International Journal of Sensor Networks (25 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
   
  
• PULP-Lite: a More Light-weighted Multi-core Framework for IoT Applications  
  by Yong Yang, Yuyu Lian, Yanxiang Zhu, Shun Li, Wenhua Gu, Ming Ling 
  Abstract: The increasing volume of data generated by real-time applications and sensors places significant performance demands on processors. Single-core processors are constrained by the inherent limitations of their architecture in terms of parallel processing capability, making it challenging to handle real-time applications. To address this, we propose parallel ultra low power lite (PULP-Lite), a tightly coupled multi-core on-chip system to efficiently handle near-sensor data analysis in the Internet of Things endpoint devices. PULP-Lite uses low-latency interconnections and an innovative address-mapping mechanism to connect central processing units (CPU), ensuring high-performance processing while maintaining flexibility with a lightweight multi-core programming. We evaluate a field-programmable gate array (FPGA) implementation of PULP-Lite with 8 cores, showing a speedup of 6
  Keywords: multi-core optimization; parallel processing; computer systems organization.
  DOI: 10.1504/IJSNET.2025.10070743
   
  
• ToI-Model: Trustworthy Objects Identification Model for Social-Internet-of-Things (S-IoT)  
  by Rahul Gaikwad, Venkatesh R 
  Abstract: The social-internet-of-things (SIoT) paradigm integrates social concepts into IoT systems. Identifying trustworthy SIoT objects, as well as managing trust, are essential for promoting cooperation among them. The current state-of-the-art methods inadequately quantify the trustworthiness of SIoT objects and fails to evaluate trustworthiness of SIoT objects. This paper comprehensively considers specific features of SIoT objects and integrates them with the theory of social trust. The proposed Trustworthy-objects identification model (ToI-Model) captures comprehensive trust, proficiency, readiness, recommendation, reputation, honesty and excellence metrics for identifying trustworthy objects in SIoT. Service requester (SR) uses trust score of service providers (SP) before initiating service delegation. A series of experiments are conducted to evaluate the proposed trust models effectiveness in the successful completion of services, convergence, accuracy, and resilience against deceitful activities. Results of experiment shows that trust model identifies trustworthy service provider that has 19.89% more trust score and a 27.61% less latency than state-of-the-art models.
  Keywords: Trustworthy; Proficiency; Readiness; Recommendation; Reputation; Honesty; and Excellence.
  DOI: 10.1504/IJSNET.2025.10070868
   
  
• A Decoupling Algorithm for Three-Dimensional Electric Field Sensors Based on Extreme Learning Machines Optimised by Bat Algorithm  
  by Wei Zhao, Zhizhong Li 
  Abstract: During measuring the spatial electric field intensity using a three-dimensional electric field sensor, due to the electric field components coupling effect caused by the electric field distortion, a certain coupling error exists in the electric field intensity components measurement. Aiming at the problem of insufficient decoupling accuracy of the traditional extreme learning machine method, an optimised extreme learning machine method based on the combination of maximum inter-class variance and the bat algorithm is proposed to decouple the three dimensional electric field sensor. The Bat algorithm optimised the extreme learning machine methods optimal initial weight and threshold. The maximum inter-class variance method was used to analyse the inherent coupling characteristics of the sensor. The coupling effect was classified according to the varying coupling contribution degree. The traditional extreme learning machine decoupling network was extended. The calibration experiments and decoupling calculations show that the extreme learning machine algorithm optimised by the bat algorithm and maximum inter-class variance can effectively reduce the error, which is between the electric field components obtained by the model calculation and the actual electric field components, and can effectively reduce the interference generated by the inter-dimensional coupling effect of the sensor, and further improve the measurement accuracy of the electric field intensity.
  Keywords: bat algorithm; decoupling; extreme learning machine; maximum inter-class variance; three-dimensional electric field sensor.
  DOI: 10.1504/IJSNET.2025.10072003
   
