International Journal of Computer Applications in Technology (14 papers in press)

Regular Issues

• Open Access

  Supply chain demand forecasting and low-altitude economic risks based on machine learning and artificial intelligence ( Free Full-text Access )

  by Jilu Liu, Yubin Ying, Shaohui Shu, Zechen Zhang 
  Abstract: To address the challenges of coupling high-frequency supply chain forecasting with low-altitude economic risk, a unified framework integrating machine learning and artificial intelligence is proposed. A Long Short-Term Memory (LSTM)-Transformer hybrid model is designed to capture multi-granularity temporal dependencies and mitigate the bullwhip effect. A Graph Attention Network (GAT)-driven risk coupling network is constructed, integrating AirSim simulations with real warehouse data for dynamic quantification. Results show that the forecast RMSE is reduced by 35.3% compared to the Autoregressive Integrated Moving Average (ARIMA) model, and the multi-granularity MAPE variance is only 0.123. Through a joint decision-making mechanism, this method provides an intelligent decision-making paradigm for multi-source heterogeneous scenarios.
  Keywords: supply chain demand forecasting; low-altitude economy; risk analysis; machine learning; artificial intelligence.
  DOI: 10.1504/IJCAT.2026.10077901
   
  

• Open Access

  Ecological management algorithm of fresh supply chain information under artificial intelligence and blockchain architecture ( Free Full-text Access )

  by Chanjuan Li 
  Abstract: This paper establishes a new supply chain management system based on artificial intelligence and blockchain technology (BT), and proposes a new supply chain solution on this basis. This paper defines the traditional centralised management mode as Supply Chain 1 and the blockchain distributed architecture as Supply Chain 2. Sample 1 consists of 500 sets of structured transaction data, while Sample 2 consists of 500 sets of unstructured sensor log data. The results showed that in sample 1, the information transmission times of supply chain 1 and supply chain 2 were 15.4 s and 3.3 s. In sample 2, when the sample size was 500, the information transmission times of SC 1 and SC 2 were 20.7 s and 4.2 s. It can be seen that SC 2 has a higher information transmission efficiency. This evaluation algorithm can be used in practice to diagnose supply chain information transmission bottlenecks and optimise resource allocation.
  Keywords: artificial intelligence; blockchain architecture; fresh supply chain; information ecological management.
  DOI: 10.1504/IJCAT.2026.10077983
   
  

• Open Access

  Simulation model of marine biological population based on multi-agent comfort drive ( Free Full-text Access )

  by Xi Deng, Wei Liang, Yupeng Zhu 
  Abstract: Current simulation models that focus on marine biological communities have not yet taken into account the effects of comfort on the behaviour of single organisms. This shortcoming makes it difficult for researchers to develop accurate simulations of how groups behave when faced with adaptive behaviour in complicated environments. A Multi-Agent System based on Reinforcement Learning (RL) is proposed in this paper, which uses Environmental Perception Mechanisms and Comfort Functions to optimise decisions made by individual agents in complex marine environments via Q-Learning and Deep Q-Networks (DQN). The proposed approach employs a collaborative strategy that combines local and global comfort to simulate collective actions. Experimental results indicate that a comfort-driven decision-making model would enable a collective of individuals to converge within 240 seconds, with 85% of the individuals congregating in areas they perceived as having the highest comfort levels. This work opens up new avenues for simulating the behaviours of marine biological communities and demonstrates the potential of a comfort-driven approach to studying the adaptive behaviour of groups.
  Keywords: marine species group simulation; comfort-driven mechanism; multi-agent system; reinforcement learning; deep Q-network.
  DOI: 10.1504/IJCAT.2026.10078256
   
  

• Open Access

  CISAA: a collective intelligence-based optimisation framework for low-altitude airspace resource allocation and engineering management ( Free Full-text Access )

  by Feng Li 
  Abstract: The rapid growth of the low-altitude economy and complex airspace operations pose challenges for efficient and adaptive resource allocation. Traditional methods struggle with multi-dimensional coupling among resources, missions, and management under uncertainty. This paper proposes a Collective Intelligence-based Spatio-Allocation and Administration (CISAA) framework that integrates swarm intelligence with hierarchical dynamic optimization to enable collaborative decision-making among UAVs, infrastructure units, and management systems. Through multi-layer coordination, CISAA optimizes task scheduling, airspace utilization, and engineering progress. Experiments in simulated low-altitude scenarios show that CISAA achieves higher resource efficiency, task completion, and system stability compared with heuristic and centralized approaches, offering a novel path toward intelligent and sustainable low-altitude airspace management.
  Keywords: low-altitude airspace management; collective intelligence optimisation; multi-agent coordination; engineering resource allocation; adaptive scheduling.
  DOI: 10.1504/IJCAT.2026.10078501
   
