The use of IoT-based semantic codecs to process complex contextual semantic information in holographic counterparts introduces significant privacy risks, as it may expose sensitive data, thereby increasing the likelihood of privacy disclosures. The diverse and dynamic nature of holographic counterparts in IoT environments exacerbates these challenges, making it more difficult for semantic codecs to effectively safeguard data privacy. This complexity further intensifies the need for privacy-preserving computation methods, as ensuring the confidentiality and security of the data processed by these codecs becomes a critical concern. However, current privacy protection strategy for multi-party training of semantic codecs relies heavily on the central server for gradient calculation, which may lead to gradient leakage issue. To address this issue, we propose PIMSeC (Privacy-Preserving Interactive Multi-Party Semantic Codec Training for IoT-Based Holographic Counterparts), a novel encryption-based technique that facilitates secure and efficient multi-party interactive training without the dependence on the central server, which enhances both data security and privacy resilience. PIMSeC not only proposes a full interactive secure multi-party deep learning model to protect data privacy during multi-party interactive training, but also, within the above deep learning model, establishes an encrypted additive gradient noise mechanism to ensure post-training semantic codec data privacy. Our theoretical analysis and experimental results demonstrate that PIMSeC promotes semantic codecs privacy protection effectively by interactive secure multi-party training. Compared to the state-of-art methods, PIMSeC achieves a 3% to 15% improvement in terms of accuracy, precision, F1-score, and recall at lower compression rates.
使用基于物联网的语义编解码器来处理全息对应物中的复杂上下文语义信息会带来重大的隐私风险，因为它可能会暴露敏感数据，从而增加隐私泄露的可能性。物联网环境中全息对应物的多样性和动态性加剧了这些挑战，使得语义编解码器更难以有效地保护数据隐私。这种复杂性进一步加强了对隐私保护计算方法的需求，因为确保这些编解码器处理的数据的机密性和安全性成为一个关键问题。然而，目前用于语义编解码器多方训练的隐私保护策略严重依赖中央服务器进行梯度计算，这可能导致梯度泄漏问题。为了解决这个问题，我们提出了PIMSeC （privacy - preserving Interactive Multi-Party Semantic Codec Training for IoT-Based Holographic对像），这是一种基于加密的新技术，可以在不依赖中央服务器的情况下促进安全高效的多方交互训练，从而提高数据安全性和隐私弹性。PIMSeC不仅提出了全交互安全的多方深度学习模型来保护多方交互训练过程中的数据隐私，而且在上述深度学习模型内，建立了加密的加性梯度噪声机制来保证训练后语义编解码器的数据隐私。理论分析和实验结果表明，PIMSeC通过交互式安全多方训练有效地促进了语义编解码器的隐私保护。与最先进的方法相比，PIMSeC在较低的压缩率下，在准确性、精密度、f1分数和召回率方面提高了3%到15%。
{"title":"Privacy-Preserving Interactive Semantic Codec Training for IoT-Based Holographic Counterparts","authors":"Jinpeng Xu;Liang Chen;Limei Lin;Xiaoding Wang;Yanze Huang;Li Xu;Md. Jalil Piran","doi":"10.1109/TCE.2025.3563921","DOIUrl":"https://doi.org/10.1109/TCE.2025.3563921","url":null,"abstract":"The use of IoT-based semantic codecs to process complex contextual semantic information in holographic counterparts introduces significant privacy risks, as it may expose sensitive data, thereby increasing the likelihood of privacy disclosures. The diverse and dynamic nature of holographic counterparts in IoT environments exacerbates these challenges, making it more difficult for semantic codecs to effectively safeguard data privacy. This complexity further intensifies the need for privacy-preserving computation methods, as ensuring the confidentiality and security of the data processed by these codecs becomes a critical concern. However, current privacy protection strategy for multi-party training of semantic codecs relies heavily on the central server for gradient calculation, which may lead to gradient leakage issue. To address this issue, we propose PIMSeC (Privacy-Preserving Interactive Multi-Party Semantic Codec Training for IoT-Based Holographic Counterparts), a novel encryption-based technique that facilitates secure and efficient multi-party interactive training without the dependence on the central server, which enhances both data security and privacy resilience. PIMSeC not only proposes a full interactive secure multi-party deep learning model to protect data privacy during multi-party interactive training, but also, within the above deep learning model, establishes an encrypted additive gradient noise mechanism to ensure post-training semantic codec data privacy. Our theoretical analysis and experimental results demonstrate that PIMSeC promotes semantic codecs privacy protection effectively by interactive secure multi-party training. Compared to the state-of-art methods, PIMSeC achieves a 3% to 15% improvement in terms of accuracy, precision, F1-score, and recall at lower compression rates.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"5287-5299"},"PeriodicalIF":10.9,"publicationDate":"2025-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144867940","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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 : 2025-04-24DOI: 10.1109/TCE.2025.3564011
Zhao Xuejian;Chen Wenxin;Wang Enliang;Hu Yekai