  
• A New Hyperchaotic Image Encryption Scheme Based on DNA Computing and SHA-512  
  by Shuliang Sun, Xiping Wang, Zihua Zhao 
  Abstract: Smartphones and digital cameras are becoming more and more widespread in the world. Massive images are generated every day in the world. They are easily transmitted on the insecure channel-Internet. Encryption technique is usually adopted to protect sensitive images during communication. A new cryptosystem is constructed by six-dimensional (6D) chaotic system and deoxyribonucleic acid (DNA) techniques. Firstly, the hash value is calculated. It keeps the encrypted result closely connected with the original image. The initial conditions of the cryptosystem are produced with the generated hash value and secret key. Secondly, the pixel is divided into four parts, forming a large matrix. Scrambling is performed on the new image. Subsequently, DNA coding, modern DNA complementary rules, DNA computing, and DNA decoding are utilised. Diffusion operation is also executed to improve the security, and the ciphered image is achieved finally. The experimental performance reveals that the designed algorithm has some advantages. It also signifies that the designed algorithm could protect against common attacks and is more secure than some existing methods.
  Keywords: hyperchaotic system; DNA computing; SHA-512.
  DOI: 10.1504/IJSNET.2025.10072159
   
  
• 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.
  
  
• Security Analysis and Improvement of Key Exchange Protocol in LoRaWAN Network  
  by Arman Amjadian, Hamid Meghdadi, Ali Shahzadi 
  Abstract: While using very low-power and inexpensive transmitters, LoRaWAN networks exhibit very high sensitivity and excellent reliability over very long ranges. Although these networks benefit from higher security performance compared to other low-power wide-area communication protocols, some aspects of their security can be greatly improved. Namely, the key exchange protocol was considered as one of the weakest links in the security of LoRaWAN networks. This issue was addressed at the second edition of the LoRaWAN protocol. However, the improvement was achieved at the cost of using much more complicated algorithms. Even then, some of the security issues of the protocol such as vulnerability against node capture attack and forward secrecy remained the same. In this paper, we demonstrate the limitations of new LoRaWAN key exchange protocols using Scyther and ProVerif security analysis tools. Then we propose a novel scheme that while requiring much less complex computations, offers a more robust security for LoRaWAN networks. We use the aforementioned tools to verify that the proposed method considerably improves the resilience of LoRaWAN against known attacks
  Keywords: IoT; LoRaWAN; Network security; OTAA; Scyther; ProVerif.
  DOI: 10.1504/IJSNET.2025.10072911
   
  
• Nonlinear Least Squares-Based Localisation Method for WSN-based Smart Agriculture Systems using Range Measurements  
  by Emad Hassan 
  Abstract: Accurate source localisation in Wireless Sensor Networks (WSN) is critical for applications requiring precise target tracking, environmental monitoring, and security surveillance. Traditional localisation techniques suffer from multipath interference, leading to degraded accuracy. This paper proposes an enhanced localisation algorithm leveraging multipath exploitation to improve position estimation. The proposed approach utilises time difference of arrival (TDOA) and direction-of-arrival (DOA) measurements, incorporating a hybrid scheme that can mitigate noise and enhance accuracy. A space division multiple access (SDMA) spread spectrum receiver is employed to extract DOA estimates, while TDOA information is utilised to differentiate between line-of-sight (LOS) and non-line-of-sight (NLOS) components. By associating multipath signals with corresponding reflectors, the scheme significantly improves localisation performance, even in environments where LOS paths are obstructed. Simulation results demonstrate that the proposed scheme significantly improves localisation accuracy compared to conventional schemes. The root mean square error (RMSE) is reduced by 30%, and the overall localisation success rate is increased by 25%, showcasing the robustness of the proposed scheme. These findings suggest that integrating multi-path components constructively rather than treating them as interference can enhance WSN localisation performance, making it suitable for real-world deployment.
  Keywords: WSNs; source localization; DOA; NLOS; TDOA; smart irrigation systems.
  DOI: 10.1504/IJSNET.2025.10072995
   