  

• Open Access

  Application of information optics experimental methods in optical sensors for wearable devices ( Free Full-text Access )

  by Yifang Li, Yingji He 
  Abstract: Traditional optical sensors struggle with high-precision physiological monitoring in dynamic environments and lack sufficient processing power for real-time wearable devices. This paper presents an information optics method using a microlens array-based optical sensor with advanced optical modulation to enhance resolution and dynamic response. A combined FM and AM signal transmission improves anti-interference and data speed. A filtering-based data processing method and real-time feedback mechanism ensure signal accuracy and timeliness. Experimental results show that under low, medium, and high noise, the transmission rate increases by 350 bps, 404 bps, and 281 bps, respectively, while bit error rates drop to 0.01, 0.03, and 0.07. In dynamic measurements of heart rate, body temperature, blood oxygen saturation, and blood pressure, accuracy exceeds 92%, and average response time is 0.20 seconds. This demonstrates the effectiveness of the information optics method in wearable optical sensors.
  Keywords: information optics experiment; wearable devices; optical sensor; microlens array; filtering algorithm.
  DOI: 10.1504/IJCAT.2026.10079101
   
  

• Open Access

  From physical isolation to disaster recovery communication guarantee based on software-defined perimeter ( Free Full-text Access )

  by Chengfei Qi, Yachao Wang, Shan Li, Yuning Zhang, Wenwen Li, Xinyue Zhang 
  Abstract: This paper proposes a layered fusion architecture to address the contradiction between the security of physical isolation and the flexibility of Software Defined Perimeter (SDP) in disaster recovery. The core layer uses Quantum Noise Channel Encryption (QNCE) and one-way optical gate hardware to ensure absolute data security and solve recovery point hysteresis. The control layer enhances dynamic defense with SDP, blocking 98.2% of lateral attacks and using a 50ms fuse mechanism. Resource scheduling via weighted fair queue improves utilization to 78.4%, reducing TCO by 32.7%. The architecture blocks 100% of APT attacks, with 48.2ms fault switching delay. It achieves 100% Tier-0 RTO and 99.2% average RTO. Under 150% overload, it has a 99.0% switching success rate and supports up to 160% load.
  Keywords: disaster recovery communication; communication security; security protection; software defined perimeter; physical isolation.
  DOI: 10.1504/IJCAT.2026.10079161
   
  

• Open Access

  Sports biomechanical performance analysis algorithm and geometric parameter equation based on reinforcement learning. ( Free Full-text Access )

  by Lian Xiao, Zhe Huang, Zhenzhen Wang 
  Abstract: The existing biomechanical performance analysis models face the problem of balancing high-precision modelling and real-time performance. Based on this, the paper proposes a biomechanical performance analysis framework based on geometric parameter equations and deep reinforcement learning (DRL) collaborative optimisation. First, this paper constructs a rigid body dynamic environment model through explicit geometric parameter equations, and then designs a reward function that comprehensively considers multi-objective requirements such as motion accuracy, energy consumption, and stability, and second, introduces geometric parameter adaptive mechanisms and combinatorial transfer learning techniques to improve the models ability to adapt to individual geometric features, ensuring its universality and personalised performance among different individuals. The experimental results show that compared with the baseline model, this algorithm reduces the motion accuracy error by about 50% in running and jumping tasks, shortens the single frame data processing time by 37%.
  Keywords: biomechanical performance; geometric parameter equation; deep reinforcement learning; deep neural network; reward function.
  DOI: 10.1504/IJCAT.2026.10079193
   
  

