitoring; blind spot recognition; mobile sensor node; coverage scheduling; cooperative evolutionary algorithm.
  DOI: 10.1504/IJSNET.2026.10078831
   
  
• Real-Time Quantification of Group Attention via Multimodal Sensors and Spatio-Temporal GNN  
  by Haoran Zhu, Bing Song 
  Abstract: This paper addresses real-time quantification of group attention in sensor-equipped classrooms, where prior work often focuses on individuals and neglects dynamic peer interactions through simplistic multimodal fusion. This paper proposes a framework that acquires data from distributed visual, acoustic, and physiological sensor nodes and integrates it with spatio?temporal graph neural networks. Head pose and gaze from video, voice activity from audio, and physiological signals are extracted to construct dynamic graphs reflecting student interactions. A graph attention network captures attention propagation among students, and a temporal convolutional network models its temporal evolution. A gated fusion mechanism adaptively combines multimodal features. On the corpus, our method achieves a Pearson correlation coefficient of 0.835, outperforming the strongest baseline by 0.047, with a root mean square error of 0.128 and an F1 score of 89.0 percent. Statistical tests confirm significance, and real?time inference reaches 32 frames per second on edge nodes.
  Keywords: sensor networks; multimodal data fusion; spatio?temporal graph neural networks; distributed sensing; real?time analytics.
  DOI: 10.1504/IJSNET.2026.10078832
   
  
• Three-Dimensional Field Reconstruction for Food Fermentation Processes Using Dynamic Sensing Network  
  by Xingyu Gao, Huiwei Lv 
  Abstract: Industrial fermentation processes are critical to biomanufacturing, where internal three-dimensional physical fields directly determine product yield and efficiency. Yet these fields remain largely unobservable due to sparse sensor deployments, forcing operators to rely on empirical judgment and leading to monitoring blind spots. This paper proposes a dynamic sensing network-implicit neural field framework that reconstructs complete 3D fields from limited measurements. The framework integrates physics-informed implicit neural representations with multi-resolution hash encoding to embed fermentation kinetics as soft constraints, while a Monte Carlo dropout-based dynamic sensing strategy iteratively optimizes sensor placements. Experiments show our method achieves a mean absolute percentage error of 4.2% with twelve sensors, substantially outperforming graph neural networks. The dynamic strategy further reduces reconstruction error by 15% compared to static layouts. This enables accurate 3D field visualization and reduces empirical judgment in industrial biomanufacturing.
  Keywords: dynamic sensing network; implicit neural representation; physics-informed learning; 3D field reconstruction; fermentation process.
  DOI: 10.1504/IJSNET.2026.10078867
   
  
• Robust Navigation for Autonomous Driving Sensor Networks Based on Variational Bayesian Cubature Kalman Filter  
  by Yuzhen Wang 
  Abstract: With autonomous driving technology gradually evolving towards complex open roads, multisource sensors have become the core support for achieving high-precision navigation. However, autonomous driving sensor networks face issues such as heavy-tailed non-Gaussian noise and error accumulation. Traditional filtering algorithms struggle to balance navigation accuracy, robustness, and real-time performance. To address this, this paper proposes a robust navigation method based on a variational Bayesian cubature Kalman filter. Heavy-tailed noise is modelled using the student's t-distribution. Noise parameters are estimated online via variational Bayesian methods. A delay compensation mechanism and seamless fusion strategy with nonlinear AutoRegressive with eXogenous inputs neural networks are designed to suppress abnormal observations and compensate for inertial drift. Experimental results show that the position root-mean-square error of the proposed method is as low as 1.82 m, the unlock divergence rate is only 0.12 m/s, and the single-computation time is merely 1.15 ms, significantly improving navigation accuracy, robustness, and real-time performance.
  Keywords: autonomous driving sensor network; robust navigation; variational Bayesian; cubature kalman filter.
  DOI: 10.1504/IJSNET.2026.10078868
   