  
• A Blockchain-Based Privacy Protection Model for a Spatial Crowdsourcing Platform  
  by Amal Albilali, Maysoon Abulkhair, Manal Bayousef 
  Abstract: Spatial crowdsourcing (SC) involves collecting geographic information from a crowd of people using mobile devices, raising critical privacy issues regarding participants' location data. In this article, we propose an efficient privacy protection task assignment model (ePPTA) as a novel method that combines centralised and decentralised platforms to achieve privacy protection for worker location, worker identity, and task location during the task assignment (TA) process. Through a centralized SC platform, we achieve privacy protection using an elliptic curve cryptography (ECC), ensuring low user computational and communication overheads. The task assignment process and its data integrity are managed via blockchain technology. We evaluate our model on a real dataset, comparing it with state-of-the-art methods. The ePPTA model demonstrates low user computational and communication overheads and theoretically prevents task-tracking and eavesdropping attacks from external entities. Performance evaluation results confirm that the proposed model's efficiency is reasonable, providing robust privacy protection for SC.
  Keywords: Crowdsourcing; Privacy; Location Privacy; Spatial Crowdsourcing (SC); Blockchain.
  
  
• Long-term Wind Power Prediction based on Feature Fusion Model and Temporal Pattern Attention Mechanism  
  by Li Liu, Ze Wang, Siwen Lei, Shengchi Liu, Hao Wang, Yue Jiang 
  Abstract: With the increasing global demand for clean energy, wind power has rapidly expanded as a renewable resource. However, the multidimensionality, long time series, and high volatility of wind power data pose significant challenges for long-term forecasting. This paper proposes a long-term wind power prediction model that utilises a feature fusion method and an attention mechanism. It integrates the strengths of the Light Gradient Boosting Machine (LightGBM) and Long Short-Term Memory (LSTM) algorithms, employing the temporal attention mechanism for fusion. The LightGBM algorithm handles multidimensional data and selects critical spatial features from wind farm data, while the LSTM network captures long-term dependencies in time-series data. The attention mechanism dynamically assigns weights to predictions based on specific conditions, allowing the model to focus on more relevant features during different periods and fluctuation regions. Experiments on data from multiple regions demonstrate that the proposed model outperforms existing methods, especially in long-term predictions.
  Keywords: wind power; long-time series; spatial multi-features; temporal attention mechanism; feature fusion model.
  DOI: 10.1504/IJSNET.2025.10073135
   
  
• OCH-MAC: an Optimised Channel Hop MAC Protocol with Dynamic Blacklist for Wireless Sensor Networks  
  by Vandenberg B. Da Paixao, Renato De Moraes 
  Abstract: This paper presents two sets of algorithms for optimizing data exchange from a sensor network's medium access control (MAC) perspective. The Optimized Channel Hopping MAC (OCH-MAC) seeks that, in case of the inoperability of the current communication channel in use, the next chosen channel presents the best conditions of the spectrum because the communicating sensor pair employs a quality scale for each available channel. Accordingly, we propose a dynamic blacklist (D-Blacklist) algorithm, where three factors determine the blocking or unblocking of the channel: the analysis of the previous and current signal-to-noise plus interference ratio (SNIR) records, the SNIR levels of the active and neighboring channels, and the seasonality in the SNIR variations, analyzing the current record and comparing it with the recorded minimum and maximum. Results show that these algorithms enhance the reliability of blocking or unblocking the channel and outperform other sensor technologies, such as WirelessHART, ISA100.11a, and IEEE802.15.4e/A-TSCH.
  Keywords: Channel blacklist; MAC protocols; sensor networks.
  DOI: 10.1504/IJSNET.2025.10073277
   
  
• A Behaviour Detection Algorithm Integrating Lightweight Networks and Feature Recombination  
  by Gen Liang, Yu Zhang, Guoxi Sun, Xinchao Li 
  Abstract: Traditional behaviour detection methods often have problems such as low accuracy and slow processing speed, making it difficult to meet the practical application needs of industrial production scenarios. This study proposes a behaviour detection algorithm that integrates lightweight networks and feature recombination. First, we replace you only look once (YOLO) backbone with an enhanced MobileNetV3, reducing model complexity and accelerating inference. Second, we introduce content-aware reassembly of features, replacing conventional upsampling to improve precision. Further, switchable atrous convolution in the neck network enhances adaptability to multi-scale features, while vision transformer with deformable attention strengthens spatial modelling. Ablation experiments demonstrate the algorithms effectiveness, achieving a 75.2% mAP, with gains of 2.8% and 4.8% in precision and recall, respectively. Compared to existing technologies, this method offers the advantages of fast speed and high accuracy, making it suitable for real-time detection scenarios, such as those in the petrochemical industry.
  Keywords: behaviour detection; lightweight network; feature reorganisation; dilated convolution; deformable attention.
  DOI: 10.1504/IJSNET.2025.10073426
   