As urbanization progresses, building surfaces increasingly suffer from degradation and structural damage due to prolonged environmental stress, raising significant safety concerns. Consumer-grade drones with embedded vision technology offer a promising approach for intelligent detection of architectural surface anomalies. However, reliance on manually designed network architectures limits their effectiveness, as these struggle to represent complex textures, reduce crack segmentation accuracy, and fail to efficiently leverage the heterogeneous computing resources of drones, hindering widespread adoption in building inspections. To address these issues, we propose a Neural Architecture Search framework based on Dynamic Particle Swarm Optimization (DPSO-NAS). This framework introduces a hardware-aware search space to dynamically adapt architectures to drone computational constraints, a dual-path feature fusion unit using anisotropic convolution to enhance crack feature extraction, and an automated evaluation mechanism to eliminate human bias and ensure optimal model convergence. Experiments show DPSO-NAS outperforms manually designed networks by 4.7–12.3 percentage points in classification accuracy on CIFAR and ImageNet16-120 datasets. In crack segmentation, it achieves a 77.4% mIoU and reduces edge localization errors by 38.6%. On mainstream drone platforms, it improves inference speed by 2.1 times and cuts power consumption by 57%, advancing efficient, scalable inspection solutions.
{"title":"DPSO-NAS: Wall Crack Detection Algorithm Based on Particle Swarm Optimization NAS","authors":"Zhao Xuejian;Chen Wenxin;Wang Enliang;Hu Yekai","doi":"10.1109/TCE.2025.3564011","DOIUrl":"https://doi.org/10.1109/TCE.2025.3564011","url":null,"abstract":"As urbanization progresses, building surfaces increasingly suffer from degradation and structural damage due to prolonged environmental stress, raising significant safety concerns. Consumer-grade drones with embedded vision technology offer a promising approach for intelligent detection of architectural surface anomalies. However, reliance on manually designed network architectures limits their effectiveness, as these struggle to represent complex textures, reduce crack segmentation accuracy, and fail to efficiently leverage the heterogeneous computing resources of drones, hindering widespread adoption in building inspections. To address these issues, we propose a Neural Architecture Search framework based on Dynamic Particle Swarm Optimization (DPSO-NAS). This framework introduces a hardware-aware search space to dynamically adapt architectures to drone computational constraints, a dual-path feature fusion unit using anisotropic convolution to enhance crack feature extraction, and an automated evaluation mechanism to eliminate human bias and ensure optimal model convergence. Experiments show DPSO-NAS outperforms manually designed networks by 4.7–12.3 percentage points in classification accuracy on CIFAR and ImageNet16-120 datasets. In crack segmentation, it achieves a 77.4% mIoU and reduces edge localization errors by 38.6%. On mainstream drone platforms, it improves inference speed by 2.1 times and cuts power consumption by 57%, advancing efficient, scalable inspection solutions.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"6055-6068"},"PeriodicalIF":10.9,"publicationDate":"2025-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868389","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Internet of Medical Things (IoMT) applications encounter issues with data protection, continual adaptation, and domain-specific knowledge retention, especially in consumer-centric IoMT scenarios. We overcome these obstacles and facilitate effective knowledge retention and task adaptation in IoMT applications. This study attempts to create a unique privacy-preserving federated learning framework that combines a hierarchical learning structure with Continual Learning (CL). Despite the advancements in Federated Learning (FL), current models have trouble integrating changing datasets in real-time while protecting privacy, as well as catastrophic forgetting, which occurs when previously learned knowledge is lost when adjusting to new tasks. We present a hierarchical learning framework that makes use of three levels of models - Junior Model (JM), Consultant Model (CM), and Senior Consultant Model (SCM) - to overcome these drawbacks. Each level of the model aids in archived retention and domain-knowledge adaptation. To guarantee that the model maintains valuable information over time and adapts to new tasks with ease, our method blends domain adaptation strategies with ongoing learning approaches like knowledge distillation and elastic weight consolidation (EWC). We compare the suggested methodology with current state-of-the-art (SOTA) models on healthcare datasets for tasks like illness diagnosis and medical image categorization. According to our findings, the hierarchical continual learning model performs better than SOTA techniques in terms of accuracy, task adaptability, and privacy protection. In the healthcare industry, our study sets a new standard for privacy-preserving, continuously adaptable federated learning systems, allowing for real-time, scalable IoMT applications that can adapt dynamically to a variety of changing datasets.
