Volume 15, Issue 1

The high development of sensors and wireless network technology has led to the widespread application of wireless sensor networks in the field of environmental monitoring. How to establish efficient, fast, and stable data collection algorithms has become a hot research field. Given this, a dynamic clustering and multi-hop data collection algorithm is proposed based on the neighbor clustering propagation algorithm, low-power adaptive layered routing protocol, and multi-hop priority strategy. The final experimental results indicated that the dynamic data collection algorithm only entered a significant decay period after 2,000 rounds of data collection, indicating that under the same conditions, the dynamic data collection algorithm had better data transmission performance. In Scenario 1, the survival rate of the dynamic data collection algorithm was still close to 80% at 300 rounds. The dynamic data collection algorithm in Scenario 2 was still close to 75% at 1,500 rounds. In Scenario 3, the remaining algorithms decreased to below 50% after 100 rounds, while the dynamic data collection algorithm remained close to 90% after 300 rounds. The remaining algorithms in Scenario 4 dropped below 50% by 500 rounds, while the dynamic data collection algorithm was still close to 70% by 1,500 rounds. The experiment fully demonstrates that the dynamic data collection algorithm has strong comprehensive performance, the best stability, and the highest energy utilization efficiency. Therefore, the dynamic algorithm proposed in the study has strong survivability and performance advantages in various scenarios.

In the current age, the mental condition of people varies rapidly due to various factors such as social relationships, eating disorders, and economic crisis. There are various factors that can be overcome, such as regular exercise, community engagement, meditation, reading books, and yoga. But there are situations where people cannot undergo the mentioned strategies. As advances in data science and big data continue, there is an increasing availability of drug review datasets. There are various manual and traditional approaches to identify the proper drug and condition with some flaws such as overtime, measurement error of the drug, and high computational complexity. Due to these barriers, cutting-edge technologies are involved in data exploration (data cleaning, data transformation, data integration, etc.) to identify the proper prescription of the condition with machine learning approaches. Furthermore, the proposed work has a threefold unique approach that includes the integration of datasets, the creation of a new dataset, and the focus on exploratory data analysis. In the final step, a novel dataset is created from multiple datasets on behavioral healthcare drug reviews that are compared with individual datasets. The main objective of the work is to satisfy the customer’s health in all aspects. The work is verified by identifying the prescription for popular health conditions such as anxiety, depression, insomnia, panic disorder, and bipolar disorder.

Automatic speech emotion recognition has become an important research subject in the area of speech signal processing. The performance of classification algorithms depends on the features extracted from speech. In this work, a new framework for emotion recognition is proposed based on the long-term average spectrum (LTAS). Our framework is evaluated through a comparative study, where classifiers such as artificial neural network, K-nearest neighbours, logistic regression, Bayesian algorithms, tree-based logistics, and support vector machine were used. The framework was experimentally tested using the well-known Toronto Emotional Speech Set database, and the results were compared against state-of-the-art alternatives, using mel frequency cepstral coefficients, filter bank energies, and chroma coefficient speech coding, on this database. Comparative experiments showed that the use of LTAS achieved higher performance, with accuracies of 96–99% in terms of correct classification of speech emotion, compared with the best performance of 97% for the state-of-the-art alternatives. Different sampling frequencies were used to extract LTAS, and the classifiers were tested individually. The main contribution of this work is to demonstrate that the new framework using LTAS significantly reduces the number of parameters down to 87.5 values per s (approximately), as opposed to the 1,200 values used in the best-performing state-of-the-art alternatives; this means that the process of feature extraction is significantly reduced and the performance in terms of correct classification is improved.

The Internet of Medical Things has enabled the development of smart healthcare systems, allowing users to collect, process, and transmit patient health data to cloud servers for further analysis. Internet of Things devices and cloud servers’ physical distance raise network latency and jitter even with ample storage and processing capability. Additionally, this connectivity raises serious challenges about security and privacy in smart healthcare environments. In this article, we focused on addressing those security concerns in a hierarchical smart healthcare system involving devices, gateway, and medical servers. To show the security issues in the existing solution, we reviewed the Das and Namasudra authentication scheme, identifying vulnerabilities such as replay attacks, device impersonation, and denial of service attacks. This article proposes a Privacy-Preserved Lightweight and Secure Multi-level Authentication (PLASMA) scheme to mitigate these vulnerabilities. PLASMA’s security and key privacy are validated through the Real-Or-Random (ROR) model and simulated using the widely accepted Scyther tool. PLASMA’s performance was evaluated by comparing its computational and communication costs with related schemes, and functional analysis was demonstrated to show the security and efficiency of the scheme. The results prove that PLASMA offers a secure, scalable solution for smart healthcare systems.

