Pub Date : 2025-05-01DOI: 10.1016/j.pmcj.2025.102061
Tongzhijun Zhu , Ying Lin , Yanzhen Qu , Zediao Liu , Yayu Luo , Tenglong Mao , Ziyi Chen
Performance fairness has always been a key issue in federated learning (FL), however, the pursuit of performance consistency can lead to a trade-off where the accuracy of well-performing clients is compromised to enhance the accuracy of poor-performing clients. To ensure equitable treatment and unbiased outcomes for all participants in the FL process, we propose FedMH, a fair and fast multi-gradient descent federated learning algorithm with reinforced gradient historical empirical information. We have conducted a theoretical analysis of FedMH from the perspectives of fairness and convergence. Extensive experiments are performed on four federated datasets, revealing significant improvements achieved by FedMH compared to state-of-the-art baselines. Moreover, the experimental findings highlight FedMH’s superior performance in fine-grained classification problems when compared to existing advanced baselines. In brief, the proper utilization of gradient historical empirical information helps improve the effectiveness and fairness of FL, making it more suitable for large-scale and heterogeneous distributed environments.
{"title":"Federated learning with empirical insights: Leveraging gradient historical experiences for performance fairness","authors":"Tongzhijun Zhu , Ying Lin , Yanzhen Qu , Zediao Liu , Yayu Luo , Tenglong Mao , Ziyi Chen","doi":"10.1016/j.pmcj.2025.102061","DOIUrl":"10.1016/j.pmcj.2025.102061","url":null,"abstract":"<div><div>Performance fairness has always been a key issue in federated learning (FL), however, the pursuit of performance consistency can lead to a trade-off where the accuracy of well-performing clients is compromised to enhance the accuracy of poor-performing clients. To ensure equitable treatment and unbiased outcomes for all participants in the FL process, we propose FedMH, a fair and fast multi-gradient descent federated learning algorithm with reinforced gradient historical empirical information. We have conducted a theoretical analysis of FedMH from the perspectives of fairness and convergence. Extensive experiments are performed on four federated datasets, revealing significant improvements achieved by FedMH compared to state-of-the-art baselines. Moreover, the experimental findings highlight FedMH’s superior performance in fine-grained classification problems when compared to existing advanced baselines. In brief, the proper utilization of gradient historical empirical information helps improve the effectiveness and fairness of FL, making it more suitable for large-scale and heterogeneous distributed environments.</div></div>","PeriodicalId":49005,"journal":{"name":"Pervasive and Mobile Computing","volume":"110 ","pages":"Article 102061"},"PeriodicalIF":3.0,"publicationDate":"2025-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143922506","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-05-01DOI: 10.1016/j.pmcj.2025.102062
Chuan Ying Peng , Wu Jun Yang , Zhi Xian Chang , Jin Ming Lv , Juan Guo
In mobile vehicular networks, when edge servers (ES) provide services to high-speed moving vehicles, the problem of service interruption is particularly prominent due to the limitation of service coverage, which seriously affects the continuity and quality of services. To solve this problem, this paper proposes a service migration target selection method based on trajectory prediction. The method first predicts the future movement trajectories of vehicles by the TS-LSTM trajectory prediction model to identify potential activity areas and their associated edge servers; then, the target server selection is optimized using Deep Q-Network (DQN), which jointly incorporate delay and load fairness into the optimization objective function. In addition, pre-replication technology is introduced during the service migration process to ensure that the original servers can continue to provide services during the service switchover, allowing the target servers to seamlessly receive tasks, effectively ensuring service continuity. The experimental results show that, compared with the current state-of-the-art, the proposed method has significant advantages in terms of convergence speed, service delay and service stability: the average end-to-end service delay is reduced by 32% and the service rejection rate is reduced by 28%.