{"title":"Hierarchical Continual Learning for Domain-Knowledge Retention in Healthcare Federated Learning","authors":"Saeed Iqbal;Xiaopin Zhong;Muhammad Attique Khan;Zongze Wu;Dina Abdulaziz AlHammadi;Weixiang Liu;Imran Arshad Choudhry","doi":"10.1109/TCE.2025.3563909","DOIUrl":"https://doi.org/10.1109/TCE.2025.3563909","url":null,"abstract":"Internet of Medical Things (IoMT) applications encounter issues with data protection, continual adaptation, and domain-specific knowledge retention, especially in consumer-centric IoMT scenarios. We overcome these obstacles and facilitate effective knowledge retention and task adaptation in IoMT applications. This study attempts to create a unique privacy-preserving federated learning framework that combines a hierarchical learning structure with Continual Learning (CL). Despite the advancements in Federated Learning (FL), current models have trouble integrating changing datasets in real-time while protecting privacy, as well as catastrophic forgetting, which occurs when previously learned knowledge is lost when adjusting to new tasks. We present a hierarchical learning framework that makes use of three levels of models - Junior Model (JM), Consultant Model (CM), and Senior Consultant Model (SCM) - to overcome these drawbacks. Each level of the model aids in archived retention and domain-knowledge adaptation. To guarantee that the model maintains valuable information over time and adapts to new tasks with ease, our method blends domain adaptation strategies with ongoing learning approaches like knowledge distillation and elastic weight consolidation (EWC). We compare the suggested methodology with current state-of-the-art (SOTA) models on healthcare datasets for tasks like illness diagnosis and medical image categorization. According to our findings, the hierarchical continual learning model performs better than SOTA techniques in terms of accuracy, task adaptability, and privacy protection. In the healthcare industry, our study sets a new standard for privacy-preserving, continuously adaptable federated learning systems, allowing for real-time, scalable IoMT applications that can adapt dynamically to a variety of changing datasets.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"5025-5035"},"PeriodicalIF":10.9,"publicationDate":"2025-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144867974","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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 : 2025-04-24DOI: 10.1109/TCE.2025.3563993
Hazrat Bilal;Muhammad Shamrooz Aslam;Yibin Tian;Inam Ullah;Sarra Ayouni;Athanasios V. Vasilakos
The adoption of agricultural robots, or agrobots, has revolutionized modern farming operations, ranging from crop monitoring to automated harvesting, significantly boosting productivity. Motivated by the rapid advancements in agrobots and their integration into smart agricultural practices, this study proposes an autonomous trajectory tracking system for wheat farms using quadcopter UAVs. To address actuator fault detection, including stuck faults and partial loss of efficiency, a TSF-$H^{infty }$ -SMO (Takagi-Sugeno Fuzzy-based $H^{infty }$ Sliding Mode Observer) fault detection framework is introduced. The approach initiates with the derivation of a TSF (Takagi-Sugeno Fuzzy) attitude control model that integrates an uncertainty term, constructed from the original nonlinear dynamics of the UAV and approximated through local linear models at four equilibrium positions. An actuator fault model is subsequently integrated to develop a comprehensive TSF-UAV model, accounting for actuator faults. The TSF-$H^{infty }$ -SMO is then designed using matrix coordinate transformation to enable precise fault detection. The fault detection capabilities of the TSF-$H^{infty }$ -SMO are evaluated through simulations on the TSF-UAV model under SISO (single-input single-output) actuator fault scenarios. The experimental results validate the proposed system, demonstrating its ability to detect a range of actuator faults accurately and promptly. The analysis reveals a proportional relationship between the amplitude of the state change and the severity of the fault, attributed to the interaction between system states and actuator flaps. This approach underscores the potential for deploying autonomous UAV-based fault detection and trajectory tracking systems in agricultural applications. Furthermore, integrating such advanced fault-tolerant control algorithms holds promise for consumer technology applications, where precision, reliability, and robustness are critical to enhancing system performance and operational efficiency.