  
• Extraction and Recognition of Emotional Features in Artistic Images using Visual Sensor Network  
  by Fei Liu 
  Abstract: Artistic images serve as an important medium for emotional expression, and the precise identification of their emotional semantics is of great significance for art appreciation. To address the current issues of poor image quality and incomplete extraction of emotional features in artistic images, this paper first designs a multi-node wireless visual sensor network architecture to enable rapid acquisition of artistic images. Then, an improved Retina-Cortex algorithm is designed to enhance artistic images. Based on this, a feature extractor is constructed using a Residual Neural Network to extract multidimensional emotional features, calculate the emotional weights of the three features, and perform feature fusion. Finally, the emotional features are used to identify the emotions in artistic images. The experimental results show that the average recognition accuracy and F1 score of the proposed model are 94.86% and 96.47%, enabling effective recognition of emotions in art images.
  Keywords: visual sensor network; artistic images; feature extraction; emotion recognition; Retina-Cortex (Retinex) algorithm.
  DOI: 10.1504/IJSNET.2025.10073591
   
  
• Sparse Signal Recovery via Reweighted Shrinkage Thresholding: Applications in Compressed Sensing MRI  
  by Yuze Liu, Xiaokun Zhou, Mingjun Feng, Da Cao, Wei Wang 
  Abstract: Compressed sensing (CS) is a powerful technique for rapid magnetic resonance imaging (MRI). The iterative shrinkage thresholding algorithm (ISTA) is widely used due to its computational efficiency, but conventional versions lack adaptive adjustment of regularisation parameters, which limits both accuracy and speed. We propose a novel algorithm that dynamically updates regularisation weights according to the convergence status of previous iterations. To mitigate staircase artefacts commonly produced by wavelet-based reconstructions, the method incorporates the contourlet transform, which more effectively captures edges and contours. The approach is designed for sensor-based MRI systems, where efficient data acquisition and processing are critical for real-time imaging. Experiments on MR images from multiple anatomical regions and sampling rates demonstrate that the method achieves faster convergence and superior reconstruction quality compared to traditional ISTA, highlighting its potential in real-time medical imaging applications, including wireless health monitoring and sensor-network-based MRI systems.
  Keywords: Compressed Sensing; Iterative Shrinkage Thresholding Algorithm; Regularization Weight Adjustment; Contourlet Transform.
  DOI: 10.1504/IJSNET.2025.10073600
   
  
• A Dynamic Interactive System for Intelligent Lighting Art Installations based on Multimodal Sensing  
  by Tingshuo Yan, Yan Chen, Shenjian Hu, Wei Deng 
  Abstract: To address the problem of incomplete effective feature information extraction in the interaction methods of intelligent lighting art installations, this paper first collects data such as light intensity and user behaviour through multimodal sensors. Then, multimodal sensor signals are mapped to a two-dimensional image space, and a lighting perception network is used to evaluate lighting conditions and obtain lighting probabilities. By comprehensively representing shallow and deep features for information measurement and assigning adaptive weights, the beneficial information of the fusion method is maximally preserved. Finally, positive and negative feedback are set based on user behaviour to train the reinforcement learning model, enabling it to adapt to the user's dynamic interaction habits gradually. Simulation results show that the control accuracy of the proposed method is 95.39%, the system interaction response time is 13.6ms, and it can effectively improve the control accuracy and real-time performance of the interaction system.
  Keywords: intelligent lighting art installation; dynamic interaction; multimodal sensing fusion; lighting perception; reinforcement learning.
  DOI: 10.1504/IJSNET.2025.10073758
   