{"title":"Trajectory prediction-based migration target selection method for vehicular network services","authors":"Chuan Ying Peng , Wu Jun Yang , Zhi Xian Chang , Jin Ming Lv , Juan Guo","doi":"10.1016/j.pmcj.2025.102062","DOIUrl":"10.1016/j.pmcj.2025.102062","url":null,"abstract":"<div><div>In mobile vehicular networks, when edge servers (ES) provide services to high-speed moving vehicles, the problem of service interruption is particularly prominent due to the limitation of service coverage, which seriously affects the continuity and quality of services. To solve this problem, this paper proposes a service migration target selection method based on trajectory prediction. The method first predicts the future movement trajectories of vehicles by the TS-LSTM trajectory prediction model to identify potential activity areas and their associated edge servers; then, the target server selection is optimized using Deep Q-Network (DQN), which jointly incorporate delay and load fairness into the optimization objective function. In addition, pre-replication technology is introduced during the service migration process to ensure that the original servers can continue to provide services during the service switchover, allowing the target servers to seamlessly receive tasks, effectively ensuring service continuity. The experimental results show that, compared with the current state-of-the-art, the proposed method has significant advantages in terms of convergence speed, service delay and service stability: the average end-to-end service delay is reduced by 32% and the service rejection rate is reduced by 28%.</div></div>","PeriodicalId":49005,"journal":{"name":"Pervasive and Mobile Computing","volume":"110 ","pages":"Article 102062"},"PeriodicalIF":3.0,"publicationDate":"2025-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143922507","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-21DOI: 10.1016/j.pmcj.2025.102048
Hooman Sarvghadi , Andreas Reinhardt , Esther A. Semmelhack
Technology has rapidly evolved over the course of the last decades, and drastically transformed our way of life. Robots are no longer just mechanical aides, but have become collaborators on many tasks. Wearable gadgets have become virtually ubiquitous due to their ability to collect data, monitor health parameters, and assist users in various day-to-day tasks. In recent years, there has been a surge in interest around the use of wearable technologies to collect human psychological parameters for human–robot collaboration. With the field of robotics advancing, there is a growing need for robots to interact with humans seamlessly. To achieve this seamless human–robot connection, robots must be able to interpret human emotions and react appropriately. While understanding human emotions and behavior is a complex task in itself, wearable sensor systems contribute valuable insights. This survey provides a comprehensive overview of wearable gadgets and technologies proposed for measuring five key human factors — trust, cognitive workload, stress, safety perception, and fatigue — within the scope of human–robot collaboration, based on the systematic review of papers published between 2015 and the end of 2024 in six major databases. Our analysis indicates that trust and cognitive workload have received greater attention from researchers in recent years, as compared to other human factors. The Empatica E4 wristband, Shimmer3 GSR+ and EPOC X EEG headset are among the most widely used wearable devices, capable of capturing essential physiological parameters widely used for human–robot collaboration, including electrodermal activity, heart rate variability, skin temperature, and electroencephalogram. Besides reviewing the potentials and capabilities of these gadgets, we highlight their shortcomings and offer directions for future research in this domain.