With the growth of informatization in education and the need to enhance comprehensive national strength, it is important to promote students’ autonomous learning ability and cultural literacy. Blended teaching (BT) is also a new way that meets the teaching reform and the context of the times. To improve teaching efficiency, a BT design model for Beijing Youmu Technology Co., Ltd (UMU) interactive learning platform (ILP) that integrates genetic algorithm (GA) is investigated. Simulated annealing algorithm is embedded on traditional GA, and the teaching design model is experimentally tested. The results showed that the maximum value of the optimal fitness of the improved GA was 0.94 and the minimum value was 0.92, which were 0.04 and 0.2 higher than the traditional GA, respectively. The minimum value of the average fitness of the improved GA was 0.86 and the maximum value was 0.94, which were 0.15 and 0.12 higher than the traditional GA, respectively. When the initial population size was 150, the performance was best and the computation time was shortest. After BT, the average score of the experimental class increased by 6.33 and the standard deviation decreased to 4.93. The research results indicate that the BT design of the studied UMU ILP has an improvement effect on teaching efficiency and has certain application potential in the field of BT.

In the digital era, computer vision technology has become a key tool in the field of sports analysis, especially widely used in basketball action recognition (BAR) technology. To improve the recognition accuracy and efficiency of technical actions in basketball, a BAR system that integrates three-dimensional convolutional neural network video recognition technology and a single-shot multibox detector (SSD) target detection algorithm is proposed. First, the basketball player’s action sequence is captured by the video recognition technique, and then the SSD target detection algorithm is utilized for real-time target detection and localization of the player’s key parts. By combining these two techniques, it is possible to more accurately recognize and classify a wide range of basketball player’s actions including but not limited to shoot, dribbling, passing, and defending. The experimental results indicated that the proposed model of the study provided the best recognition effect in continuous video images with 16 frames, as well as faster response and convergence speeds, and displayed good stability in the computational process. In the application effect comparison experiments with the comparison algorithms, the average recognition rate of the proposed model was 89.5 and 92.5% on the original and cropped frames, respectively, and the average accuracy mean value was 91.69%. Except for the shoot action, the recognition misjudgment rate for the remaining five basketball actions was the lowest among the three methods, which significantly improved the accuracy and effectiveness of action recognition. In addition, the accuracy of the proposed model in practical applications reached 94.50%, and the real-time performance reached 95 ms/frame, both significantly better than the other two compared models, indicating that the model is suitable for real-time BAR tasks and can provide efficient and accurate results. Graphical abstract

Electric vehicles (EVs) are emerging as major energy consumers, offering numerous environmental and operational advantages such as reduced greenhouse gas emissions and lower reliance on fossil fuels. As the adoption of EVs accelerates globally, accurate forecasting of EV charging demand becomes increasingly critical for maximizing the efficiency, reliability, and profitability of charging infrastructure. However, many existing forecasting models fall short by neglecting the complex and dynamic influence of external factors – particularly weather conditions and calendar variables – which can significantly affect usage patterns. This study presents a robust forecasting framework that integrates historical charging data with both temporal and meteorological information to comprehensively evaluate their individual and combined impacts on EV charging behavior. Leveraging long short-term memory networks – effective in modeling time-series data – we evaluate the impact of contextual features on forecasting performance. Results show that calendar information notably improves accuracy, surpassing the effect of weather data. These insights help EV station operators optimize scheduling, reduce uncertainty in day-ahead energy planning, and support sustainability and grid stability.

