Pub Date : 2025-06-02DOI: 10.1109/TCE.2025.3575761
Ashu Taneja;Shalli Rani
Emerging consumer electronics ecosystem leverage on Internet-of-Things (IoT) for real-time data flow among large number of interconnected devices. The integration of uncrewed aerial vehicles (UAVs) offers enhanced mobility, automation and innovative aerial perspectives with applications in environmental monitoring, object tracking, crop monitoring and surveillance. But the UAV assisted consumer IoT suffers from the challenge of poor communication support and poor resource management. This paper presents a UAV aided consumer IoT network which utilises double faced active intelligent reflecting surface (DFA-IRS). The DFA-IRS makes use of element pairs on opposite faces to extend communication support to multiple consumer nodes. Further, an intelligent resource utilization algorithm is proposed that associates each IoT node to each DFA-IRS element using optimal phase shifts and maximum signal-to-interference-plus-noise-ratio (SINR). The mathematical formulations for DFA-IRS signal modelling and achievable rate are also obtained. It is observed that DFA-IRS system with per element maximum amplification gain $alpha ^{max}_{n}$ of 30 dB offers an improvement of 15.4% in achieved data rate over $alpha ^{max}_{n}$ of 20 dB. Also, the achievable rate improves by 7.8% with optimal phase shifts $theta _{opt,n}$ . In the end, the performance of DFA-IRS assisted UAV system scenario is compared with other UAV systems aided with conventional IRSs designs.
新兴的消费电子生态系统利用物联网（IoT）在大量互联设备之间实现实时数据流。无人驾驶飞行器（uav）的集成提供了增强的机动性、自动化和创新的空中视角，应用于环境监测、目标跟踪、作物监测和监视。但无人机辅助消费物联网面临通信支持差和资源管理差的挑战。提出了一种利用双面主动智能反射面（DFA-IRS）的无人机辅助消费物联网网络。DFA-IRS利用相反面的元素对将通信支持扩展到多个消费者节点。此外，提出了一种智能资源利用算法，该算法使用最优相移和最大信噪比（SINR）将每个物联网节点与每个DFA-IRS元素关联。给出了DFA-IRS信号建模和可实现率的数学表达式。结果表明，当单元件最大放大增益$alpha ^{max}_{n}$为30 dB时，DFA-IRS系统的增益提高了15.4倍% in achieved data rate over $alpha ^{max}_{n}$ of 20 dB. Also, the achievable rate improves by 7.8% with optimal phase shifts $theta _{opt,n}$ . In the end, the performance of DFA-IRS assisted UAV system scenario is compared with other UAV systems aided with conventional IRSs designs.
{"title":"Intelligent Resource Utilization in UAV-Assisted Consumer IoT Using DFA-IRS","authors":"Ashu Taneja;Shalli Rani","doi":"10.1109/TCE.2025.3575761","DOIUrl":"https://doi.org/10.1109/TCE.2025.3575761","url":null,"abstract":"Emerging consumer electronics ecosystem leverage on Internet-of-Things (IoT) for real-time data flow among large number of interconnected devices. The integration of uncrewed aerial vehicles (UAVs) offers enhanced mobility, automation and innovative aerial perspectives with applications in environmental monitoring, object tracking, crop monitoring and surveillance. But the UAV assisted consumer IoT suffers from the challenge of poor communication support and poor resource management. This paper presents a UAV aided consumer IoT network which utilises double faced active intelligent reflecting surface (DFA-IRS). The DFA-IRS makes use of element pairs on opposite faces to extend communication support to multiple consumer nodes. Further, an intelligent resource utilization algorithm is proposed that associates each IoT node to each DFA-IRS element using optimal phase shifts and maximum signal-to-interference-plus-noise-ratio (SINR). The mathematical formulations for DFA-IRS signal modelling and achievable rate are also obtained. It is observed that DFA-IRS system with per element maximum amplification gain <inline-formula> <tex-math>$alpha ^{max}_{n}$ </tex-math></inline-formula> of 30 dB offers an improvement of 15.4% in achieved data rate over <inline-formula> <tex-math>$alpha ^{max}_{n}$ </tex-math></inline-formula> of 20 dB. Also, the achievable rate improves by 7.8% with optimal phase shifts <inline-formula> <tex-math>$theta _{opt,n}$ </tex-math></inline-formula>. In the end, the performance of DFA-IRS assisted UAV system scenario is compared with other UAV systems aided with conventional IRSs designs.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"3913-3920"},"PeriodicalIF":10.9,"publicationDate":"2025-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868100","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 : 2025-04-30DOI: 10.1109/TCE.2025.3565850
Mohammed Ayub;El-Sayed M. El-Alfy