在过去的几十年里，科技迅速发展，彻底改变了我们的生活方式。机器人不再仅仅是机械助手，而是在许多任务中成为合作者。由于可穿戴设备能够收集数据、监测健康参数并协助用户完成各种日常任务，因此它们几乎无处不在。近年来，人们对使用可穿戴技术来收集人类心理参数以进行人机协作的兴趣激增。随着机器人领域的发展，人们对机器人与人类无缝互动的需求越来越大。为了实现这种无缝的人机连接，机器人必须能够理解人类的情感并做出适当的反应。虽然理解人类的情绪和行为本身就是一项复杂的任务，但可穿戴传感器系统提供了有价值的见解。本调查基于对2015年至2024年底在六个主要数据库中发表的论文的系统综述，全面概述了可穿戴设备和技术，这些设备和技术用于测量人机协作范围内的五个关键人为因素——信任、认知工作量、压力、安全感知和疲劳。我们的分析表明，与其他人为因素相比，信任和认知工作量近年来受到了研究人员的更多关注。Empatica E4腕带、Shimmer3 GSR+和EPOC X EEG耳机是应用最广泛的可穿戴设备，能够捕获广泛用于人机协作的基本生理参数，包括皮电活动、心率变率、皮肤温度和脑电图。除了回顾这些小工具的潜力和能力外，我们还指出了它们的不足之处，并提出了该领域未来的研究方向。
{"title":"A survey of wearable devices to capture human factors for human-robot collaboration","authors":"Hooman Sarvghadi , Andreas Reinhardt , Esther A. Semmelhack","doi":"10.1016/j.pmcj.2025.102048","DOIUrl":"10.1016/j.pmcj.2025.102048","url":null,"abstract":"<div><div>Technology has rapidly evolved over the course of the last decades, and drastically transformed our way of life. Robots are no longer just mechanical aides, but have become collaborators on many tasks. Wearable gadgets have become virtually ubiquitous due to their ability to collect data, monitor health parameters, and assist users in various day-to-day tasks. In recent years, there has been a surge in interest around the use of wearable technologies to collect human psychological parameters for human–robot collaboration. With the field of robotics advancing, there is a growing need for robots to interact with humans seamlessly. To achieve this seamless human–robot connection, robots must be able to interpret human emotions and react appropriately. While understanding human emotions and behavior is a complex task in itself, wearable sensor systems contribute valuable insights. This survey provides a comprehensive overview of wearable gadgets and technologies proposed for measuring five key human factors — trust, cognitive workload, stress, safety perception, and fatigue — within the scope of human–robot collaboration, based on the systematic review of papers published between 2015 and the end of 2024 in six major databases. Our analysis indicates that trust and cognitive workload have received greater attention from researchers in recent years, as compared to other human factors. The Empatica E4 wristband, Shimmer3 GSR+ and EPOC X EEG headset are among the most widely used wearable devices, capable of capturing essential physiological parameters widely used for human–robot collaboration, including electrodermal activity, heart rate variability, skin temperature, and electroencephalogram. Besides reviewing the potentials and capabilities of these gadgets, we highlight their shortcomings and offer directions for future research in this domain.</div></div>","PeriodicalId":49005,"journal":{"name":"Pervasive and Mobile Computing","volume":"110 ","pages":"Article 102048"},"PeriodicalIF":3.0,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143869712","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-12DOI: 10.1016/j.pmcj.2025.102047
Mengxi Liu, Vitor Fortes Rey, Lala Shakti Swarup Ray, Bo Zhou, Paul Lukowicz
While IMU-based Human Activity Recognition (HAR) has achieved significant success in wearable and pervasive computing areas over the past decade, the potential for further improvement of IMU-based HAR performance through the contrastive representation method enhanced by other sensing modalities remains underexplored. In this work, we propose a contrastive representation learning framework to demonstrate that bio-impedance can enhance IMU-based fitness activity recognition beyond the common sensor fusion method, which requires all sensing modalities to be available during both training and inference phases. Instead, in our proposed framework, only the target sensing modality (IMU) is required at inference time. To evaluate our method, we collected both IMU and bio-impedance sensing data through an experiment involving ten subjects performing six types of upper-body and four kinds of lower-body exercises over five days. The bio-impedance-alone classification model achieved an average Macro F1 score of 75.49% and 71.57% for upper-body and lower-body fitness activities, respectively, which was lower than that of the IMU-alone model (83.10% and 78.61%). However, with our proposed method, significant performance improvement (2.66% for upper-body activities and 3.2% for lower-body activities) was achieved by the IMU-only classification model. This improvement leverages the contrastive representation learning framework, where the information from bio-impedance sensing guides the training procedure of the IMU-only model. The results highlight the potential of contrastive representation learning as a valuable tool for advancing fitness activity recognition, with bio-impedance playing a pivotal role in augmenting the capabilities of IMU-based systems.