{"title":"A Consumer Electronics-Enhanced UAV System for Agricultural Farm Tracking With Fuzzy SMO and Actuator Fault Detection Control Algorithms","authors":"Hazrat Bilal;Muhammad Shamrooz Aslam;Yibin Tian;Inam Ullah;Sarra Ayouni;Athanasios V. Vasilakos","doi":"10.1109/TCE.2025.3563993","DOIUrl":"https://doi.org/10.1109/TCE.2025.3563993","url":null,"abstract":"The adoption of agricultural robots, or agrobots, has revolutionized modern farming operations, ranging from crop monitoring to automated harvesting, significantly boosting productivity. Motivated by the rapid advancements in agrobots and their integration into smart agricultural practices, this study proposes an autonomous trajectory tracking system for wheat farms using quadcopter UAVs. To address actuator fault detection, including stuck faults and partial loss of efficiency, a TSF-<inline-formula> <tex-math>$H^{infty }$ </tex-math></inline-formula>-SMO (Takagi-Sugeno Fuzzy-based <inline-formula> <tex-math>$H^{infty }$ </tex-math></inline-formula> Sliding Mode Observer) fault detection framework is introduced. The approach initiates with the derivation of a TSF (Takagi-Sugeno Fuzzy) attitude control model that integrates an uncertainty term, constructed from the original nonlinear dynamics of the UAV and approximated through local linear models at four equilibrium positions. An actuator fault model is subsequently integrated to develop a comprehensive TSF-UAV model, accounting for actuator faults. The TSF-<inline-formula> <tex-math>$H^{infty }$ </tex-math></inline-formula>-SMO is then designed using matrix coordinate transformation to enable precise fault detection. The fault detection capabilities of the TSF-<inline-formula> <tex-math>$H^{infty }$ </tex-math></inline-formula>-SMO are evaluated through simulations on the TSF-UAV model under SISO (single-input single-output) actuator fault scenarios. The experimental results validate the proposed system, demonstrating its ability to detect a range of actuator faults accurately and promptly. The analysis reveals a proportional relationship between the amplitude of the state change and the severity of the fault, attributed to the interaction between system states and actuator flaps. This approach underscores the potential for deploying autonomous UAV-based fault detection and trajectory tracking systems in agricultural applications. Furthermore, integrating such advanced fault-tolerant control algorithms holds promise for consumer technology applications, where precision, reliability, and robustness are critical to enhancing system performance and operational efficiency.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"6910-6923"},"PeriodicalIF":10.9,"publicationDate":"2025-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868333","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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 : 2025-04-23DOI: 10.1109/TCE.2025.3563421
Yi Huang;Jisong Zeng;Yanting Wei;Miaojiang Chen;Wenjing Xiao;Yang Yang;Zhiquan Liu;Ahmed Farouk;Houbing Herbert Song
In recent years, the widespread use of edge agricultural consumer electronics has greatly contributed to the level of intelligence in agricultural production, bringing higher efficiency and quality. However, offloading all tasks to the cloud incurs significant latency and resource waste, while relying solely on edge computing fails to meet the computational demands of the entire system. To solve the above problems, we introduce the device-edge-cloud (DEC) three-layer architecture, where agri-consumer electronics devices can partially offload tasks to the edge, and the edge can partially offload tasks to the cloud, i.e., agri-consumer electronics can realize device-edge-cloud collaborative computation. Second, we model the joint computation offloading and resource allocation optimization problem as a non-convex optimization and propose a novel Self-Attention Policy Optimization (SAPO) algorithm to solve it. Experiments show that the joint optimization performance of the proposed SAPO exceeds the baseline, and it is suitable for many different models. Compared with fully connected networks, it has better convergence and robustness, with a convergence speed 50% faster than the fully connected networks. The proposed SAPO algorithm has good scalability and adaptability, and has the potential to be extended to smart agricultural computing scenarios with non-convex optimization.