Accurately identifying household appliances from power consumption data collected via smart meters opens up new possibilities for improving energy management in smart homes and providing substantial benefits to both utilities and consumers. It enables real-time optimization of energy use, offers personalized savings recommendations, enhances demand forecasting, provides detailed appliance load profiling, and supports the promotion of energy-efficient technologies. While low-resolution consumption data are preferred due to the limited processing capabilities of residential smart meters, they lack the granularity needed to capture detailed consumption patterns, resulting in performance degradation in many cases. This paper explores a novel approach based on a revised version of a vision transformer for household appliance identification using low-resolution and low-volume data. To maintain superior algorithmic performance, we first fuse different time-series imaging to augment and compensate for features that might be missed by a single technique, enabling efficient and robust feature representation. Next, real-time data augmentation and pretrained weights from Hugging Face transformers are leveraged and fine-tuned through transfer learning to enhance model performance with limited data, accelerate the training process, and improve model generalization. We compare three variants of our proposed solution: (i) multi-class classification problem, (ii) multi-label classification problem, and (iii) multi-target appliance-specific classification problem. Extensive experiments on four public datasets (ENERTALK, UK-DALE, iWAE, and REFIT) demonstrate that our proposed multimodal data fusion vision transformer outperforms non-fusion baseline models. It can achieve near-perfect results across multi-class, multi-label, and multi-target tasks, with overall F1 scores above 97% and perfect scores for several appliances. Several cross-house and cross-dataset experiments are also conducted to assess the generalization capability of the models on data from previously unseen households and datasets. Additionally, an ablation study demonstrates the model’s scalability, as well as its computational and energy efficiency under different appliance combinations.
{"title":"Household Appliance Identification Using Vision Transformers and Multimodal Data Fusion","authors":"Mohammed Ayub;El-Sayed M. El-Alfy","doi":"10.1109/TCE.2025.3565850","DOIUrl":"https://doi.org/10.1109/TCE.2025.3565850","url":null,"abstract":"Accurately identifying household appliances from power consumption data collected via smart meters opens up new possibilities for improving energy management in smart homes and providing substantial benefits to both utilities and consumers. It enables real-time optimization of energy use, offers personalized savings recommendations, enhances demand forecasting, provides detailed appliance load profiling, and supports the promotion of energy-efficient technologies. While low-resolution consumption data are preferred due to the limited processing capabilities of residential smart meters, they lack the granularity needed to capture detailed consumption patterns, resulting in performance degradation in many cases. This paper explores a novel approach based on a revised version of a vision transformer for household appliance identification using low-resolution and low-volume data. To maintain superior algorithmic performance, we first fuse different time-series imaging to augment and compensate for features that might be missed by a single technique, enabling efficient and robust feature representation. Next, real-time data augmentation and pretrained weights from Hugging Face transformers are leveraged and fine-tuned through transfer learning to enhance model performance with limited data, accelerate the training process, and improve model generalization. We compare three variants of our proposed solution: (i) multi-class classification problem, (ii) multi-label classification problem, and (iii) multi-target appliance-specific classification problem. Extensive experiments on four public datasets (ENERTALK, UK-DALE, iWAE, and REFIT) demonstrate that our proposed multimodal data fusion vision transformer outperforms non-fusion baseline models. It can achieve near-perfect results across multi-class, multi-label, and multi-target tasks, with overall F1 scores above 97% and perfect scores for several appliances. Several cross-house and cross-dataset experiments are also conducted to assess the generalization capability of the models on data from previously unseen households and datasets. Additionally, an ablation study demonstrates the model’s scalability, as well as its computational and energy efficiency under different appliance combinations.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"2774-2792"},"PeriodicalIF":10.9,"publicationDate":"2025-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144867741","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}