This study aims to assess the effectiveness of various algorithms in the realms of supervised and semi-supervised learning applied to three multiclass facial image datasets: JAFFE, Georgia tech, and Yale. The datasets were partitioned into proportions of 80:20, 75:25, and 50:50 for supervised learning, while semi-supervised learning was conducted with labelled and unlabeled data ratios of 20:80, 25:75, and 50:50. The evaluated algorithms include convolutional neural networks (CNNs), decision tree, long short-term memory, K-nearest neighbors (K-NNs), multilayer perceptron, and support vector classifier (SVC), each with varying parameters. Experimental outcomes reveal that the performance of models depends on the dataset partitioning strategies and the type of algorithms used. Specifically, linear and polynomial SVC consistently yield favorable results in supervised learning, particularly demonstrating efficacy on the Georgia tech dataset. Conversely, on the JAFFE and Yale dataset, linear SVC and K-NN emerge as optimal choices. The inclusion of semi-supervised learning enhances insights, particularly evident in the Georgia tech dataset, where the combination of labeled and unlabeled data significantly improves accuracy, especially when leveraging linear SVC algorithm. Although there are some instances of sub-optimal performance in certain algorithms like CNN on specific datasets, this research provides comprehensive insights into the effectiveness of various models in contexts of limited-label learning. The implications of these findings are crucial in advancing the development of adaptive and robust facial recognition systems, especially in navigating datasets characterized by diverse variations and complexities.

In response to the current problems in teaching aesthetic education, such as a single teaching method and lack of innovation, the study uses digital media technology to improve it. First, Maya software is used to construct a three-dimensional (3D) model of ancient architecture, and then the UNity3D engine is used to create a roaming scene for students to experience the heritage and elegance of ancient architecture in the aesthetic education classroom. To enable students to achieve an immersive teaching experience, the study uses Kinect somatosensory technology to enable students to interact with the simulated scenes. The students were first tracked by a combination of random forest algorithm and mean shift algorithm, and then the dynamic time warping algorithm was used for dynamic gesture recognition to ensure that students could move forward and backward through the gestures while using the simulation to navigate through the building. Through the above operations, the study completes the design of a virtual recreation system for ancient cultural buildings based on digital media technology. Through the experiment, it was obtained that 94.81% of the total number of students surveyed believed that this method could stimulate interest in learning about aesthetic education classes. The system designed by the research provides new ideas for the reform of the teaching mode of aesthetic education in universities.

In web service composition, the Quality of Service (QoS) prediction applications based on the statistical point estimation method in accuracy consist of many challenges. Aiming at allowing users to select the web service composition based on their requirements, this study proposed a method based on Bootstrap to estimate and predict the QoS confidence interval for web service composition (QCI-WSC). The QCI-WSC first indicates the structure of the web service composition and simplifies the structure model. Apart from that, the QoS estimation interval can be calculated by the historical QoS data, which are invoked by users. Meanwhile, the user similarity is calculated, and the QoS of web service invoked by the similar users is used to predict QCI-WSC. Finally, the results of user-invoked web service composition QoS are verified by the average interval coverage rate, compared to the actual QoS values and prediction values of the other methods, such as adaptive QoS prediction method based on collaborative filtering (QACF) and QoS-Aware web service recommendation (WSRec). Additionally, in this work, dataset1 in WSDream is adopted to estimate and predict the QCI-WSC. Experiments show that the QoS confidence interval estimation results conform to the exponential distribution, and the validity of the QCI-WSC is proved. Furthermore, the average interval probability of the prediction algorithm was more than 75%. The QCI-WSC can accurately cover the actual QoS values of the web service composition and most of the accurate QoS values predicted by QACF and WSRec. It effectively improves the selectivity of service, which provides web service composition featuring better quality for users.

Line segment detection can offer essential technical assistance for various visual tasks, aiding computer systems in comprehending image content more effectively and executing advanced analysis and applications. Focusing on the fact that most current line segment detection methods predict the center and the displacement maps of two endpoints of a line segment separately without exploiting the coupling between them, this article proposed a dynamic label assignment strategy and an improved deformable convolution for center prediction using displacement priors, which enhances the model’s line segment sensing capability and effectively improves the detection performance. The predicted displacements are used as a priori information to guide the centroid label assignment and deformable convolution sampling of center branches, which significantly improves the performance of centroid prediction. In addition, HRNet and UNet3+ are introduced to enhance the feature expression capability of the backbone network. Finally, experiments show that the s A P 10 sA{P}^{10} metrics of the proposed one-stage model are 0.8 and 6.6% higher than those of the two-stage model efficient line segment detector and descriptor with the best performance in line segment detection on the Wireframe dataset and YorkUrban dataset, respectively.