{"title":"Self-Attention Policy Optimization for Task Offloading and Resource Allocation in Low-Carbon Agricultural Consumer Electronic Devices","authors":"Yi Huang;Jisong Zeng;Yanting Wei;Miaojiang Chen;Wenjing Xiao;Yang Yang;Zhiquan Liu;Ahmed Farouk;Houbing Herbert Song","doi":"10.1109/TCE.2025.3563421","DOIUrl":"https://doi.org/10.1109/TCE.2025.3563421","url":null,"abstract":"In recent years, the widespread use of edge agricultural consumer electronics has greatly contributed to the level of intelligence in agricultural production, bringing higher efficiency and quality. However, offloading all tasks to the cloud incurs significant latency and resource waste, while relying solely on edge computing fails to meet the computational demands of the entire system. To solve the above problems, we introduce the device-edge-cloud (DEC) three-layer architecture, where agri-consumer electronics devices can partially offload tasks to the edge, and the edge can partially offload tasks to the cloud, i.e., agri-consumer electronics can realize device-edge-cloud collaborative computation. Second, we model the joint computation offloading and resource allocation optimization problem as a non-convex optimization and propose a novel Self-Attention Policy Optimization (SAPO) algorithm to solve it. Experiments show that the joint optimization performance of the proposed SAPO exceeds the baseline, and it is suitable for many different models. Compared with fully connected networks, it has better convergence and robustness, with a convergence speed 50% faster than the fully connected networks. The proposed SAPO algorithm has good scalability and adaptability, and has the potential to be extended to smart agricultural computing scenarios with non-convex optimization.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"6969-6980"},"PeriodicalIF":10.9,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868309","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Mobile edge computing is considered as a key technology for consumer Internet of Vehicles networks, which provides low-latency, high-reliability network services for end-users. Service migration need to address where to migrate and how to implement service migration procedure based on user mobility. The existing reactive migrating solutions lead to overlong service migration time and end to end latency. The proactive service migration method obtains the target server in advance through mobility prediction, and the service migration procedure starts before communication handover. Based on the above observation, a Mobility Aware Proactive edge Service Migration framework (MAPSM) is proposed in this paper. MAPSM includes the key aspects: 1) predicting the next location of an end user based on ensemble learning method combining recurrent neural networks and geographical embedding Markov chain predictors; 2) using the mobility prediction result to determine the target edge server of service migration, a proactive migration-handover coordinated method is proposed by performing container pre-migration, memory state migration and communication handover. The time planning scheme in the procedure is also designed. Experimental results demonstrate that MAPSM can greatly improve migration performance, effectively reduce end-to-end latency and significantly reduce service migration time. MAPSM outperforms other baseline service migration approaches.