The rapid advancements of technology have encouraged the growth of the Internet of Things (IoT), which has transformed how individuals interact with their environments. Among its many branches, Consumer IoT (CIoT) has emerged as a leading force by integrating IoT elements into everyday devices, enhancing user experiences, and offering intelligent services. In particular, smart home environments powered by CIoT devices are improving the quality of life, specifically for the elderly and individuals with disabilities, through automation and behaviour monitoring. To efficiently analyze human behaviour in such settings, this study proposes a novel lightweight computer vision technique, LCNNCV-HBA (Lightweight Convolutional Neural Network-Based Computer Vision for Human Behavior Analysis), specifically optimized for resource-constrained CIoT devices. The proposed method begins with Median Filtering (MF) to eliminate noise, followed by ConvNeXtTiny, a compact yet effective deep learning architecture used for feature extraction by capturing key spatial patterns from images with minimal resource consumption. For behaviour classification, a stacked denoising autoencoder (SDAE) is employed, while an Improved Sparrow Search Algorithm (ISSA) is used to fine-tune hyperparameters and enhance model performance. Experimental validation conducted on a benchmark image dataset demonstrates the effectiveness of the proposed LCNNCV-HBA approach, achieving a superior accuracy of 98.56%, outperforming existing methods in both efficiency and precision.
{"title":"Lightweight Convolutional Neural Network-Based Computer Vision Model for Human Behavior Analysis on Consumer Internet of Things Devices","authors":"Mohamed Elhoseny;E. Laxmi Lydia;S. Rama Sree;Elvir Akhmetshin;K. Shankar","doi":"10.1109/TCE.2025.3564127","DOIUrl":"https://doi.org/10.1109/TCE.2025.3564127","url":null,"abstract":"The rapid advancements of technology have encouraged the growth of the Internet of Things (IoT), which has transformed how individuals interact with their environments. Among its many branches, Consumer IoT (CIoT) has emerged as a leading force by integrating IoT elements into everyday devices, enhancing user experiences, and offering intelligent services. In particular, smart home environments powered by CIoT devices are improving the quality of life, specifically for the elderly and individuals with disabilities, through automation and behaviour monitoring. To efficiently analyze human behaviour in such settings, this study proposes a novel lightweight computer vision technique, LCNNCV-HBA (Lightweight Convolutional Neural Network-Based Computer Vision for Human Behavior Analysis), specifically optimized for resource-constrained CIoT devices. The proposed method begins with Median Filtering (MF) to eliminate noise, followed by ConvNeXtTiny, a compact yet effective deep learning architecture used for feature extraction by capturing key spatial patterns from images with minimal resource consumption. For behaviour classification, a stacked denoising autoencoder (SDAE) is employed, while an Improved Sparrow Search Algorithm (ISSA) is used to fine-tune hyperparameters and enhance model performance. Experimental validation conducted on a benchmark image dataset demonstrates the effectiveness of the proposed LCNNCV-HBA approach, achieving a superior accuracy of 98.56%, outperforming existing methods in both efficiency and precision.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"5645-5652"},"PeriodicalIF":10.9,"publicationDate":"2025-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144867663","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 : 2025-04-30DOI: 10.1109/TCE.2025.3565962
Misba Sikandar;Ikram Ud Din;Ahmad Almogren;Joel J. P. C. Rodrigues