Achieving high-quality three-dimensional (3D) reconstruction has been a challenging problem due to factors such as motion blur. In this article, we first construct a mathematical model of an iterative relaxation method in reconstructing images, including iterative method and relaxation method. Then, the iterative image derivation model with relaxation factors is constructed by introducing relaxation factors. Next, a motion blur model based on the iterative relaxation method is proposed and combined with the 3D reconstruction method of non-coding points. Finally, the 3D reconstruction in some real scenes is carried out using the motion blur non-coded target 3D reconstruction method based on the iterative relaxation method with error analysis. The results show that the reconstruction accuracy under the optimized path has been improved by two orders of magnitude compared with that under the initial path, and the reconstruction error is basically maintained at about 0.042 mm, with the maximum not exceeding 0.05 mm. This indicates that the proposed method can effectively reduce the reconstruction error and achieve a high reconstruction accuracy. The motion blur non-coded target 3D reconstruction method based on the iterative relaxation method proposed in this article has certain practicality and promotion value.

Cold-start recommendation helps personalize user experiences and improve the relevance of the recommendation system. Despite its importance, cold-start solutions are difficult to develop because new users and items lack interaction data, making user preferences and item relevance prediction difficult. Cold-start recommendation systems face challenges because new users/items often lack historical data, resulting in suboptimal recommendation performance. This study presents the cold-start recommendation network (CSRNet) model to address the critical issue of inadequate data in new user and item recommendations, a common limitation in traditional recommendation systems. Our novel approach uses advanced machine learning techniques to create a dynamic model that adapts to no historical interaction data. Hierarchical density-based clustering groups’ subtle similarities improve recommendation accuracy when combined with transfer learning’s predictive power and Bi-GRU’s sequential data handling. Synergistic techniques optimize cold-start recommendations in this integration. By methodically overcoming data sparsity and improving recommendation quality without historical data, CSRNet sets a new standard for adaptive, accurate, and efficient recommendation systems. We tested the proposed method on a pre-processed, well-defined dataset, dividing it into training and test sets to ensure data quality and model resilience. We evaluated the CSRNet model using accuracy, precision, recall, F 1-score, Root Mean Square Error (RMSE), and Mean Absolute Error. Our model has 90.90% accuracy, 90.2% precision, 90.9% recall, 90.9% F 1-score, 1.0059 RMSE, and 0.8012 MAE, outperforming leading cold-start recommendation approaches. This shows that our model can handle the complexities of cold-start recommendation and apply it across domains to provide personalized and relevant recommendations without extensive historical interaction data.

Considering the shortcomings of traditional video multi-target recognition technology in rapidly identifying criminal suspects in complex scenes, a multi-target recognition optimization method based on video tracking technology is proposed. This method constructs a multi-target recognition algorithm based on video feature matching, and introduces the Kalman filter algorithm to improve the accuracy and real-time recognition of criminal suspects through the definition of feature vector and similarity function. Experiments showed that the model proposed in the study performed exceptionally well in terms of tracking error; the highest precision was 94.75%, the recall rate was 96.59%, the tracking error of the horizontal axis was only 3.75%, and the tracking error of the vertical axis was 3.27%. In the crime detection video application, the accuracy–recall curve of the model was 0.94, and the feature recall rate was 94.83%, verifying the effectiveness and robustness of the model in complex and fast scenes. The results show that the proposed model has good feasibility and robustness in rapidly identifying criminal suspects. In addition, the work offered new technical concepts for improving target tracking precision and adapting to real-time scene changes, opening new research avenues in the field of multi-target recognition.

The nonconvex second-order cone (nonconvex SOC) is a nonconvex extension to the convex second-order cone, in the sense that it consists of any vector divided into two sub-vectors for which the Euclidean norm of the first sub-vector is at least as large as the Euclidean norm of the second sub-vector. This cone can be used to reformulate nonconvex quadratic programs in conic format and can arise in real-world applications. In an attempt to obtain an approximate solution for optimization problems over the nonconvex SOC, in this article, we use a heuristic algorithm based on the alternating direction method of multipliers to solve them, which is the core result of our study. More specifically, the approach is built in two steps: a convex optimization problem comes first, followed by a nonconvex conic optimization. The problem in the second phase can lead to an inexact solution. Our strategy makes use of an approximate projection onto the nonconvex cone. The question of convergence remains open.