{"title":"MAPSM: Mobility-Aware Proactive Service Migration Framework for Mobile-Edge Computing in Consumer Internet of Vehicles","authors":"Xuhui Zhao;Yan Shi;Shanzhi Chen;Jianghui Liu;Baofeng Ji;Shahid Mumtaz","doi":"10.1109/TCE.2025.3563627","DOIUrl":"https://doi.org/10.1109/TCE.2025.3563627","url":null,"abstract":"Mobile edge computing is considered as a key technology for consumer Internet of Vehicles networks, which provides low-latency, high-reliability network services for end-users. Service migration need to address where to migrate and how to implement service migration procedure based on user mobility. The existing reactive migrating solutions lead to overlong service migration time and end to end latency. The proactive service migration method obtains the target server in advance through mobility prediction, and the service migration procedure starts before communication handover. Based on the above observation, a Mobility Aware Proactive edge Service Migration framework (MAPSM) is proposed in this paper. MAPSM includes the key aspects: 1) predicting the next location of an end user based on ensemble learning method combining recurrent neural networks and geographical embedding Markov chain predictors; 2) using the mobility prediction result to determine the target edge server of service migration, a proactive migration-handover coordinated method is proposed by performing container pre-migration, memory state migration and communication handover. The time planning scheme in the procedure is also designed. Experimental results demonstrate that MAPSM can greatly improve migration performance, effectively reduce end-to-end latency and significantly reduce service migration time. MAPSM outperforms other baseline service migration approaches.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"3753-3766"},"PeriodicalIF":10.9,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144867616","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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 : 2025-04-23DOI: 10.1109/TCE.2025.3563723
Lei Nie;Dandan Qi;Bingyi Liu;Peng Li;Haizhou Bao;Heng He
Efficient traffic signal control is a cost-effective way to ease urban traffic congestion. Multi-agent reinforcement learning (MARL) has become a widely adopted method for optimizing traffic signal control (TSC). However, existing MARL-based methods often focus on a single optimization objective, lacking a comprehensive consideration of traffic efficiency, environmental pollution, and traffic safety. Simultaneously, these methods often fail to effectively capture the dynamic and complex interactions among agents in multi-intersection scenarios, which negatively impacts traffic efficiency. In this article, we propose a collaborative MARL-based method for multi-objective TSC, called CMRM. First, we introduce a multi-objective reward mechanism that integrates traffic efficiency, environmental impact, and safety to guide agents toward more comprehensive optimization. Second, we design a cooperation enhancement module (CEM) based on the graph attention mechanism to dynamically capture neighboring agents’ state information. This mitigates the partial observability problem in independent proximal policy optimization (IPPO) and enhances the model’s ability to capture dynamic and complex interactions among agents. Finally, we assess the performance of the proposed CMRM method using SUMO on two real traffic networks. Experimental results demonstrate that our method significantly improves traffic efficiency while reducing environmental pollution and enhancing traffic safety, compared to the best performing baseline, our method reduces CO2 emission by approximately 17.53% and 9.57%, and lowers vehicle collision risks by 44.39% and 42.85% in two different traffic networks.
{"title":"CMRM: Collaborative Multi-Agent Reinforcement Learning for Multi-Objective Traffic Signal Control","authors":"Lei Nie;Dandan Qi;Bingyi Liu;Peng Li;Haizhou Bao;Heng He","doi":"10.1109/TCE.2025.3563723","DOIUrl":"https://doi.org/10.1109/TCE.2025.3563723","url":null,"abstract":"Efficient traffic signal control is a cost-effective way to ease urban traffic congestion. Multi-agent reinforcement learning (MARL) has become a widely adopted method for optimizing traffic signal control (TSC). However, existing MARL-based methods often focus on a single optimization objective, lacking a comprehensive consideration of traffic efficiency, environmental pollution, and traffic safety. Simultaneously, these methods often fail to effectively capture the dynamic and complex interactions among agents in multi-intersection scenarios, which negatively impacts traffic efficiency. In this article, we propose a collaborative MARL-based method for multi-objective TSC, called CMRM. First, we introduce a multi-objective reward mechanism that integrates traffic efficiency, environmental impact, and safety to guide agents toward more comprehensive optimization. Second, we design a cooperation enhancement module (CEM) based on the graph attention mechanism to dynamically capture neighboring agents’ state information. This mitigates the partial observability problem in independent proximal policy optimization (IPPO) and enhances the model’s ability to capture dynamic and complex interactions among agents. Finally, we assess the performance of the proposed CMRM method using SUMO on two real traffic networks. Experimental results demonstrate that our method significantly improves traffic efficiency while reducing environmental pollution and enhancing traffic safety, compared to the best performing baseline, our method reduces CO2 emission by approximately 17.53% and 9.57%, and lowers vehicle collision risks by 44.39% and 42.85% in two different traffic networks.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"2793-2805"},"PeriodicalIF":10.9,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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 : 2025-04-23DOI: 10.1109/TCE.2025.3563674
Yang Li;Shichao Liu;Li Zhu;Hongwei Wang
Securing secondary frequency control against increasing false data injection (FDI) attacks is crucial in microgrid systems. Although various detection systems (DSs) have been proposed for microgrids, false positives (FPs) and false negatives (FNs) in DSs introduce imperfect observations to the cyber defense system. Improper defense actions may reduce the system performance due to additional time delay and/or resource utilization. This paper designs a decentralized optimal decision-making scheme for cyber-layer defense to secure microgrid secondary frequency control against rational FDI attacks. Besides the capability of tackling imperfect observations from DSs, the proposed optimal defense decision-making scheme can maximize the long-term reward rather than a one-shot reward in response to FDI attacks. A multi-stage security game model is formulated, and cyber-physical states and controllability Gramians are jointly considered in the payoff function. The strategy realization-equivalent rule and Nash equilibrium (NE) are introduced to derive the optimal defense policy. A neural fictitious self-play (NFSP) is introduced to learn the optimal defense strategy. Simulation results show that the proposed method increases the successful defense ratio by 21.29% compared with the stochastic game solution when imperfect observations of DSs are considered.