Stomach Adenocarcinoma (STAD) significantly contributes to global cancer mortality, underscoring the urgent need for precise diagnostic methods. Traditionally, artificial intelligence (AI) methods have relied heavily on imaging techniques like CT, PET, and MRI. However, genomic data represents an underutilized resource for identifying STAD-related genetic mutations. This study explores genomic potential by analyzing amino acid sequences of genes impacted by STAD. We propose a framework integrating Huffman Encoding (HMC) for feature extraction and a Differentially Private Federated Long Short-Term Memory (DPFLSTM) model within a federated learning (FL) setting enhanced by differential privacy (DP). The DPFLSTM framework is specifically tailored for consumer-centric Internet of Medical Things (IoMT) environments, facilitating secure collaboration among diverse consumer medical IoT devices. Our DPFLSTM model achieves notable accuracies of 0.93 in testing and 0.99 in training, highlighting a significant improvement in diagnostic precision and data privacy. Additionally, the HMC feature set improves not only DPFLSTM but also conventional ML models (PSVM, RSVM, RF, BNB, DT). This research establishes a new standard for secure and effective genomic diagnostics, promoting multi-institutional collaboration and integration into IoMT-based clinical decision support systems (CDSS).
{"title":"Differentially Private Federated Learning for Genomic-Based Stomach Adenocarcinoma Detection in Consumer Medical IoT","authors":"Misba Sikandar;Ikram Ud Din;Ahmad Almogren;Joel J. P. C. Rodrigues","doi":"10.1109/TCE.2025.3565962","DOIUrl":"https://doi.org/10.1109/TCE.2025.3565962","url":null,"abstract":"Stomach Adenocarcinoma (STAD) significantly contributes to global cancer mortality, underscoring the urgent need for precise diagnostic methods. Traditionally, artificial intelligence (AI) methods have relied heavily on imaging techniques like CT, PET, and MRI. However, genomic data represents an underutilized resource for identifying STAD-related genetic mutations. This study explores genomic potential by analyzing amino acid sequences of genes impacted by STAD. We propose a framework integrating Huffman Encoding (HMC) for feature extraction and a Differentially Private Federated Long Short-Term Memory (DPFLSTM) model within a federated learning (FL) setting enhanced by differential privacy (DP). The DPFLSTM framework is specifically tailored for consumer-centric Internet of Medical Things (IoMT) environments, facilitating secure collaboration among diverse consumer medical IoT devices. Our DPFLSTM model achieves notable accuracies of 0.93 in testing and 0.99 in training, highlighting a significant improvement in diagnostic precision and data privacy. Additionally, the HMC feature set improves not only DPFLSTM but also conventional ML models (PSVM, RSVM, RF, BNB, DT). This research establishes a new standard for secure and effective genomic diagnostics, promoting multi-institutional collaboration and integration into IoMT-based clinical decision support systems (CDSS).","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"5007-5014"},"PeriodicalIF":10.9,"publicationDate":"2025-04-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868314","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}
In a variety of human-machine interaction (HMI) applications, the high-level techniques based on audio-driven talking face generation are often challenged by the issues of temporal misalignment and low-quality outputs. Recent solutions have sought to improve synchronization by maximizing the similarity between audio-visual pairs. However, the temporal disturbances introduced during the inference phase continue to limit the enhancement of generative performance. Inspired by the intrinsic connection between the segmented static facial image and the stable appearance representation, in this study, two strategies, Manual Temporal Segmentation (MTS) and Static Facial Reference (SFR), are proposed to improve performance during the inference stage. The corresponding functionality consists of: MTS involves segmenting the input video into several clips, effectively reducing the complexity of the inference process, and SFR utilizes static facial references to mitigate the temporal noise generated by dynamic sequences, thereby enhancing the quality of the generated outputs. Substantial experiments on the LRS2 and VoxCeleb2 datasets have demonstrated that the proposed strategies are able to significantly enhance inference performance with the LSE-C and LSE-D metrics, without altering the network architecture or training strategy. For effectiveness validation in realistic scenario applications, a deployment has also been conducted on the healthcare devices with the proposed solution.