• Open Access

  Automatic recognition system of athletic injury images based on deep learning ( Free Full-text Access )

  by Long Wang, Shuping Xu 
  Abstract: Traditional athletic injury image recognition relies on manual feature extraction, which struggles with subtle lesions like meniscus tears in complex backgrounds, resulting in low precision. This paper proposes a Dense Net-GAP (densely connected convolutional networks-global average pooling) model to improve recognition precision and efficiency. First, nonparametric intensity normalization enhances image quality by matching histogram distributions. DenseNet's dense connection mechanism strengthens feature transfer and fusion, effectively capturing subtle lesion information. Replacing fully connected layers with GAP significantly reduces model parameters and computational burden. Softmax classifies various injury types, while RMSProp (Root Mean Square Propagation) optimization ensures rapid convergence. Experiments on seven injury types achieve 97.28% precision, 96.85% recall, 97.06% F1-score, and 0.97 AUC (area under the curve), with only 0.79% false positive rate (FPR), 1.6% false negative rate (FNR), and 30.79ms recognition time, effectively solving the low precision problem of traditional methods in complex backgrounds.
  Keywords: deep learning; densely connected convolutional networks; image recognition for sports injuries; medical image classification; small-sample learning.
  DOI: 10.1504/IJCAT.2026.10080055
   
  
• FPGA implementation and Multisim simulation of a new four-dimensional two-scroll hyperchaotic system with coexisting attractors  
  by Sundarapandian Vaidyanathan, Esteban Tlelo-Cuautle, Khaled Benkouider, Aceng Sambas, Ciro Fabian Bermudez-Marquez, Samy Abdelwahab Safaan 
  Abstract: Field-programmable gate array (FPGA) design of a new four-dimensional two-scroll hyperchaotic system is investigated in this work. A detailed system modelling of the new system with a hyperchaotic attractor begins this work with phase plots, which is followed by a bifurcation study of the new system. Special dynamic properties such as multistability and symmetry are also investigated for the new system. Using Multisim software, a circuit model is designed and simulated for the new hyperchaotic system. FPGA design and Multisim simulation of the new system enable practical applications in science and engineering. The implementation of the FPGA design in this work is carried out by applying two numerical schemes, viz. Forward Euler and Trapezoidal methods. Experimental attractors observed in the oscilloscope show good match with the Matlab signal plots.The FPGA hardware resources are detailed for both numerical methods.
  Keywords: hyperchaos; bifurcation; symmetry; phase plots; hyperchaotic system;rnparameters; stability; multistability; circuit model; FPGA implementation.
  
  
• Improving hybrid-layer convolutional neural network system for lung cancer nodule classification using enhanced weight optimisation algorithm  
  by Vikul Pawar, P. Premchand 
  Abstract: In recent times, lung cancer is evolving as a highly life-threatening disease for human beings. According to the WHO, lung cancer disease is the second largest cause of deaths as compared to all other types of cancer. The prevailing available technology is striving to get more exposure in the field of medical science using Computer Assisted Diagnosis (CAD), where image processing is playing a crucial role for detecting the cancerous nodules in computer tomographic images. Augmenting the machine learning techniques with image processing algorithms is becoming a more comprehensive examination of cancer disease in proposed CAD systems. This paper is describes a heuristic approach for lung cancer nodule detection, and the proposed model predominantly consists of the following tasks, which are image enhancement, segmenting ROI (Region of Interest), features extraction, and nodule classification. In pre-processing, primarily the Adaptive Median Filter (AMF) filtering method is applied to eliminate the speckle noise from input CT images of Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI): in the LIDC-IDRI dataset, the quality of input image is improved by applying Histogram Equalization (HE) technique with Contrast-Limited Adaptive (CLA) approach. Secondly, in the successive stage the Improved Level-Set (ILS) algorithm is used to segment the ROI. Furthermore, the third step of the projected work is applied to extract the definite learnable texture features and statistical features from the segmented ROI. The extracted features in the subsequent stage of classification are applied to Hybrid-Layer Convolutional Neural Network (HL-CNN) architecture to classify the lung cancer nodule as either benign or malignant. Principally this research is carried out by contributing to each stage of it, where the novel concept of the improved Hybrid-Layer Convolutional Neural Network (HL-CNN) is employed by optimising and selecting the optimal weight using the Enhanced Cat Swarm Optimisation (ECSO) algorithm. The experimental result of the proposed HL-CNN using the weight optimisation algorithm ECSO is achieved an accuracy of 93%, which is comparatively efficient with respect to existing models such as DBN, SVM, CNN, WOA, MFO, and CSO. Moreover, the proposed model conclusively gives a decision on the detected nodule as either benign or malignant.
  Keywords: Computer Assisted Diagnosis (CAD); Computer Vision; Cancer Diagnosis; Image Classification; Image Enhancement; Image Segmentation; Feature Extraction.
  