{"title":"Neural Fictitious-Self Play-Based Cyber-Layer Defense for Frequency Control in Microgrids Against FDI Attacks","authors":"Yang Li;Shichao Liu;Li Zhu;Hongwei Wang","doi":"10.1109/TCE.2025.3563674","DOIUrl":"https://doi.org/10.1109/TCE.2025.3563674","url":null,"abstract":"Securing secondary frequency control against increasing false data injection (FDI) attacks is crucial in microgrid systems. Although various detection systems (DSs) have been proposed for microgrids, false positives (FPs) and false negatives (FNs) in DSs introduce imperfect observations to the cyber defense system. Improper defense actions may reduce the system performance due to additional time delay and/or resource utilization. This paper designs a decentralized optimal decision-making scheme for cyber-layer defense to secure microgrid secondary frequency control against rational FDI attacks. Besides the capability of tackling imperfect observations from DSs, the proposed optimal defense decision-making scheme can maximize the long-term reward rather than a one-shot reward in response to FDI attacks. A multi-stage security game model is formulated, and cyber-physical states and controllability Gramians are jointly considered in the payoff function. The strategy realization-equivalent rule and Nash equilibrium (NE) are introduced to derive the optimal defense policy. A neural fictitious self-play (NFSP) is introduced to learn the optimal defense strategy. Simulation results show that the proposed method increases the successful defense ratio by 21.29% compared with the stochastic game solution when imperfect observations of DSs are considered.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"6109-6119"},"PeriodicalIF":10.9,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868117","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Embedding multi-modal brain network analysis technology into consumer electronics, such as smart wearables, helps enable early intelligent diagnosis of brain diseases. Recent studies confirm that the functional-structural coupling in certain regions is more tightly correlated than in others. However, existing multi-modal methods often directly fuse functional brain networks (FBN) and structural brain networks (SBN), ignoring the regional heterogeneity between them. Additionally, identity information encoded in brain networks may interfere with disease diagnosis. In this paper, we develop a multi-modal brain network fusion method with regional heterogeneity constraints, and design a feature decoupling module to alleviate disease-irrelevant information. Specifically, we first divide FBN and SBN into multiple subnetworks, and introduce penalty weights to reduce the communication cost between cross-modal brain regions within the subnetwork while increasing the cost between different subnetworks. Then, under the regional heterogeneity constraints, we adopt optimal transport to simulate the transfer of brain region hubness from FBN to SBN, thereby effectively integrating the complex cross-modal interactions. Furthermore, we design a feature decoupling module to suppress ineffective features and enhance the discrimination between modality-specific features and multi-modal features. Experimental results show that the proposed method has promising performance and can identify multi-modal biomarkers for brain disease diagnosis.