{"title":"Audio-Driven Talking Face Generation With Segmented Static Facial References for Customized Health Device Interactions","authors":"Zige Wang;Yashuai Wang;Tianyu Liu;Peng Zhang;Lei Xie;Yangming Guo","doi":"10.1109/TCE.2025.3565518","DOIUrl":"https://doi.org/10.1109/TCE.2025.3565518","url":null,"abstract":"In a variety of human-machine interaction (HMI) applications, the high-level techniques based on audio-driven talking face generation are often challenged by the issues of temporal misalignment and low-quality outputs. Recent solutions have sought to improve synchronization by maximizing the similarity between audio-visual pairs. However, the temporal disturbances introduced during the inference phase continue to limit the enhancement of generative performance. Inspired by the intrinsic connection between the segmented static facial image and the stable appearance representation, in this study, two strategies, Manual Temporal Segmentation (MTS) and Static Facial Reference (SFR), are proposed to improve performance during the inference stage. The corresponding functionality consists of: MTS involves segmenting the input video into several clips, effectively reducing the complexity of the inference process, and SFR utilizes static facial references to mitigate the temporal noise generated by dynamic sequences, thereby enhancing the quality of the generated outputs. Substantial experiments on the LRS2 and VoxCeleb2 datasets have demonstrated that the proposed strategies are able to significantly enhance inference performance with the LSE-C and LSE-D metrics, without altering the network architecture or training strategy. For effectiveness validation in realistic scenario applications, a deployment has also been conducted on the healthcare devices with the proposed solution.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"5404-5413"},"PeriodicalIF":10.9,"publicationDate":"2025-04-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868101","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 : 2025-04-29DOI: 10.1109/TCE.2025.3565105
Muhammad Shamrooz Aslam;Hazrat Bilal;Wen-Jer Chang;Neeraj Kumar;Izaz Ahmad Khan;Athanasios V. Vasilakos
This article investigates the design of robust $H_{infty }$ filters for Takagi–Sugeno (T–S) fuzzy systems with time-varying delays, with a critical challenge in many consumer electronics applications. We extend existing research by incorporating Markovian jump parameters to model system uncertainties and considering internal-like time-varying delays, which are prevalent in real-world scenarios such as wireless communication and networked control systems in consumer devices. To address the complexities introduced by time-varying delays, we employ a three-term approximation model that more accurately captures the system dynamics compared to traditional approaches. Furthermore, we incorporate the time derivative of the membership functions (MFs) into the filter design, and its impact on system stability and performance. To ensure filter stability, we derive novel stability conditions based on the mode-dependent Lyapunov-Krasovskii functional approach, incorporating constraints on the higher bounds of the MF time derivatives. Leveraging the small-gain theorem with scaling techniques, we propose the design of both full-order and reduced-order $H_{infty }$ filters, expressed in terms of linear matrix inequalities (LMIs). These LMI-based solutions provide a systematic and computationally efficient approach for filter synthesis. Finally, we validate the effectiveness of our proposed filtering strategies through illustrative examples drawn from relevant consumer electronics applications, such as noise cancellation in audio devices, image stabilization in cameras, and vibration suppression in wearable devices. The results demonstrate the enhanced robustness and performance of the proposed filters in the presence of time-varying delays and uncertainties.