{"title":"Contrastive-representation IMU-based fitness activity recognition enhanced by bio-impedance sensing","authors":"Mengxi Liu, Vitor Fortes Rey, Lala Shakti Swarup Ray, Bo Zhou, Paul Lukowicz","doi":"10.1016/j.pmcj.2025.102047","DOIUrl":"10.1016/j.pmcj.2025.102047","url":null,"abstract":"<div><div>While IMU-based Human Activity Recognition (HAR) has achieved significant success in wearable and pervasive computing areas over the past decade, the potential for further improvement of IMU-based HAR performance through the contrastive representation method enhanced by other sensing modalities remains underexplored. In this work, we propose a contrastive representation learning framework to demonstrate that bio-impedance can enhance IMU-based fitness activity recognition beyond the common sensor fusion method, which requires all sensing modalities to be available during both training and inference phases. Instead, in our proposed framework, only the target sensing modality (IMU) is required at inference time. To evaluate our method, we collected both IMU and bio-impedance sensing data through an experiment involving ten subjects performing six types of upper-body and four kinds of lower-body exercises over five days. The bio-impedance-alone classification model achieved an average Macro F1 score of 75.49% and 71.57% for upper-body and lower-body fitness activities, respectively, which was lower than that of the IMU-alone model (83.10% and 78.61%). However, with our proposed method, significant performance improvement (2.66% for upper-body activities and 3.2% for lower-body activities) was achieved by the IMU-only classification model. This improvement leverages the contrastive representation learning framework, where the information from bio-impedance sensing guides the training procedure of the IMU-only model. The results highlight the potential of contrastive representation learning as a valuable tool for advancing fitness activity recognition, with bio-impedance playing a pivotal role in augmenting the capabilities of IMU-based systems.</div></div>","PeriodicalId":49005,"journal":{"name":"Pervasive and Mobile Computing","volume":"110 ","pages":"Article 102047"},"PeriodicalIF":3.0,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143844323","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-08DOI: 10.1016/j.pmcj.2025.102045
Sara Lazzaro , Vincenzo De Angelis , Anna Maria Mandalari , Francesco Buccafurri
In the context of consumer Internet of Things (IoT) devices, the identification of vulnerabilities is becoming increasingly relevant. In this paper, we propose a scalable black-box assessment methodology for identifying authentication and reliability issues in IoT devices without the need for prior knowledge of device models or communication protocols. Our methodology consists of a suite of five black-box tests focusing on two specific aspects: authentication and reliability. One of these tests required the development of a tool, called REPLIOT, specifically aimed at discovering replay attacks on the local network. To the best of our knowledge, the development of such a tool is a significant contribution, as there was no similar tool previously available in the literature. We applied these tests to a testbed consisting of 51 consumer IoT devices. Our experiments reveal that 88% of the tested devices fail at least one of the proposed tests. Further manual investigation reveals severe implications of these results in terms of privacy, security, and reliability. Our findings underline a strong need to improve consumer IoT devices security practices to minimize these potential risks and protect smart home environments.
