Search Results (18,363)

Background: Indoor Positioning Systems (IPS) have gained increasing relevance in logistics, offering solutions for safety enhancement, intralogistics management, and material flow control across various environments such as industrial facilities, offices, hospitals, and supermarkets. This study aims to evaluate IPS technologies’ performance and applicability to guide practitioners in selecting systems suited to specific contexts. Methods: The study systematically reviews key IPS technologies, positioning methods, data types, filtering methods, and hybrid technologies, alongside real-world examples of IPS applications in various testing environments. Results: Our findings reveal that radio-based technologies, such as Radio Frequency Identification (RFID), Ultra-wideband (UWB), Wi-Fi, and Bluetooth (BLE), are the most commonly used, with UWB offering the highest accuracy in industrial settings. Geometric methods, particularly multilateration, proved to be the most effective for positioning and are supported by advanced filtering techniques like the Extended Kalman Filter and machine learning models such as Convolutional Neural Networks. Overall, hybrid approaches that integrate multiple technologies demonstrated enhanced accuracy and reliability, effectively mitigating environmental interferences and signal attenuation. Conclusions: The study provides valuable insights for logistics practitioners, emphasizing the importance of selecting IPS technologies suited to specific operational contexts, where precision and reliability are critical to operational success. Full article

In a greenhouse environment where tomatoes are grown, pests in yellow sticky traps need to be detected in order to control the pest population. However, tomato pests typically found on yellow sticky traps are small in size and lack distinct visual features, making it difficult for convolutional networks to extract sufficient contextual information, thereby rendering the tasks of localization and classification exceptionally challenging. In this work, an improved approach based on the advanced object detection model You Only Look Once version 7-tiny (YOLOv7-tiny) is introduced, aiming to enhance the accuracy of detecting small tomato pests while maintaining computational complexity. Firstly, a context information extraction block (CIE) based on a Transformer encoder is proposed, and this block aims to capture global context, explore potential relationships between features, and emphasize important characteristics. Secondly, an Tiny-ELAN fusion network is introduced, which enhanced the feature fusion ability of the network. Thirdly, the feature fusion part takes the P2 feature layer into account and adds a P2 small target detection head. Finally, the SCYLLA-IoU (SIoU) loss function is introduced, and its components are redefined to incorporate direction information, which enhances the model’s learning ability and convergence performance. Experimental results show that our method can accurately detect three insects: whitefly (WF), macrolophus (MR), and nesidiocoris (NC) in the yellow sticky trap images of tomato crops. Compared with Faster R-CNN, SSD, YOLOv3-tiny, YOLOv5s, YOLOv7-tiny, YOLOv7, YOLOv7-x, YOLOv8n, YOLOv8s, YOLOv10n, and RT-DETR, the mean average precision of our method increased by 3.14%, 11.8%, 4.7%, 4.7%, 4.4%, 3.5%, 2.9%, 4.6%, 4.4%, 4.2%, and 4.2%, respectively. Full article

Community of practice (CoP) has been seen as a pivotal support for higher education institutions to implement their social responsibilities. Even though this model is widely admired, assessing its effectiveness and sustainability still faces many challenges: (1) the absence of an appropriate index reveals the significance of CoP; (2) the difficulty of realizing quantitative assessment; and (3) the strategies to improve contribution sustainably by considering CoP development. To address these challenges, a comprehensive Higher Education Social Responsibility Contribution Index (HESRCI) is constructed by taking into account the CoP key influence factors. An FBN model is further developed for the purpose of assessing the various corresponding contributions quantitatively and investigating the potential interdependencies between influence factors. The effectiveness of the proposed approach is evidenced by the quantitative indication of CoP’s contributions to priorities. Research findings also highlight the significance of CoP governance, the mechanism of resource allocation, and team development, in particular, in facilitating the synergy between university development and sustainable socio-economic growth. In addition, it provides data support and a theoretical basis for higher education institutions to make more informed decisions when implementing industry-education integration strategies. Full article

This paper addresses the challenge of preserving user privacy within the Internet of Things (IoT) ecosystem using blockchain technology. Several approaches consider using blockchain and encryption to enhance the privacy of IoT applications and constrained IoT devices. However, existing blockchain platforms such as Ethereum and Hyperledger Fabric already use encryption to store data blocks and secure communication. Therefore, introducing an additional cryptographic layer on top of these platforms could potentially increase processing overhead and reduce response time. In this work, we investigate the integration of IoT and blockchain for privacy preservation. More specifically, we propose a new model that leverages the properties of private blockchain and smart contracts to ensure user data privacy when shared with others. We define policy-based algorithms and notations to assist users in managing smart contracts responsible for registering and controlling their IoT devices. We also specify multiple smart contracts designed to enhance privacy by creating a private channel for communication between the user and the blockchain network. Full article