{"title":"H∞Delayed Filtering of Markov Jump Fuzzy Systems in Consumer Electronics: Input–Output Analysis","authors":"Muhammad Shamrooz Aslam;Hazrat Bilal;Wen-Jer Chang;Neeraj Kumar;Izaz Ahmad Khan;Athanasios V. Vasilakos","doi":"10.1109/TCE.2025.3565105","DOIUrl":"https://doi.org/10.1109/TCE.2025.3565105","url":null,"abstract":"This article investigates the design of robust <inline-formula> <tex-math>$H_{infty }$ </tex-math></inline-formula> filters for Takagi–Sugeno (T–S) fuzzy systems with time-varying delays, with a critical challenge in many consumer electronics applications. We extend existing research by incorporating Markovian jump parameters to model system uncertainties and considering internal-like time-varying delays, which are prevalent in real-world scenarios such as wireless communication and networked control systems in consumer devices. To address the complexities introduced by time-varying delays, we employ a three-term approximation model that more accurately captures the system dynamics compared to traditional approaches. Furthermore, we incorporate the time derivative of the membership functions (MFs) into the filter design, and its impact on system stability and performance. To ensure filter stability, we derive novel stability conditions based on the mode-dependent Lyapunov-Krasovskii functional approach, incorporating constraints on the higher bounds of the MF time derivatives. Leveraging the small-gain theorem with scaling techniques, we propose the design of both full-order and reduced-order <inline-formula> <tex-math>$H_{infty }$ </tex-math></inline-formula> filters, expressed in terms of linear matrix inequalities (LMIs). These LMI-based solutions provide a systematic and computationally efficient approach for filter synthesis. Finally, we validate the effectiveness of our proposed filtering strategies through illustrative examples drawn from relevant consumer electronics applications, such as noise cancellation in audio devices, image stabilization in cameras, and vibration suppression in wearable devices. The results demonstrate the enhanced robustness and performance of the proposed filters in the presence of time-varying delays and uncertainties.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"7002-7013"},"PeriodicalIF":10.9,"publicationDate":"2025-04-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868088","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 : 2025-04-29DOI: 10.1109/TCE.2025.3565011
Tun Wang;Yuan He;Mengyan Hao
This paper proposes an intelligent hybrid framework based on Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks with Particle Swarm Optimization (PSO) to enhance real-time cyber threat detection in smart cities. It tries to address cybersecurity challenges posed by the Internet of Things (IoT) devices in smart cities and the necessity for timely responses to emerging threats. The model incorporates the collection and preprocessing of sequential data from network traffic logs, followed by the design and implementation of an RNN-LSTM model tailored for temporal pattern recognition. PSO is deployed to optimize the model’s hyperparameters when offline, achieving significant improvements in detection accuracy and latency. The results indicate an appropriate detection accuracy of 96% and a recall rate of 95.4%, demonstrating the effectiveness of the proposed framework. This research shows the importance of dynamic optimization techniques in adapting to the evolving security landscape of smart cities. It also highlights the critical role of machine learning in safeguarding urban infrastructure and enhancing the resilience of smart city environments against cyber threats.
{"title":"Real-Time Cyber Threat Detection in Smart Cities Using Artificial Intelligence","authors":"Tun Wang;Yuan He;Mengyan Hao","doi":"10.1109/TCE.2025.3565011","DOIUrl":"https://doi.org/10.1109/TCE.2025.3565011","url":null,"abstract":"This paper proposes an intelligent hybrid framework based on Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks with Particle Swarm Optimization (PSO) to enhance real-time cyber threat detection in smart cities. It tries to address cybersecurity challenges posed by the Internet of Things (IoT) devices in smart cities and the necessity for timely responses to emerging threats. The model incorporates the collection and preprocessing of sequential data from network traffic logs, followed by the design and implementation of an RNN-LSTM model tailored for temporal pattern recognition. PSO is deployed to optimize the model’s hyperparameters when offline, achieving significant improvements in detection accuracy and latency. The results indicate an appropriate detection accuracy of 96% and a recall rate of 95.4%, demonstrating the effectiveness of the proposed framework. This research shows the importance of dynamic optimization techniques in adapting to the evolving security landscape of smart cities. It also highlights the critical role of machine learning in safeguarding urban infrastructure and enhancing the resilience of smart city environments against cyber threats.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"4744-4750"},"PeriodicalIF":10.9,"publicationDate":"2025-04-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144868155","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}
Large language models (LLMs) have made substantial advancements in knowledge reasoning and are increasingly utilized in specialized domains such as code completion, legal analysis, and medical transcription, where accuracy is paramount. In such applications, document-specific precision is more critical than general reasoning capabilities. This paper proposes a novel approach based on Retrieval-Augmented Fine-Tuning (RAFT) to enhance model-generated outputs, particularly in code transformation tasks. RAFT integrates domain-specific knowledge, optimizing in-domain retrieval-augmented generation by training the model to discern the relationship between prompts, retrieved documents, and target outputs. This enables the model to extract relevant information while mitigating the impact of noise. Experimental results demonstrate that the proposed method improves accuracy of 2.4% and CodeBLEU of 1.3% for VB-to-C# code conversion, highlighting its effectiveness in domain-specific applications.