  
• Multi-Agent Deep Reinforcement Learning Edge Task Scheduling Algorithm with Migratable Service Environment  
  by Zengwei Lyu, Yu Zhang, Zhenchun Wei, Juan Xu, Lei Shi, Yuqi Fan 
  Abstract: To address the problem that computing task offloading cannot dynamically schedule the service environment according to the task preferences of different regions, we implement real-time scheduling of service environments and computing tasks based on container technology. This approach can dynamically meet the service demands of users in different regions. In order to solve the coupling problem of the scheduling process and the replacement problem of storage content, we integrate task offloading, service environment migration, and storage content replacement into a task scheduling action. Additionally, we design an attention-based decentralised-actor centralised-critic network to address this problem. The proposed algorithm uses a centralised training and distributed execution framework and optimises the policy learning process by introducing an attention mechanism. Simulation experiments demonstrate that the proposed algorithm can effectively improve the load balancing level of the multi-edge collaborative system and achieve a lower average task execution delay.
  Keywords: Multi-Agent Reinforcement Learning; Edge Computing; Task Scheduling; Edge Computing; Migratable Service Environment.
  DOI: 10.1504/IJSNET.2025.10073784
   
  
• A Conflict-Balanced SF Allocation for a LoRa Network  
  by Junming Guan, Yancheng Yao, Chuanxin Zhao, Changzhi Wu 
  Abstract: Long Range (LoRa) has emerged as one of the preferred communication solutions for connecting Internet of Things (IoT) devices due to its low power consumption and long-distance transmission capabilities. The LoRa network supports communication between multiple nodes and gateways over several kilometres. To reduce energy consumption and enhance the communication quality, the configuration of LoRa network is important. In this paper, the signal communication in a LoRa network is first analysed, and a nonlinear integer programming problem is developed to configure the spreading factor (SF) of nodes. The problem is formulated by maximizing the volume of successfully transmitted data and reducing the probability of data collisions to minimise system energy consumption. Additionally, the performance of the proposed scheme is evaluated through extensive simulations. The results demonstrate that it achieves a higher number of delivered packets with lower energy consumption, outperforming both the equal-interval-based (EIB) and equal-area-based (EAB) SF allocation mechanisms. Finally, the network scalability is analysed by considering factors such as network configuration time, node density, and network radius.
  Keywords: Lora networks?spreading factor?allocation scheme?EAB?EIB.
  DOI: 10.1504/IJSNET.2025.10073865
   
  
• An Edge Computing Offloading Strategy based on Multi-Dimensional Attributes and Distributed Deep Learning  
  by Dong She 
  Abstract: With the explosive growth of smart terminal devices and the wide adoption of latency-sensitive applications, the traditional cloud computing model is difficult to meet the demands of ultra-low latency and high privacy protection. Therefore, this paper proposes a distributed deep reinforcement learning offloading strategy based on multi-dimensional joint modeling of task, device, and environment attributes. A state space integrating multi-dimensional attributes is constructed to achieve comprehensive awareness of the system state; a distributed asynchronous deep Q-network framework is designed to realize knowledge sharing through local model co-training among multiple edge nodes. Experimental results show that this approach can reduce the average task processing latency by 4.8% and the overall system energy consumption by 5.3%. This research provides a practical solution for computational offloading in resource-constrained edge scenarios that balances efficiency and energy consumption.
  Keywords: multi-dimensional attributes; distributed; deep learning; edge computing offloading.
  DOI: 10.1504/IJSNET.2025.10073899
   