With the emergence of new applications in mobile networks, users demand higher network stability and lower data transmission delays. When the network address of a mobile user changes, the data transmission path in the wired network may need to be switched to maintain service continuity. Traditional mobility support methods typically rely on a single switching path for all mobile data flows. However, if this path cannot meet the requirements of all the flows, it may lead to network congestion or a decline in user experience. To overcome this limitation, this paper proposes a mobility handover decision method based on multi-topology. It enables the dynamic allocation of mobile data flows across different switching paths within multiple logical topologies. The method models a multi-topology selection problem aimed at minimizing average packet transmission delay and packet loss rate, while considering network conditions and the Quality of Service (QoS) requirements for each flow. By solving the dual problem of the original optimization, a near-optimal solution is achieved. The proposed scheme and algorithm were implemented and tested using the Mininet network simulator. Results show that the proposed approach achieves low average packet transmission delay, low average packet loss rate, and high throughput, compared to traditional single-path switching methods and existing multipath routing methods. Full article

Recently, the radar high-resolution range profiles (HRRPs) have gained significant attention in the field of radar automatic target recognition due to their advantages of being easy to acquire, having a small data footprint, and providing rich target structural information. However, existing recognition methods typically focus on single-domain features, utilizing either the raw HRRP sequence or the extracted feature sequence independently. To fully exploit the multi-domain information present in HRRP sequences, this paper proposes a novel target feature fusion recognition approach. By combining a convolutional long short-term memory (ConvLSTM) network with a cascaded gated recurrent unit (GRU) structure, the proposed method leverages multi-domain and temporal information to enhance recognition performance. Furthermore, a multi-input framework based on learnable parameters is designed to improve target representation capabilities. Experimental results of 6 ship targets demonstrate that the fusion recognition method achieves superior accuracy and faster convergence compared to methods relying on single-domain sequences. It is also found that the proposed method consistently outperforms the other previous methods. And the recognition accuracy is up to 93.32% and 82.15% for full polarization under the SNRs of 20 dB and 5 dB, respectively. Therefore, the proposed method consistently outperforms the previous methods overall. Full article

Drainage is an important aspect of effective water management in row-crop agriculture. Drainage systems can be broadly categorized as either subsurface or surface drainage. A significant amount of design goes into subsurface drainage installations, such as tile networks, and permanent surface drainage installations, such as waterways and berms. However, many farmers also implement temporary surface drainage installations to drain localized areas within their fields each year. This practice involves creating shallow water paths, typically using spinner ditchers, and it is especially commonplace in areas with poor soil permeability. However, this practice is traditionally performed using only observations by farmers and without any data-based workflows. The objective of this study was to analyze the potential yield benefits from a more data-informed approach to surface drainage on a production row-crop farm by exploring corn and soybean yield data from 2008–2021 from two fields where a data-informed approach to surface drainage was implemented. Field topography and drainage information were combined with yield maps from prior years with traditional ad hoc drainage and the years following the incorporation of the data-informed approach to better understand the impact of the workflow. Geospatial distribution of the average normalized crop yields and elevation maps for the fields were analyzed to isolate the yield impacts of the areas affected by the data-informed on-farm surface drainage artifacts. In the years after implementation of the data-informed surface drainage approach, Field 1 and Field 2 showed respective increases of 18.3% and 13.9% in average corn yields. Further analysis isolating three areas affected by the surface drainage using topography and drainage layout showed that all three isolated areas improved more than the field averages, ranging from 15.9–26.5% for Field 1 and 21.4–40.2% for Field 2. Similarly, soybean yields were also higher in the isolated affected areas after the data-informed drainage ditch construction. The findings highlight the effectiveness of data-informed on-farm surface drainage, a relatively straightforward approach that proved beneficial for both soybean and corn production. Full article