{"title":"Multi-Modal Brain Network Fusion for Intelligent Diagnostic Devices","authors":"Shengrong Li;Qi Zhu;Liang Sun;Kai Ma;Yixin Ji;Shile Qi;Daoqiang Zhang","doi":"10.1109/TCE.2025.3563691","DOIUrl":"https://doi.org/10.1109/TCE.2025.3563691","url":null,"abstract":"Embedding multi-modal brain network analysis technology into consumer electronics, such as smart wearables, helps enable early intelligent diagnosis of brain diseases. Recent studies confirm that the functional-structural coupling in certain regions is more tightly correlated than in others. However, existing multi-modal methods often directly fuse functional brain networks (FBN) and structural brain networks (SBN), ignoring the regional heterogeneity between them. Additionally, identity information encoded in brain networks may interfere with disease diagnosis. In this paper, we develop a multi-modal brain network fusion method with regional heterogeneity constraints, and design a feature decoupling module to alleviate disease-irrelevant information. Specifically, we first divide FBN and SBN into multiple subnetworks, and introduce penalty weights to reduce the communication cost between cross-modal brain regions within the subnetwork while increasing the cost between different subnetworks. Then, under the regional heterogeneity constraints, we adopt optimal transport to simulate the transfer of brain region hubness from FBN to SBN, thereby effectively integrating the complex cross-modal interactions. Furthermore, we design a feature decoupling module to suppress ineffective features and enhance the discrimination between modality-specific features and multi-modal features. Experimental results show that the proposed method has promising performance and can identify multi-modal biomarkers for brain disease diagnosis.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"3654-3666"},"PeriodicalIF":10.9,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868424","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Unified power quality conditioner (UPQC) can comprehensively address power quality issues related to voltages and currents, but its control system is usually dominated by proportional-integral (PI) controllers, which may affect the system’s operation performance in the face of uncertain interferences. Therefore, in this paper, a control strategy based on passivity fractional-order sliding mode control (PFOSMC) is proposed to enhance the performance and robustness of UPQC. The passivity of system is demonstrated by establishing the Euler-Lagrange model, and then the passivity-based control (PBC) is designed to accelerate the convergence speed of system to errors. In practice, changes in the stable equilibrium points may cause disturbances to the passive control law, so a SMC is employed to optimize PBC by utilizing the advantages of SMC in resisting internal and external disturbances. To suppress the chattering, a fractional-order term is introduced into SMC, which promotes the system to approach the sliding surface more smoothly. After that, the control law of PFOSMC is designed, and the overall control strategy of UPQC is given based on this control law. To verify the superiority of the proposed PFOSMC in operation performance, the comparative experiments on PI, PBC, SMC and PFOSMC are tested under various working conditions from the aspects of overshoot, dynamic response and total harmonic distortion. Experimental results indicate that the proposed PFOSMC-based control strategy can more effectively enhance the UPQC’s operation performance.
{"title":"Performance Enhancement for Unified Power Quality Conditioner Using Passivity Fractional-Order Sliding Mode Control","authors":"Xiaojun Zhao;Haodong Dang;Mengwei Li;Xiaohuan Wang;Hao Ding;Xiaoqiang Guo","doi":"10.1109/TCE.2025.3563356","DOIUrl":"https://doi.org/10.1109/TCE.2025.3563356","url":null,"abstract":"Unified power quality conditioner (UPQC) can comprehensively address power quality issues related to voltages and currents, but its control system is usually dominated by proportional-integral (PI) controllers, which may affect the system’s operation performance in the face of uncertain interferences. Therefore, in this paper, a control strategy based on passivity fractional-order sliding mode control (PFOSMC) is proposed to enhance the performance and robustness of UPQC. The passivity of system is demonstrated by establishing the Euler-Lagrange model, and then the passivity-based control (PBC) is designed to accelerate the convergence speed of system to errors. In practice, changes in the stable equilibrium points may cause disturbances to the passive control law, so a SMC is employed to optimize PBC by utilizing the advantages of SMC in resisting internal and external disturbances. To suppress the chattering, a fractional-order term is introduced into SMC, which promotes the system to approach the sliding surface more smoothly. After that, the control law of PFOSMC is designed, and the overall control strategy of UPQC is given based on this control law. To verify the superiority of the proposed PFOSMC in operation performance, the comparative experiments on PI, PBC, SMC and PFOSMC are tested under various working conditions from the aspects of overshoot, dynamic response and total harmonic distortion. Experimental results indicate that the proposed PFOSMC-based control strategy can more effectively enhance the UPQC’s operation performance.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"2675-2688"},"PeriodicalIF":10.9,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144867850","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：