{"title":"A black-box assessment of authentication and reliability in consumer IoT devices","authors":"Sara Lazzaro , Vincenzo De Angelis , Anna Maria Mandalari , Francesco Buccafurri","doi":"10.1016/j.pmcj.2025.102045","DOIUrl":"10.1016/j.pmcj.2025.102045","url":null,"abstract":"<div><div>In the context of consumer Internet of Things (IoT) devices, the identification of vulnerabilities is becoming increasingly relevant. In this paper, we propose a scalable black-box assessment methodology for identifying authentication and reliability issues in IoT devices without the need for prior knowledge of device models or communication protocols. Our methodology consists of a suite of five black-box tests focusing on two specific aspects: authentication and reliability. One of these tests required the development of a tool, called REPLIOT, specifically aimed at discovering replay attacks on the local network. To the best of our knowledge, the development of such a tool is a significant contribution, as there was no similar tool previously available in the literature. We applied these tests to a testbed consisting of 51 consumer IoT devices. Our experiments reveal that 88% of the tested devices fail at least one of the proposed tests. Further manual investigation reveals severe implications of these results in terms of privacy, security, and reliability. Our findings underline a strong need to improve consumer IoT devices security practices to minimize these potential risks and protect smart home environments.</div></div>","PeriodicalId":49005,"journal":{"name":"Pervasive and Mobile Computing","volume":"110 ","pages":"Article 102045"},"PeriodicalIF":3.0,"publicationDate":"2025-04-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143821505","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-07DOI: 10.1016/j.pmcj.2025.102049
Jay Kumar Jain , Dipti Chauhan
Wireless communication is pivotal in the modern era, enabling seamless connectivity across diverse applications. However, the increasing complexity and sophistication of cyber threats pose significant challenges to the security of wireless communication systems. This paper proposes an innovative approach to enhance wireless communication security through integrating artificial intelligence (AI) techniques. First, we construct the network using the Horizontal Partitioning Sierpinski Triangle to reduce the network's high traffic and perform the authentication process. After successful authentication, we perform the clustering process and Game Theory-Driven Clustering (GT-DC) allows nodes to strategically optimize energy utilization while forming clusters as rational entities in a cooperative game. Perform the beacon injection and detect the attacks using the Improved Random Forest (IRF) that signifies the accurate identification of cyber-attacks, IRF is improving the Bootstrap Sampling, Class Weights, and Anomaly Score Threshold. In Routing implement Improved Cache LEACH Protocol (ICLP) which discovers the effective routing establishing the Cache nodes (Cn), to obtain optimal routing by lowering latency, improving data access, enhancing data reliability, and reducing data redundancy. The proposed work is compared with evaluation metrics such as authentication time, throughput, attack detection rate, energy consumption, packet delivery rate, and delay.
{"title":"Optimized secure and energy-efficient approach for IoT-enabled wireless sensor networks","authors":"Jay Kumar Jain , Dipti Chauhan","doi":"10.1016/j.pmcj.2025.102049","DOIUrl":"10.1016/j.pmcj.2025.102049","url":null,"abstract":"<div><div>Wireless communication is pivotal in the modern era, enabling seamless connectivity across diverse applications. However, the increasing complexity and sophistication of cyber threats pose significant challenges to the security of wireless communication systems. This paper proposes an innovative approach to enhance wireless communication security through integrating artificial intelligence (AI) techniques. First, we construct the network using the Horizontal Partitioning Sierpinski Triangle to reduce the network's high traffic and perform the authentication process. After successful authentication, we perform the clustering process and Game Theory-Driven Clustering (GT-DC) allows nodes to strategically optimize energy utilization while forming clusters as rational entities in a cooperative game. Perform the beacon injection and detect the attacks using the Improved Random Forest (IRF) that signifies the accurate identification of cyber-attacks, IRF is improving the Bootstrap Sampling, Class Weights, and Anomaly Score Threshold. In Routing implement Improved Cache LEACH Protocol (ICLP) which discovers the effective routing establishing the Cache nodes (Cn), to obtain optimal routing by lowering latency, improving data access, enhancing data reliability, and reducing data redundancy. The proposed work is compared with evaluation metrics such as authentication time, throughput, attack detection rate, energy consumption, packet delivery rate, and delay.</div></div>","PeriodicalId":49005,"journal":{"name":"Pervasive and Mobile Computing","volume":"110 ","pages":"Article 102049"},"PeriodicalIF":3.0,"publicationDate":"2025-04-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143829093","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-04DOI: 10.1016/j.pmcj.2025.102044
Fahrurrozi Rahman, Martin Schiemer, Andrea Rosales Sanabria, Juan Ye
Human activity recognition (HAR) is a key enabler for many applications in healthcare, factory automation, and smart home. It detects and predicts human behaviours or daily activities via a range of wearable sensors or ambient sensors embedded in an environment. As more and more HAR applications are deployed in the real-world environments, there is a pressing need for the ability of continually and incrementally learning new activities over time without retraining the HAR model. Recently, various continual learning techniques have been applied to HAR; however, most of them commit to a large architecture, which might not suit to devices that deploy HAR models. In addition, these techniques often require to deploy the same large architecture on the devices and cannot customise the architecture for different requirements. To tackle this challenge, we present a dynamic mixture-of-experts approach, which grows an expert for each new task and allows flexible composition of experts to suit individual needs of applications. We have empirically evaluated our technique on 4 third-party, publicly available datasets and compared with 11 state-of-the-art continual learning techniques. Our results demonstrate that our technique can achieve better or comparable performance but with much less parameter spaces and training time.