Intrusion detection systems (IDSs) are crucial for identifying cyberattacks on in-vehicle networks. To enhance IDS robustness and preserve user data privacy, researchers are increasingly adopting federated learning (FL). However, traditional FL-based IDSs depend on a single central aggregator, creating performance bottlenecks and introducing a single point of failure, thereby compromising robustness and scalability. To address these limitations, this paper proposes a Hierarchical Federated Learning (H-FL) framework to deploy and evaluate the performance of the IDS. The H-FL framework incorporates multiple edge aggregators alongside the central aggregator, mitigating single-point failure risks, improving scalability, and efficiently distributing computational load. We evaluate the proposed IDS using the H-FL framework on two car hacking datasets under realistic non-independent and identically distributed (non-IID) data scenarios. Experimental results demonstrate that deploying the IDS within an H-FL framework can enhance the F1-score by up to 10.63%, addressing the limitations of edge-FL in dataset diversity and attack coverage. Notably, H-FL improved the F1-score in 16 out of 24 evaluated scenarios. By enabling the IDS to learn from diverse data, driving conditions, and evolving threats, this approach substantially strengthens cybersecurity in modern vehicular systems. Full article

This paper addresses the problem of side lobe interference in 5G networks by proposing a unique collaborative beamforming strategy based on Deep Q-Network (DQN) reinforcement learning. Our method, which operates in the sub-6 GHz band, maximizes beam steering and power management by using a two-antenna system with DQN-controlled phase shifters. We provide an OFDM cellular network environment where inter-cell interference is managed while many base stations serve randomly dispersed customers. In order to reduce interference strength and improve signal-to-interference-plus-noise ratio (SINR), the DQN agent learns to modify the interference angle. Our model integrates experience replay memory with a long short-term memory (LSTM) recurrent neural network for time series prediction to enhance learning stability. The outcomes of our simulations show that our suggested DQN approach works noticeably better than current DQN and Q-learning methods. In particular, our technique reaches a maximum of 29.18 dB and a minimum of 5.15 dB, whereas the other approaches only manage 0.77–27.04 dB. Additionally, we significantly decreased the average interference level to 5.42 dB compared to competing approaches of 38.84 dB and 34.12 dB. The average sum-rate capacity is also increased to 3.90 by the suggested strategy, outperforming previous approaches. These findings demonstrate how well our cooperative beamforming method reduces interference and improves overall network performance in 5G systems. Full article

A temporal graph network (TGN) algorithm is introduced to identify fraudulent activities within a digital platform. The central premise is that digital transactions can be modeled via a graph network where various entities interact. The data used to build an event-based temporal graph (ETG) were sourced from an online payment platform and include details such as users, cards, devices, bank accounts, and features related to all these entities. Based on these data, seven distinct graphs were created; the first three represent individual interaction events (card registration, device registration, and bank account registration), while the remaining four are combinations of these graphs (card–device, card–bank account, device–bank account, and card–device–bank account registration). This approach was adopted to determine if the graph’s structure influenced the detection of fraudulent transactions. The results demonstrate that integrating more interaction events into the graph enhances the metrics, meaning graphs containing more interaction events yield superior fraud detection results than those based on individual events. In addition, the data used in this work correspond to Latin American payment transactions, which is relevant in the context of fraud detection since this region has the highest fraud rate in the world, yet few studies have focused on this issue. Full article

Hyper-redundant manipulators based on bionic structures offer superior dexterity due to their large number of degrees of freedom (DOFs) and slim bodies. However, controlling these manipulators is challenging because of infinite inverse kinematic solutions. In this paper, we present a novel reinforcement learning-based control method for hyper-redundant manipulators, integrating path and configuration planning. First, we introduced a deep reinforcement learning-based control method for a multi-target approach, eliminating the need for complicated reward engineering. Then, we optimized the network structure and joint space target points sampling to implement precise control. Furthermore, we designed a variable-reset cycle technique for a continuous multi-target approach without resetting the manipulator, enabling it to complete end-effector trajectory tracking tasks. Finally, we verified the proposed control method in a dynamic simulation environment. The results demonstrate the effectiveness of our approach, achieving a success rate of 98.32% with a 134% improvement using the variable-reset cycle technique. Full article

