Internet Technology Letters最新文献

It is vitally important for intelligent transportation systems (ITS) to make use of vehicular ad hoc networks (VANETs) to improve road safety, traffic management, and communication. The mobility of vehicles and dynamic traffic conditions continue to pose challenges to network congestion. In this paper, we suggest a load-balancing strategy for reducing network congestion through the optimization of control packet overhead and the enhancement of data dissemination between roadside units (RSUs) and vehicles. As part of the proposed methodology, real-time traffic conditions, predictive modeling, and intelligent routing algorithms are incorporated to achieve an efficient load distribution among vehicular nodes. Urban VANETs can use the framework to measure performance metrics like packet delivery ratios (PDRs) and energy consumption (EC). Simulations indicate the proposed approach improves overall performance over existing load-balancing approaches by reducing processing delays, minimizing network congestion, and minimizing resource utilization. Network operations need to optimize control packet overhead to achieve a balance between communication efficiency and network stability.

New methods for communication in precision agriculture are required to enable better crop productivity, minimize the leakage of personal information, and reduce computational complexity. Real-time monitoring and automated decision making in precision agriculture are leading to better crop productivity, but current communication approaches face challenges in terms of computational complexity, leakage of personal information and energy efficiency. In smart agricultural networks, a lightweight and robust authentication and communication framework is presented to safeguard signaling and data transmission between users, controllers, and sensors. The proposed protocol uses the lightweight cryptographic primitives, dynamic session-key establishment and privacy-preserving authentication mechanism to achieve the security of communication between users, controllers and sensor nodes. It minimizes computation and communication overhead by inhibiting replay and impersonation attacks, as well as man-in-the-middle attacks. By using the proposed model, a low latency, minimal energy consumption, and minimum communication costs can be achieved while maintaining security and scalability in comparison with existing methods. As an IoT-based agricultural application framework and as part of an autonomous farming system, the framework is particularly suitable.

Human activity recognition (HAR) is one of the most important components of modern healthcare monitoring systems, particularly for providing assistance to the elderly, rehabilitating patients, and continuously monitoring their health. The traditional method of activity recognition often relies on only a single sensing modality, limiting its accuracy and robustness when used in the real world. A hybrid LiDAR-wearable sensor framework is proposed in this study to address this challenge. Using LiDAR sensors and wearable devices to collect motion-related signals, the proposed system integrates spatial information and motion-related data. Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) are used to learn spatial and temporal features from fusion sensor data. Activity classification is improved by extracting features from sensor signals and fusing multimodal features. A wide variety of human activities are accurately detected by the proposed model in experimental tests, such as walking, sitting, standing, jogging, and climbing stairs. CNN, LSTM, and CNN-Bi-LSTM are outperformed by the model, which achieves 99.3% of classification accuracy. Accordingly, the proposed method based on the integration of deep learning algorithm and the usage of LiDAR sensing together with wearable sensor data increases activity recognition systems' accuracy and reliability. Human Activity Recognition (HAR) plays an important role in modern healthcare monitoring, especially for elderly care, rehabilitation, and continuous patient supervision. This study proposes a hybrid LiDAR and wearable sensor framework that combines spatial and motion data for accurate activity recognition. CNN and LSTM models are used to extract spatial and temporal features from fused sensor data. Experimental results show that the proposed model accurately recognizes activities such as walking, sitting, standing, jogging, and stair climbing, achieving 99.3% classification accuracy and outperforming existing CNN, LSTM, and CNN-Bi-LSTM methods. The integration of LiDAR and wearable sensors with deep learning improves the accuracy, reliability, and efficiency of healthcare-oriented HAR system.

The fast expansion of artificial intelligence (AI) technologies across the ecommerce ecosystems has radically changed the processes of how firms interact with consumers. The paper explores the complex effect of AI-based personalization on consumer decision making and the firm-level performance outcomes, based on e-commerce sites. Our proposed Personalization Impact Optimization Framework (PIOF) represents a blend of collaborative filtering, deep neural networks, and reinforcement learning as a framework to model the user preference dynamics in real-time. The mathematical model integrates a multi-objective optimization criterion taking into consideration both consumer utility and revenue of the firm. We show that conversion rates increase with AI-based engines of personalization by a maximum of 34.7%, consumer search costs with AI-based personalization engines decrease by 28.3% and the average value of orders with AI-based personalization engines by 19.6%, compared to controls without personalization. Moreover, our econometric model shows that there is a statistically significant positive correlation between the extent of personalization and the amount of gross merchandise volume (GMV) by the firm. The combination of reinforcement learning and fairness-aware multi-objective optimization led to better performance than sequential recommenders based on transformers. Such results have significant implications on platform strategy, algorithmic governance and the consumer welfare policy.