{"title":"Enhancing Code Transformation in Large Language Models Through Retrieval-Augmented Fine-Tuning","authors":"Jing-Ming Guo;Po-Yang Liu;Yi-Chong Zeng;Ting-Ju Chen","doi":"10.1109/TCE.2025.3565294","DOIUrl":"https://doi.org/10.1109/TCE.2025.3565294","url":null,"abstract":"Large language models (LLMs) have made substantial advancements in knowledge reasoning and are increasingly utilized in specialized domains such as code completion, legal analysis, and medical transcription, where accuracy is paramount. In such applications, document-specific precision is more critical than general reasoning capabilities. This paper proposes a novel approach based on Retrieval-Augmented Fine-Tuning (RAFT) to enhance model-generated outputs, particularly in code transformation tasks. RAFT integrates domain-specific knowledge, optimizing in-domain retrieval-augmented generation by training the model to discern the relationship between prompts, retrieved documents, and target outputs. This enables the model to extract relevant information while mitigating the impact of noise. Experimental results demonstrate that the proposed method improves accuracy of 2.4% and CodeBLEU of 1.3% for VB-to-C# code conversion, highlighting its effectiveness in domain-specific applications.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 1","pages":"2342-2346"},"PeriodicalIF":4.3,"publicationDate":"2025-04-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144308333","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 : 2025-04-29DOI: 10.1109/TCE.2025.3565390
Qi Zhang;Yuwei Ding;Weiqi Zhang;Yian Zhu;Bob Zhang;Jerry Chun-Wei Lin
Image denoising has been used in various edge computing scenarios such as consumer electronics to improve the image quality and user experience. Existing image denoising methods based on Convolutional Neural Networks (CNNs) and vision Transformers achieve good performance by empirically utilizing residual neural network (ResNet) as basic component. However, ResNet lacks interpretability in network design and also long-term memory of features, potentially restricting the performance of denoising networks. In this paper, we propose an Implicit Multi-scale Swin Transformer Network (IMSNet) for image denoising, which introduces the implicit Euler scheme from the feature memory perspective. Specifically, densely connected implicit feature extraction blocks (IFEBs) are designed to learn the residual mapping between noisy and clean images. The IFEB reformulates the initial skip connection of ResNet based on the implicit Euler discretization, providing both network interpretability and long-term memory. In IFEB, the multi-scale swin Transformer block (MSB) is designed as the implicit layer to capture spatial details and non-local contextual information at different scales. Additionally, a cross-layer feature fusion block (CLFF) is proposed to further improve feature reuse capabilities. Compared to existing denoising networks, the extensive experiments demonstrate the superior performance of our IMSNet in various image denoising tasks, and the flexibility in the practical applications with proposed light model on resource-restricted platforms such as consumer electronic devices.