{"title":"Continual learning in sensor-based human activity recognition with dynamic mixture of experts","authors":"Fahrurrozi Rahman, Martin Schiemer, Andrea Rosales Sanabria, Juan Ye","doi":"10.1016/j.pmcj.2025.102044","DOIUrl":"10.1016/j.pmcj.2025.102044","url":null,"abstract":"<div><div>Human activity recognition (HAR) is a key enabler for many applications in healthcare, factory automation, and smart home. It detects and predicts human behaviours or daily activities via a range of wearable sensors or ambient sensors embedded in an environment. As more and more HAR applications are deployed in the real-world environments, there is a pressing need for the ability of continually and incrementally learning new activities over time without retraining the HAR model. Recently, various continual learning techniques have been applied to HAR; however, most of them commit to a large architecture, which might not suit to devices that deploy HAR models. In addition, these techniques often require to deploy the same large architecture on the devices and cannot customise the architecture for different requirements. To tackle this challenge, we present a dynamic mixture-of-experts approach, which grows an expert for each new task and allows flexible composition of experts to suit individual needs of applications. We have empirically evaluated our technique on 4 third-party, publicly available datasets and compared with 11 state-of-the-art continual learning techniques. Our results demonstrate that our technique can achieve better or comparable performance but with much less parameter spaces and training time.</div></div>","PeriodicalId":49005,"journal":{"name":"Pervasive and Mobile Computing","volume":"110 ","pages":"Article 102044"},"PeriodicalIF":3.0,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143792171","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-01DOI: 10.1016/j.pmcj.2025.102046
Jun Hu , Feiyan Cheng , Meng Liu , Xuanle Xu , Xiaojing Li
Falls are a major public health concern for the aging population, leading to significant injuries, loss of independence, and increased healthcare costs. While wearable devices present promising solutions, existing algorithms are often hindered by the limitations of microcontroller units (MCU) in terms of computational power, memory, and energy consumption. To overcome these challenges, we introduce MicroFallNet, a lightweight convolutional neural network designed for accurate and efficient fall detection. MicroFallNet features a novel FireModel architecture, incorporating Squeeze and Expand layers to optimize computational efficiency and enhance feature extraction. The proposed algorithm demonstrates superior performance on the UMAFALL and FallAllD datasets, achieving geometric mean accuracies of 97.91 % and 97.86 %, respectively, significantly surpassing traditional methods. Additionally, MicroFallNet showcases excellent deployment efficiency across various microcontrollers, particularly excelling on the ESP32 smart wristband platform, where it achieves an inference time of just 30.3 milliseconds. This capability makes MicroFallNet ideally suited for real-time fall detection applications, advancing the development of wearable devices for the elderly and contributing substantially to the field of smart health monitoring. Our code will be publicly available at https://github.com/qwer12330/MicroFallNet-A-Lightweight-Model-for-Real-Time-Fall-Detection-on-Smart-Wristbands-Using-Sm.