  
• A Dynamic Sensing-Based, Adaptive Energy-Efficient Transmission Scheme for Wireless Sensor Networks in Perishable Goods Monitoring  
  by Li Chen 
  Abstract: The cold chain is a temperature-controlled supply chain crucial for preserving perishable goods such as pharmaceuticals by maintaining low temperatures throughout storage and transport. Wireless sensor networks play a crucial role in monitoring these conditions in real-time. However, their performance is hindered by harsh environments, limited node energy, and dynamic channel conditions, which result in a shortened network lifetime and unreliable transmissions. Existing transmission schemes often employ static parameters and lack adaptability, leading to energy waste and poor reliability under fluctuating cold chain conditions. To overcome these limitations, this paper proposes an adaptive energy-efficient transmission scheme based on dynamic sensing. The core approach involves constructing a sensing model to quantify key parameters and designing a fuzzy logic-based adaptive transmission algorithm. Experimental results demonstrate that the proposed method reduces energy consumption by up to 32%, extends network lifetime by 40%, and maintains a data transmission success rate of over 98%, outperforming conventional methods.
  Keywords: wireless sensor networks; dynamic sensing; adaptive transmission; energy efficiency.
  DOI: 10.1504/IJSNET.2025.10073900
   
  
• Stochastic Network Calculus based Energy Harvesting Model for Underwater Agriculture  
  by Vignesh S. R, Rajeev Sukumaran 
  Abstract: An Underwater Wireless Sensor Network (UWSN) is a specialised Wireless Sensor Network for marine monitoring, where energy sustainability is critical since node batteries are difficult to replace. Energy harvesting (EH) from the underwater environment provides a practical solution. This research presents a novel EH framework for underwater agriculture monitoring by applying Stochastic Network Calculus (SNC) to model and analyse the impact of delay and storage constraints on Energy Harvesting Rate (EHR). The objective is to ensure reliable energy availability for continuous monitoring of parameters such as pH, salinity, temperature, dissolved oxygen, and water quality. The main contribution lies in integrating SNC with piezoelectric-based EH to derive probabilistic delay and energy performance bounds, an approach rarely explored in underwater agriculture. Simulation results using discrete-event simulators validate the analytical model, showing that larger packet sizes increase the minimum EHR, while stricter delay requirements lower the minimum EHR under a fixed traffic rate.
  Keywords: UWSN; Energy Harvesting; SNC; Energy Storage Constrain; Sensor Node.
  DOI: 10.1504/IJSNET.2025.10073911
   
  
• A novel offloading algorithm for cost-sensitive tasks in VEC networks using deep reinforcement learning  
  by Benhong Zhang, Hao Xu, Xiang Bi, Qiwei Hu 
  Abstract: Vehicular edge computing (VEC) provides a fundamental condition for the fast and complete realisation of complex intelligent functions in autonomous driving vehicles. However, different tasks have different requirements on time cost, for example, the tasks related to safe driving have strict requirements on real-time, which brings challenges to task offloading and resource allocation in VEC. This paper first defines different utility evaluation functions that measure the delay requirements of different tasks. Then, an optimisation problem is presented by considering the task types, the dynamic generation feature of tasks and the price cost that measures the willingness of the service providers. Finally, the task offloading and resource allocation process is modelled as a Markov decision process (MDP) and a D3QN-based algorithm is designed to solve our problem. Simulation results show that the proposed algorithm has better performance on utility and task success rate compared to other algorithms.
  Keywords: vehicle edge computing; VEC; task offloading; resource allocation; Markov decision process; MDP; duelling double deep Q network; D3QN.
  DOI: 10.1504/IJSNET.2025.10072414
   
  
• A path optimisation method based on dynamic clustering strategy and non-dominated sorting genetic algorithm II in wireless sensor networks  
  by Liying Zhao, Jin Zhu, Chao Liu, Yu Wang, Sinan Shi, Chao Lu, Qi Luan 
  Abstract: Traditional wireless sensor network data transmission path selection often focuses on single- objective optimisation, resulting in unbalanced node load and insufficient path reliability. This study proposes a routing optimisation method that combines dynamic clustering strategy with non- dominated sorting genetic algorithm II. MiniBatchKMeans is used to divide network nodes into clusters of different scales, and cluster heads are elected by comprehensively evaluating node residual energy, distance to cluster centre and historical load. The algorithm generates Pareto optimal paths with objectives of minimising energy consumption and transmission delay and maximising signal strength. Simulation results show that the proposed method extends network lifetime by 12.2%, increases total data throughput by 21.1%, and improves load-balancing performance.
  Keywords: clustering strategy; non-dominated sorting genetic algorithm II; NSGA-II; path optimisation; wireless sensor network; WSN.
  DOI: 10.1504/IJSNET.2025.10072468
   