Detecting malignancy in pulmonary nodules holds significant clinical importance, yet existing image classification methods often struggle with inadequate feature integration and ineffective loss functions. This study proposes two innovative strategies to address these limitations: first, we introduce a multiscale feature weighted fusion technique that enhances the integration of features across different scales, allowing the model to prioritize critical pixel locations essential for accurate diagnosis. Second, we combine contrastive loss with binary cross-entropy within our training framework to improve learning from both similarities and differences among paired samples, which fosters better discrimination between similar nodules while maintaining sensitivity to variations across classes. Besides, our proposed methodologies demonstrate promising performance improvements in detecting pulmonary nodule malignancy, leading to enhanced performance and reliability compared to conventional approaches.

The rapidly increasing photovoltaic (PV) technology is one of the key renewable energies expected to mitigate the impact of climate change and the energy crisis, which has been widely installed in the past few years. However, the variability of PV power generation creates different negative impacts on the electric grid systems, and a resilient and predictable PV power generation is crucial to stabilize and secure grid operation and promote large-scale PV power integration. This article proposed machine learning-based short-term PV power generation forecasting techniques by using XGBoost, SARIMA, and long short-term memory network (LSTM) algorithms. The experimental results demonstrated that the proposed resilient LSTM solution can accurately predict (around 90% R 2 {R}^{2} and 0.028 root mean squared error) PV power generation with minimum input data.

The Internet of Things (IoT) encompasses a network of interconnected devices that collect, analyze, and exchange vast amounts of data. However, this connectivity creates opportunities for various types of cyberattacks, making IoT systems vulnerable and potentially leading to the compromise of sensitive information. Therefore, developing effective intrusion detection system (IDS) is one of the key challenges in IoT network security. The aim of this study is to develop a machine learning (ML) model for network traffic classification and attack detection in IoT environments. Through a comparative analysis of different algorithms, the study seeks to identify the model with the best performance, which could serve as a foundation for efficient IDS solutions tailored to the specific characteristics of IoT networks. The RT-IoT2022 dataset was used for experimental analysis, providing realistic framework for testing ML models, including k-nearest neighbors, Random Forest, XGBoost, multilayer perceptron, and various 1D convolutional neural network architectures. The study examines preprocessing techniques, focusing on dimensionality reduction (principal component analysis, variance inflation factor, Pearson’s test), outlier detection (interquartile range, Z -score, Isolation Forest), and transformation methods (Box–Cox, RobustScaler, Winsorization). Based on the results of the experiment, the most effective model and preprocessing technique were proposed.

Regulating traffic in cities plays a crucial role in addressing climate change. The rapid advancement of artificial intelligence technologies offers new opportunities for managing urban traffic within the framework of Smart Cities. One common method for classifying traffic is the level of service (LoS) criterion, which evaluates traffic quality based on factors like density, speed, and location-specific characteristics. Because of these variations, LoS must be assessed individually for each location, often with expert assistance. In this article, we propose and compare several approaches for LoS classification using neural networks, fuzzy sets, and high-dimensional random vectors. Our goal is to reduce the reliance of LoS determination on local conditions, making the methodology adaptable across different locations. The results show that all three used methods achieved sufficient accuracy, which supports their potential integration with meteorological and pollution data for further applications.

Ensuring the smooth operation of a railway transport node, such as the maintenance depot, is highly demanding due to the complexity of the system. Many components are closely interconnected, and numerous constraints must be considered. Moreover, all workflow processes are frequently disrupted by a significant number of unexpected interruptions, such as delays, technical failures, or personnel shortages. These challenges highlight the necessity of optimizing both operational planning and real-time control. The field of operational research provides a wide range of optimization methods to improve efficiency in the management of railway transport nodes. However, exact mathematical models often prove unsuitable due to their computational complexity and the time required to obtain solutions, especially when dealing with large-scale operational problems. Instead, heuristic and metaheuristic approaches have emerged as practical alternatives, offering a balance between solution quality and computational time. One of the important aspects of depot management is the assignment of staff, particularly train drivers, to various tasks and activities. Efficient staff allocation is essential to minimize delays, optimize workload distribution, and reduce operational costs. In our research, we focus on two heuristic methods – SA and genetic algorithms – as potential tools for solving the driver assignment problem. These methods have demonstrated their effectiveness in achieving near-optimal solutions within a reasonable time frame, making them well-suited for real-world applications in railway maintenance depots.