Industrialization is growing as a threat to the quality of water in urban lakes. In this paper, a hybrid machine learning system (combining remote sensing [RS] data with a stacked ensemble model) is used to predict the turbidity of Saki Lake within the Patancheruvu industrial belt in Hyderabad. As a result of the nonlinear relationship between RS spectral characteristics and in situ turbidity (NTU), conventional models are ineffective. XGBR + NN + GBR as base learners with Ridge Regression as meta-learner minimally decreased MAE to 3.06 NTU (compared to 5.11, 4.17, and 3.81 NTU as un-minimized MAE of individual models). An additional error-correction step following stacking with the XGBR was used to further decrease MAE to 2.18 NTU (R² = 0.86). The two-step framework shows that ensemble learning with specific fault correction can provide scaled water quality monitoring applicable to several contaminated lentic water bodies.

The distinctive features of dynamic topologies, sparse contacts and evolving adversaries, necessitate that trust models for Vehicles Ad-hoc Networks (VANETs) be adaptable. We introduced: a new trust-based model for trust management in VANETs, which is a fully decentralized, modified Q-Learning approach, and enables dynamic assessments of trust for vehicles (nodes). The system has three primary components: (1) a Q-Learning Trust Calculation module which assesses trust in a vehicle based on previous interactions, (2) a Malicious Node Detection (MND) module which detects adversarial nodes through the use of adaptive thresholds, and (3) a Malicious Node Removal (MNR) module which removes adversarial nodes from the network through a process known as collaborative revocation. The combined contributions of these components is that the trust model increases overall network security and trust, and increases the reliability of the data in the network by providing a mechanism that is adaptable and resilient to the majority of attacks encountered in vehicular networks.

Underwater acoustic (UWA) networks suffer from long propagation delays and highly dynamic channels, making transmission scheduling difficult. While cooperative multi-agent reinforcement learning (MARL) can improve performance, its conventional learning framework requires a central scheduler (CS) to collect full state information from all agents, incurring heavy communication overhead especially costly in UWA environments. In this study, we propose a cooperative MARL framework with compressed state sharing. Each agent encodes its local history, that is, queue length, CSI, and past transmission strategy, using a deep learning model to generate a compact state representation, which is then sent to the CS. The CS uses the aggregated compressed state to guide the joint scheduling and power allocation of all the agents. Simulation results confirm that the proposed method significantly reduces communication overhead while preserving high scheduling performance.

With the rapid development of technology, English learning and education are gradually transitioning towards digitalization. Speech sensing recognition technology has brought new possibilities to English teaching. The aim of this study is to explore a blended learning method for English based on speech sensor recognition and data analysis, in order to improve learning effectiveness and student learning experience. Based on the goals of English learning and the needs of students, research is conducted to arrange speech sensors in the learning environment to ensure that they can capture students' speech information. By using speech sensor technology, collect students' speech data and their behavioral data during the learning process to evaluate their learning effectiveness. By tracking the learning process of students, problems can be identified and solved in a timely manner, and indicators such as improvement in pronunciation, learning motivation, and participation can be analyzed to comprehensively understand the effectiveness of teaching methods and student learning performance. By combining speech sensor technology with English learning and utilizing data analysis technology to deeply track and evaluate students' learning process, this comprehensive application provides strong support for blended learning in English, helping to improve learning effectiveness and promote continuous innovation in education.

With the popularization of smart devices, English E-learning, as a convenient and efficient learning method, is receiving more and more attention from people. However, traditional E-learning methods have problems such as dull learning and lack of interactivity, which limit their popularity and application effectiveness among users. Therefore, this article aims to design an English E-learning gamification system based on intelligent devices, and evaluate its user experience and teaching effectiveness through testing. We have researched and constructed an English E-learning gamified mobile learning system, and designed the system architecture. Then, using gamified learning mode and knowledge tracking algorithm, an interactive learning environment was provided. In order to implement the system on smart devices, a data processing model was designed, client data interaction was implemented, and a data stream classification algorithm was developed to improve the performance and user experience of the system. Finally, a user experience test was conducted on the system, including usability testing and teaching effectiveness comparison testing. The E-learning system showed significant advantages in improving English learning ability. Therefore, the gamification design of English E-learning based on intelligent devices has great potential for improving user experience and teaching effectiveness.

With the rapid progress of educational informatization and computer networks, online course teaching resources play a very important role in actual classrooms. Currently, people from all walks of life are paying increasing attention to online teaching. As a new technology for online learning, online teaching will become an important model for future classroom teaching reform. In the network teaching environment, it is important to cultivate learners' innovative thinking and critical thinking ability, and emotional change is a crucial factor. However, due to the inability to conduct face-to-face teaching, learners' emotions are difficult to accurately recognize. Now, the above research results can be used in the evaluation of English classroom quality, and the decision tree can be used as a basis to better build the evaluation index set of the English online teaching system. The algorithm proposed in this paper has shown significant improvement in the accuracy of emotional recognition for learners in English online teaching. Experiments have shown that the method proposed in this paper is effective and can more accurately reflect the learners' emotional state. By identifying the emotions of English online teaching learners, their learning effectiveness can be evaluated, which can greatly improve the progress and quality of English teaching.