  
• Spacetime graph convolution driven sensing edge node collaboration for photovoltaic fault diagnosis  
  by Dingyou Wang 
  Abstract: Focusing on issues of insufficient feature characterisation and low accuracy of fault diagnosis in current photovoltaic fault diagnosis methods, this paper firstly designs an improved sensing edge node arrangement strategy to collect photovoltaic fault data comprehensively. Then, the similarity graph modelling method is used to construct the sensing edge node graph, and the one-dimensional convolutional network and residual structure are used to build the characteristic pre-extraction function to form the spatio-temporal graph data. The graph convolutional network and the characteristic pre-extraction function are connected to form a feature extractor to capture the time-spatial correlation features at different scales in the spatio-temporal map data. Finally, a multi-class classifier is designed to classify fault features to realise the diagnosis of photovoltaic faults. Experimental outcome indicates that the diagnosis accuracy of the designed model is improved by at least 4%, which attests to the model's superior performance.
  Keywords: photovoltaic fault diagnosis; sensing edge node; residual structure; convolutional neural network; spatiotemporal graph convolutional neural network.
  DOI: 10.1504/IJSNET.2025.10073085
   
  
• MultiNeigh-GNN: a multi-order neighbouring data fusion graph neural network for IoT intrusion detection  
  by Zhihao Yin 
  Abstract: Contemporary internet of things (IoT) intrusion detection techniques encounter difficulties in intricate networks: typically, they do not account for cross-hop dependencies outside immediate neighbours, and label sparsity obstructs the learning process from malicious traffic, hence hindering the detection of fresh attacks. We present a multi-neigh graph neural network (GNN) technique that integrates multi-order neighbouring data to overcome these challenges. This approach effectively captures multi-hop dependences and cross-hop attack propagation paths by integrating 1st, 2nd, and 3rd order neighbouring data via a multi-order neighbouring data fusion module, thereby substantially enhancing the model's capacity to recognise intricate attack patterns. The developed cross-neighbouring graph mutual-exclusion learning module efficiently identifies distinctive characteristics from sparse harmful traffic samples. Experimental findings on several publicly accessible IoT intrusion detection datasets demonstrate that MultiNeigh-GNN substantially surpasses current benchmark approaches, particularly in addressing cases characterised by intricate attack patterns and sparse harmful traffic.
  Keywords: cross-neighbouring graph mutual-exclusion learning; graph neural networks; GNNs; IoT intrusion detection; multi-hop dependences; multi-order neighbouring.
  DOI: 10.1504/IJSNET.2025.10073463
   
  
• Advanced air quality prediction modelling using intelligent optimisation algorithm in urban regions  
  by Wendi Tan, Zhisheng Li 
  Abstract: Ambient air contamination is a significant environmental challenge threatening human well-being and quality of life, especially in urban areas. Technical breakthroughs in artificial intelligence models offer more accurate air quality predictions by analysing significant data sources, including meteorological factors like humidity, wind speed, and pollution data. Also, existing methods of air quality prediction often lack due to their dependency on statistical models that may not adequately capture the complexities of environmental data like pollutants and meteorological factors. Thus, the research introduces a grey wolf optimised variational autoencoder to enhance the air quality prediction by effectively capturing complex relationships in environmental data. The model acquires the probabilistic nature of variational latent representations from historical air quality input data and prevents overfitting. The relevant features are selected using the grey wolf technique, identifying the appropriate variables to enhance the data quality. Additionally, it optimises critical hyperparameters like learning rates and greedy layer sizes, leading to better convergence during model training and improved performance in air quality index prediction. Experimental results demonstrate improved prediction accuracy, reduced error rate, and faster convergence.
  Keywords: air quality index; AQI; prediction model; optimisation; artificial intelligence; urban region; environmental factors.
  DOI: 10.1504/IJSNET.2025.10070252