Surface soil moisture (SSM) is crucial for understanding terrestrial hydrological processes. Despite its widespread use since 2015, the Soil Moisture Active and Passive (SMAP) SSM dataset faces challenges due to its inherent low spatial resolution and data gaps. This study addresses these limitations through a deep learning approach aimed at interpolating missing values and downscaling soil moisture data. The result is a seamless, daily 1 km resolution SSM dataset for China, spanning from 1 January 2016 to 31 December 2022. For the original 9 km daily SMAP products, a convolutional neural network (CNN) with residual connections was employed to achieve the spatially seamless 9 km SSM data, integrating multi-source remote sensing data. Subsequently, auxiliary data including land cover, land surface temperatures, vegetation indices, vegetation temperature drought indices, elevation, and soil texture were integrated into the CNN-based downscaling model to generate the spatially seamless 1 km SSM. Comparative analysis of the spatially seamless 9 km and 1 km SSM datasets with ground observations yielded unbiased root mean square error values of 0.09 cm³/cm³ for both, demonstrating the effectiveness of the downscaling method. This approach provides a promising solution for generating high-resolution, spatially seamless soil moisture data to meet the needs of hydrological, meteorological, and agricultural applications. Full article

Traditional multimodal contrastive learning brings text and its corresponding image closer together as a positive pair, where the text typically consists of fixed sentence structures or specific descriptive statements, and the image features are generally global features (with some fine-grained work using local features). Similar to unimodal self-supervised contrastive learning, this approach can be seen as enforcing a strict identity constraint in a multimodal context. However, due to the inherent complexity of remote sensing images, which cannot be easily described in a single sentence, and the fact that remote sensing images contain rich ancillary information beyond just object features, this strict identity constraint may be insufficient. To fully leverage the characteristics of remote sensing images, we propose a multimodal contrastive learning method for remote sensing image feature extraction, based on positive sample tripartite relaxation, where the model is relaxed in three aspects. The first aspect of relaxation involves both the text and image inputs. By introducing learnable parameters in the language and image branches, instead of relying on fixed sentence structures and fixed image features, the network can achieve a more flexible description of remote sensing images in text and extract ancillary information from the image features, thereby relaxing the input constraints. Second relaxation is achieved through multimodal alignment of various features. By aligning semantic information with the corresponding semantic regions in the images, the method allows for the relaxation of local image features under semantic constraints. This approach addresses the issue of selecting image patches in unimodal settings, where there is no semantic constraint. The proposed method for remote sensing image feature extraction has been validated on four datasets. On the PatternNet dataset, it achieved a 91.1% accuracy with just one-shot. Full article

Hospital overcrowding, driven by both structural management challenges and widespread medical emergencies, has prompted extensive research into machine learning (ML) solutions for predicting patient length of stay (LOS) to optimize bed allocation. While many existing models simplify the LOS prediction problem to a classification task, predicting broad ranges of hospital days, an exact day-based regression model is often crucial for precise planning. Additionally, available data are typically limited and heterogeneous, often collected from a small patient cohort. To address these challenges, we present a novel multimodal ML framework that combines imaging and clinical data to enhance LOS prediction accuracy. Specifically, our approach uses the following: (i) feature extraction from chest CT scans via a convolutional neural network (CNN), (ii) their integration with clinically relevant tabular data from patient exams, refined through a feature selection system to retain only significant predictors. As a case study, we applied this framework to pneumonia patient data collected during the COVID-19 pandemic at two hospitals in Naples, Italy—one specializing in infectious diseases and the other general-purpose. Under our experimental setup, the proposed system achieved an average prediction error of only three days, demonstrating its potential to improve patient flow management in critical care environments. Full article

Universities and schools rely heavily on the ability to forecast student performance, as it enables them to develop efficient strategies for enhancing academic results and averting student attrition. The automation of processes and the management of large datasets generated by technology-enhanced learning tools can facilitate the analysis and processing of these data, which provides crucial insights into the knowledge of students and their engagement with academic endeavors. The method under consideration aims to forecast the academic achievement of students through an ensemble of deep neural networks. The proposed method presents a new feature-ranking mechanism based on existing approaches. This mechanism is effective in identifying the most pertinent features and their correlation with the academic performance of students. The proposed method employs an optimization strategy to concurrently configure and train the deep neural networks within our ensemble system. Furthermore, the proposed ensemble model uses weighted voting among its learning components for more accurate prediction. Put simply, the suggested approach enhances the accuracy of academic performance predictions for students not only by employing weighted ensemble techniques, but also by optimizing the parameters of deep learning models. These experimental outcomes provide evidence that the proposed method outperformed the alternative approaches, accurately predicting student performance with a root-mean-square error (RMSE) value of 1.66, a Mean Absolute Percentage Error (MAPE) value of 9.75, and an R-squared value of 0.7430. These results show a significant improvement compared to the null model (RMSE = 4.05, MAPE = 24.89, and R-squared = 0.2897) and prove the efficiency of the techniques employed in the proposed method. Full article