{"title":"MicroFallNet: A lightweight model for real-time fall detection on smart wristbands","authors":"Jun Hu , Feiyan Cheng , Meng Liu , Xuanle Xu , Xiaojing Li","doi":"10.1016/j.pmcj.2025.102046","DOIUrl":"10.1016/j.pmcj.2025.102046","url":null,"abstract":"<div><div>Falls are a major public health concern for the aging population, leading to significant injuries, loss of independence, and increased healthcare costs. While wearable devices present promising solutions, existing algorithms are often hindered by the limitations of microcontroller units (MCU) in terms of computational power, memory, and energy consumption. To overcome these challenges, we introduce MicroFallNet, a lightweight convolutional neural network designed for accurate and efficient fall detection. MicroFallNet features a novel FireModel architecture, incorporating Squeeze and Expand layers to optimize computational efficiency and enhance feature extraction. The proposed algorithm demonstrates superior performance on the UMAFALL and FallAllD datasets, achieving geometric mean accuracies of 97.91 % and 97.86 %, respectively, significantly surpassing traditional methods. Additionally, MicroFallNet showcases excellent deployment efficiency across various microcontrollers, particularly excelling on the ESP32 smart wristband platform, where it achieves an inference time of just 30.3 milliseconds. This capability makes MicroFallNet ideally suited for real-time fall detection applications, advancing the development of wearable devices for the elderly and contributing substantially to the field of smart health monitoring. Our code will be publicly available at <span><span>https://github.com/qwer12330/MicroFallNet-A-Lightweight-Model-for-Real-Time-Fall-Detection-on-Smart-Wristbands-Using-Sm</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":49005,"journal":{"name":"Pervasive and Mobile Computing","volume":"109 ","pages":"Article 102046"},"PeriodicalIF":3.0,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143759571","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-31DOI: 10.1016/j.pmcj.2025.102043
Bhabani Sankar Gouda , Trilochan Panigrahi , Sudhakar Das , Meenakshi Panda , Linga Reddy Cenkeramaddi
Sensor nodes in wireless sensor networks (WSNs) for several remote applications are deployed in harsh environments and are coupled with low-cost components. Because of these factors, sensor nodes are becoming faulty, resulting in serious data inaccuracy in the network if not diagnosed in a timely manner. The current approaches to centralized or distributed fault detection algorithms are based on statistical methods or machine learning algorithms. Statistical methods require more data to achieve the desired detection accuracy and may be impractical for sparse networks. Machine learning approaches are computationally demanding. We know that the mean and variance of data from a faulty node differ from those from a healthy node. As a result, simultaneous likelihood ratio statistics are proposed here to determine the sensor node’s fault status in WSNs. The proposed hybrid method, in which the faulty status of the node connected to the anchor node is diagnosed by the anchor node, assumes that the anchor node has statistics for all connected nodes. During the diagnosis time, the simultaneous likelihood ratio statistics (SLRS) are computed using the data received by the anchor node over a specific time period. The fault status is determined by comparing the likelihood ratio to a predetermined threshold based on the tolerance limit. The algorithm’s performance is determined and compared to state-of-the-art algorithms using real-time measured data from the literature.
{"title":"Distributed fault detection in sparse wireless sensor networks utilizing simultaneous likelihood ratio statistics","authors":"Bhabani Sankar Gouda , Trilochan Panigrahi , Sudhakar Das , Meenakshi Panda , Linga Reddy Cenkeramaddi","doi":"10.1016/j.pmcj.2025.102043","DOIUrl":"10.1016/j.pmcj.2025.102043","url":null,"abstract":"<div><div>Sensor nodes in wireless sensor networks (WSNs) for several remote applications are deployed in harsh environments and are coupled with low-cost components. Because of these factors, sensor nodes are becoming faulty, resulting in serious data inaccuracy in the network if not diagnosed in a timely manner. The current approaches to centralized or distributed fault detection algorithms are based on statistical methods or machine learning algorithms. Statistical methods require more data to achieve the desired detection accuracy and may be impractical for sparse networks. Machine learning approaches are computationally demanding. We know that the mean and variance of data from a faulty node differ from those from a healthy node. As a result, simultaneous likelihood ratio statistics are proposed here to determine the sensor node’s fault status in WSNs. The proposed hybrid method, in which the faulty status of the node connected to the anchor node is diagnosed by the anchor node, assumes that the anchor node has statistics for all connected nodes. During the diagnosis time, the simultaneous likelihood ratio statistics (SLRS) are computed using the data received by the anchor node over a specific time period. The fault status is determined by comparing the likelihood ratio to a predetermined threshold based on the tolerance limit. The algorithm’s performance is determined and compared to state-of-the-art algorithms using real-time measured data from the literature.