  
• Prediction model for total amount of coke oven gas generation based on FCM-RBF  
  by Lili Feng, Jun Peng, Zhaojun Huang 
  Abstract: The rational use of Coke Oven Gas (COG) is of great significance to improve the economic efficiency of enterprises. In this paper, a COG generation prediction model based on fuzzy C-mean clustering (FCM) and radial basis function (RBF) neural network is proposed to address the problems such as the difficulty of accurate modelling of COG generation process and the difficulty of real-time flow prediction. Firstly, the coke oven production process is analysed and correlation analysis is used to select the influencing factors. Secondly, the FCM is used to classify the working conditions of the coke oven, and the appropriate number of working conditions is selected through experiments. Finally, the prediction models under different working conditions are established separately by using RBF. The experiments were carried out using actual industrial production data, and the experimental results showed that the model could provide guidance reference for the dispatchers.
  Keywords: coking oven process; fuzzy C-means clustering; prediction model; radial basis function neural network.
  
  
• Slope stability assessment based on a combined weight-cloud model  
  by Decheng Zhang, Jielu He, Yunjun Yang 
  Abstract: To evaluate the stability of slopes accurately and reduce the risk of landslide accidents during slope treatment, this study focuses on four aspects: slope morphology, rock mass characteristics, geological structure, and triggering factors. Based on these factors and the intrinsic relationships among indicators, the study refines them into 12 secondary indicators to establish a risk assessment system for slope instability. This research aims to address limitations in previous studies, where the combined influence of subjective judgment and objective data was not fully integrated, and uncertainties in slope evaluation indicators were not effectively managed. To improve the accuracy of weighting, a combination of subjective weighting and objective weighting methods is adopted. The cloud model is used to solve the uncertainty and imprecision of the impact evaluation index, and the slope stability grade is determined according to the principle of maximum certainty. Finally, the method is applied to an engineering example, and the slope stability grade obtained is consistent with the actual slope stability grade, demonstrating that this method can accurately evaluate slope stability.
  Keywords: slope instability; stability evaluation; combined weighting; cloud model.
  DOI: 10.1504/IJCAT.2025.10078410
   
  
• Improving autonomous networks using predictive offloading in mobile-edge computing  
  by Himanshi Babbar, Shalli RANI 
  Abstract: The integration of Mobile Edge Computing (MEC) with Vehicular Edge Computing (VEC) to enhance the efficiency of autonomous networks is presented in this research. Traditional offloading strategies rely on reactive approaches, which fail to anticipate workload variations, leading to high latency, suboptimal resource allocation, and increased energy consumption. To address these challenges, we propose an intelligent predictive offloading method that dynamically distributes computing workloads among edge servers and vehicles. The method leverages Vehicle-to-Infrastructure (V2I) and Vehicle-to-Vehicle (V2V) communications to optimise task transfer and mobility prediction. Our approach reduces computational latency by up to 40% and decreases communication costs by 25%, as demonstrated through simulations in a high-speed vehicular environment. Additionally, we show that predictive offloading improves resource utilisation efficiency by 90%, compared to 65% in reactive offloading methods. The proposed framework significantly enhances real-time performance for applications such as traffic management and autonomous driving by minimising delays and ensuring seamless task execution. These results establish predictive offloading as a viable solution for next-generation intelligent transportation systems.
  Keywords: autonomous networks; vehicles communication; mobile edge computing; performance evaluation.
  DOI: 10.1504/IJCAT.2026.10079442
   
  
• An effective and efficient local features reduction scheme for time series classification  
  by Manel Khadiche, Bachir Boucheham, Salah Bougueroua 
  Abstract: Dynamic Time Warping (DTW), the most famous time series alignment method, is known for its high complexity and, often, its weakness in dealing with local details. We propose a variant of the DTW with the aim of enhancing its performance on both above-mentioned aspects. The proposed method is based on two operators: One-Dimensional Local Binary Pattern (1D-LBP), for retaining the most relevant and discriminative features, and the Ramer Douglas Peucker (RDP) line simplification algorithm, for data reduction. These operators are followed by application of the DTW on the so-obtained reduced features. Accordingly, the proposed method is called LBP-RDPDTW. The LBP-RDPDTW is fed to the K-Nearest Neighbors (KNN) classifier, for classification purposes. When applied to the 29 image outlines datasets from the UCR (University of California at Riverside) time series classification and clustering benchmark, the so-obtained classifier outperformed competitive existing methods, including the DTW based KNN.
  Keywords: DTW; one-dimensional local binary pattern; Ramer–Douglas–Pecker algorithm; K-nearest neighbours classifier; UCR time series classification; clustering benchmark.
  DOI: 10.1504/IJCAT.2026.10080346