  
• SecCal-ALGT: a Security-Calibrated Adaptive Local  
  by Liu Binbin, Wenshan Li, Tao Li, Jiangchuan Chen 
  Abstract: Although Transformer-based models have shown promise in time-series anomaly detection, ex-isting approaches still struggle with non-stationary behavioral shifts, insufficient sensitivity to short-burst anomalies, and high computational overhead in resource-constrained IoT environ-ments. To address these challenges, we pro-pose SecCal-ALGT, a security-calibrated adaptive local-global Transformer framework for IoT sensor anomaly detection. The proposed framework combines adaptive positional encoding for non-stationary anomaly localization, local-global attention fusion to model both short-term burst patterns and long-range dependencies, lightweight residual connections for efficient deployment, and a mixed-calibration thresholding strategy to balance false alarms and missed detections under severe class imbalance. Experiments on the NetFlow-BoT-IoT (NF-BoT-IoT) and UNSW-NB15 benchmark datasets show that SecCal-ALGT achieves superior precision-recall trade-offs com-pared with several representative unsupervised baselines. Overall, the proposed framework pro-vides an effective and deployable solution for security-oriented anomaly detection in IoT sens-ing environments.
  Keywords: IoT security; anomaly detection; Transformer; time-series analysis; sensor data protection.
  DOI: 10.1504/IJSNET.2026.10078870
   
  
• A Network Public Opinion Analysis System based on Natural Language Processing  
  by Guanlin Chen, Zhouyi Xu, Rui Huang, Wenyong Weng, Wujian Yang 
  Abstract: To enhance the intelligence of urban governance, this paper proposes a novel Network Public Opinion Analysis System by conceptualising social media platforms as Virtual Sensor Networks. The System's framework is structured into a classic three-tier sensor network architecture: First, a Perception Layer that employs multi-threaded software sensors for high-concurrency signal acquisition; Second, a Processing Layer that utilizes Natural Language Processing for signal denoising and sentiment feature extraction, converting unstructured observations into quantifiable state estimates; Third, an Application Layer for trend awareness. To address the challenges of data sparsity and noise in public opinion sensing, we adopt an Intelligent Adaptive-Window Rolling Forecast algorithm based on the Autoregressive Integrated Moving Average Model. Experimental results demonstrate that our proposed framework effectively facilitates the end-to-end process of situational awareness for public opinion. Furthermore, it exhibits superior predictive fidelity, offering robust decision support for urban sensory intelligence systems.
  Keywords: Social Sensing; Mobile Crowd Sensing; Virtual Sensor Networks; Adaptive; Natural Language Processing; Public Opinion Analysis.
  DOI: 10.1504/IJSNET.2026.10078871
   
  
• Dynamic Task Scheduling for Self-Powered Sensor Nodes Based on Smart Textile Materials  
  by Huan Chen, Mengchao Chen, Jialin Wu, Jie Gao 
  Abstract: Self-powered sensing nodes integrated into smart textiles face unreliable energy harvesting due to human motion variability, leading to frequent node failures and energy waste. To prevent energy depletion and ensure sustained operation, this paper proposes a prediction-enhanced Lyapunov scheduling framework deployed directly on the smart textile sensor nodes. This framework integrates real-time energy forecasting into online task allocation, enabling autonomous, on-node decision-making without reliance on external servers. An attention-based long short-term memory network predicts future harvested energy from motion data, and the prediction, weighted by confidence, is embedded in a Lyapunov optimization framework with real-time updates to allocate tasks while maintaining energy queue stability dynamically. Experiments on the physical activity monitoring for aging people 2 dataset demonstrate that the proposed method achieves an average utility of 4.27 J per time slot, outperforming state-of-the-art baselines by 9.8%. These results validate its effectiveness in ensuring energy-neutral operation under dynamic energy harvesting conditions.
  Keywords: smart textile; energy harvesting; task scheduling; Lyapunov optimization; attention Long Short-Term Memory (LSTM).
  DOI: 10.1504/IJSNET.2026.10078873