Vector database management systems have been recognized as a crucial innovation in the era dominated by artificial intelligence, where vast and high-dimensional datasets are generated at unprecedented scales. These systems are designed to efficiently handle, store, retrieve, and analyze high-dimensional vector data, while uncovering patterns within unstructured and heterogeneous datasets. The ability of vector database management systems to perform fast and accurate similarity searches allows contextual data retrieval. Access to vector databases is often facilitated through application code tailored to proprietary application programming interfaces and query languages, varying in syntax and terminology used among vector database management systems of different vendors. A state of tight coupling, interoperability challenges, and difficulties during transitions between vector database management systems that ultimately affect usability is thereby produced. To address these issues, we propose a model-driven software development solution that incorporates vecDSL – a domain-specific language serving as its central component – to provide a uniform approach to accessing vector databases. The goal of the proposed solution is to have vector database management simplified and its interactions streamlined, thereby ensuring that end-user efficiency is enhanced through the utilization of vecDSL. Concepts uniformly used in vecDSL are expected to ease learning and eliminate database-specific adjustments, while abstraction provided by the language aims to simplify testing and enable efficient performance assessments across different vector database management systems. In this article, we describe the syntax and usage of vecDSL, as well as the application of the MDSD-based solution in supporting the interaction with diverse vector databases. We also include the evaluation of the proposed vecDSL syntax, to examine its ability in addressing current issues and explore its potential for further development.

Applications of modern Internet of Things (IoT) devices generate many tasks that are often real-time and require short response times and elimination of delays. Edge and Fog computing are suitable platforms for processing IoT applications. The contribution of this article is the design and implementation of a new algorithm, improved earliest deadline first (EDF), which provides pseudo-optimal scheduling of IoT tasks on microcloud computing machines by minimizing the completion time of the last task in the schedule (makespan). We experimentally evaluated the algorithm on scaled sets of up to 5000 tasks. The new algorithm ensures that all scheduled IoT tasks meet their deadlines as required and optimizes the completion time of the last task of the schedule significantly better than the original EDF algorithm. In the next part of this article, we focus on the analysis of scheduling large, scaled sets of up to 20000 tasks. Our novel approach consists of a comprehensive analysis and a method for finding task schedules that satisfy the required task deadlines while optimizing two other key parameters, which are makespan and the total power consumption required to compute the tasks according to a given schedule. We designed experiments and based on simulations, measurements, visualizations, computations, and analyses, we found optimized schedules for scaled sets of dimensional task datasets, which is the main contribution of this work.

This study analyzes the resilience of open source smart home platforms, namely, Home Assistant, RaspberryMatic, HomeBridge, Nymea, and OpenHABian, against distributed denial of service (DDoS) attacks such as TCP SYN flood, UDP flood, and Internet Control Message Protocol (ICMP) flood in IPv4 and IPv6 networks. As the IoT ecosystem grows, so does the importance of cybersecurity for smart home platforms. The research evaluates the impact of different attack intensities on the availability and stability of the platforms, comparing their performance in both network protocols. Experimental results show differences in the resilience of each platform. IPv6 showed higher resilience to high frequency DDoS attacks, while IPv4 showed higher stability at moderate load levels. The results highlight the need to optimize network protocols and security mechanisms to increase the reliability and resilience of smart homes to DDoS attacks.

Data structures typically use sequential or hierarchical arrangements of elements. Hierarchical data structures are commonly referred to as trees. Trees can be implemented and applied in various ways from simple to relatively sophisticated structures. Undoubtedly, the most well-known tree structure is the binary tree. Such a tree is essential for efficient implementations of (binary) search trees, which typically provide operations like find, insert, and remove. In addition to these, the operation of tree traversal – that sequentially accesses all the elements – is also very important. Data structure libraries mostly implement traversal using the iterator design pattern. The article examines different approaches to implementing an iterator for a binary tree. The article’s main contribution lies in an experimental comparison of various iterator implementations. The comparison also includes an AI-generated iterator. The results show that the simple, straightforward approach commonly used in standard libraries is the fastest and that the AI-generated iterator performs reasonably well.