{"title":"Implicit Multi-Scale Swin Transformer Network for Image Denoising","authors":"Qi Zhang;Yuwei Ding;Weiqi Zhang;Yian Zhu;Bob Zhang;Jerry Chun-Wei Lin","doi":"10.1109/TCE.2025.3565390","DOIUrl":"https://doi.org/10.1109/TCE.2025.3565390","url":null,"abstract":"Image denoising has been used in various edge computing scenarios such as consumer electronics to improve the image quality and user experience. Existing image denoising methods based on Convolutional Neural Networks (CNNs) and vision Transformers achieve good performance by empirically utilizing residual neural network (ResNet) as basic component. However, ResNet lacks interpretability in network design and also long-term memory of features, potentially restricting the performance of denoising networks. In this paper, we propose an Implicit Multi-scale Swin Transformer Network (IMSNet) for image denoising, which introduces the implicit Euler scheme from the feature memory perspective. Specifically, densely connected implicit feature extraction blocks (IFEBs) are designed to learn the residual mapping between noisy and clean images. The IFEB reformulates the initial skip connection of ResNet based on the implicit Euler discretization, providing both network interpretability and long-term memory. In IFEB, the multi-scale swin Transformer block (MSB) is designed as the implicit layer to capture spatial details and non-local contextual information at different scales. Additionally, a cross-layer feature fusion block (CLFF) is proposed to further improve feature reuse capabilities. Compared to existing denoising networks, the extensive experiments demonstrate the superior performance of our IMSNet in various image denoising tasks, and the flexibility in the practical applications with proposed light model on resource-restricted platforms such as consumer electronic devices.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"5584-5594"},"PeriodicalIF":10.9,"publicationDate":"2025-04-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144867641","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 : 2025-04-29DOI: 10.1109/TCE.2025.3565596
Guangwei Zhang;Zhenyu Wu
This paper proposes an AI-based decision support system for enhancing energy efficiency in autonomous systems within consumer electronics. The primary objective is to optimize energy consumption by leveraging advanced Recurrent Neural Networks (RNNs) combined with Adaptive Moment Estimation (Adam). The proposed intelligent framework incorporates real-time data processing, prediction, and adaptive decision-making to reduce energy usage across a variety of devices, such as smart thermostats, robotic vacuum cleaners, and IoT-enabled lighting systems. Key methodologies include the integration of RNNs to model temporal energy consumption patterns and the Adam optimizer for efficient model training. The results demonstrate significant energy savings with improvements in prediction accuracy compared to traditional approaches. The findings suggest that the AI-based system can provide substantial reductions in energy consumption while maintaining or improving device performance. The research has important implications for the future of energy-efficient consumer electronics, particularly as the demand for smart devices grows.
{"title":"Intelligent Decision Support Systems for Energy-Efficient Autonomous Systems in Consumer Electronics","authors":"Guangwei Zhang;Zhenyu Wu","doi":"10.1109/TCE.2025.3565596","DOIUrl":"https://doi.org/10.1109/TCE.2025.3565596","url":null,"abstract":"This paper proposes an AI-based decision support system for enhancing energy efficiency in autonomous systems within consumer electronics. The primary objective is to optimize energy consumption by leveraging advanced Recurrent Neural Networks (RNNs) combined with Adaptive Moment Estimation (Adam). The proposed intelligent framework incorporates real-time data processing, prediction, and adaptive decision-making to reduce energy usage across a variety of devices, such as smart thermostats, robotic vacuum cleaners, and IoT-enabled lighting systems. Key methodologies include the integration of RNNs to model temporal energy consumption patterns and the Adam optimizer for efficient model training. The results demonstrate significant energy savings with improvements in prediction accuracy compared to traditional approaches. The findings suggest that the AI-based system can provide substantial reductions in energy consumption while maintaining or improving device performance. The research has important implications for the future of energy-efficient consumer electronics, particularly as the demand for smart devices grows.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"71 2","pages":"4714-4720"},"PeriodicalIF":10.9,"publicationDate":"2025-04-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144867642","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：