</div></div>","PeriodicalId":49005,"journal":{"name":"Pervasive and Mobile Computing","volume":"110 ","pages":"Article 102043"},"PeriodicalIF":3.0,"publicationDate":"2025-03-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143785287","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-24DOI: 10.1016/j.pmcj.2025.102031
Sumit Kumar , Rajeev Tiwari
The emergence of Internet-of-Things (IoT) has revolutionized communication among devices. IoT devices autonomously collect and disseminate contents to end-users via network routers. There is growing interest in integrating IoT communications with Named Data Networking (NDN) architecture to retrieve and distribute content efficiently. The content caching characteristics of NDN are pivotal in improving Quality-of-Service (QoS) for IoT. However, unlike multimedia content traffic, which tends to remain static, IoT-generated content is inherently transient in nature, and each content has a finite lifespan. As a result, without efficient caching solutions for IoT contents, the network efficiency and user experience would be degraded. Existing caching approaches often overlook the importance of IoT content freshness, its access pattern and the position of routers during content placement decisions in the IoT networks. In this paper, a novel Popularity and Freshness-based Caching (PFC) scheme has been proposed that aims to strategically cache popular and fresh IoT contents on routers located close to the end-user devices. In the proposed solution, the popularity of content is determined using the request history queue deployed on all network routers. For efficient caching decisions, the hop count metric favors routers in close proximity to end-users. Rigorous simulations with realistic network parameters are performed on the realistic IoT network topologies. The simulation results demonstrate that the PFC approach outperforms existing state-of-the-art caching approaches (LCE, LCC, Consumer-Driven, Consumer-Cache, etc.) on several performance parameters: cache hit ratio, network delay, hop count, network traffic, and energy consumption. This makes the PFC caching approach well-suited for NDN-based IoT networks by enabling efficient content caching decisions.
{"title":"Navigating transient content: PFC caching approach for NDN-based IoT networks","authors":"Sumit Kumar , Rajeev Tiwari","doi":"10.1016/j.pmcj.2025.102031","DOIUrl":"10.1016/j.pmcj.2025.102031","url":null,"abstract":"<div><div>The emergence of Internet-of-Things (IoT) has revolutionized communication among devices. IoT devices autonomously collect and disseminate contents to end-users via network routers. There is growing interest in integrating IoT communications with Named Data Networking (NDN) architecture to retrieve and distribute content efficiently. The content caching characteristics of NDN are pivotal in improving Quality-of-Service (QoS) for IoT. However, unlike multimedia content traffic, which tends to remain static, IoT-generated content is inherently transient in nature, and each content has a finite lifespan. As a result, without efficient caching solutions for IoT contents, the network efficiency and user experience would be degraded. Existing caching approaches often overlook the importance of IoT content freshness, its access pattern and the position of routers during content placement decisions in the IoT networks. In this paper, a novel Popularity and Freshness-based Caching (PFC) scheme has been proposed that aims to strategically cache popular and fresh IoT contents on routers located close to the end-user devices. In the proposed solution, the popularity of content is determined using the request history queue deployed on all network routers. For efficient caching decisions, the hop count metric favors routers in close proximity to end-users. Rigorous simulations with realistic network parameters are performed on the realistic IoT network topologies. The simulation results demonstrate that the PFC approach outperforms existing state-of-the-art caching approaches (LCE, LCC, Consumer-Driven, Consumer-Cache, etc.) on several performance parameters: cache hit ratio, network delay, hop count, network traffic, and energy consumption. This makes the PFC caching approach well-suited for NDN-based IoT networks by enabling efficient content caching decisions.</div></div>","PeriodicalId":49005,"journal":{"name":"Pervasive and Mobile Computing","volume":"109 ","pages":"Article 102031"},"PeriodicalIF":3.0,"publicationDate":"2025-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143697403","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号