IEEE Transactions on Consumer Electronics最新文献

{"title":"A Wrist-Worn Multimodal System for Seamless Pedestrian Navigation and Continuous Health Monitoring: Sensor Fusion in the Wild","authors":"Jiale Wang;Kunlei Li;Ming Xia;Chuang Shi","doi":"10.1109/TCE.2026.3677281","DOIUrl":"https://doi.org/10.1109/TCE.2026.3677281","url":null,"abstract":"Wearable consumer devices have evolved into essential gateways for everyday health insights and location-based services. However, achieving clinical-grade physiological monitoring and seamless positioning on wrist-worn platforms remains challenging due to motion artifacts and signal degradation in complex environments. This paper proposes a robust, lightweight multimodal system that integrates biosensors (photoplethysmography, PPG; electrocardiography, ECG) and motion sensors (inertial measurement unit, IMU) to achieve continuous health monitoring and seamless indoor-outdoor navigation. To mitigate the impact of user movement on consumer-grade sensors, we introduce a subspace projection-based artifact removal method using Hankel-SVD, significantly enhancing heart rate estimation robustness. Furthermore, we propose a factor graph-based framework that tightly fuses global navigation satellite system (GNSS) and Pedestrian Dead Reckoning (PDR) to ensure accurate trajectory tracking across environment transitions. Experimental results in real-world walking scenarios demonstrate that the proposed system maintains an average heart rate estimation error within 2.3 bpm (with >93% of samples <5 bpm) and blood pressure estimation error within 3–6 mmHg. In complex urban settings, the integrated localization method reduces average positioning errors by 34.1% and 64.3% compared to standalone GNSS and PDR, respectively. Characterized by low computational cost and high stability, this solution provides a viable paradigm for next-generation intelligent health management and emergency response in consumer wearables.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"72 3","pages":"7502-7514"},"PeriodicalIF":9.9,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837208","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}

{"title":"LLM-Enhanced Explanatory Signals for Improved API Recommendation in Consumer Electronics Software Development","authors":"Huaizhen Kou;Zhenhua Chen;Weiyi Zhong;Jian Xu;Qun Jin;Xiaokang Zhou","doi":"10.1109/TCE.2026.3663581","DOIUrl":"https://doi.org/10.1109/TCE.2026.3663581","url":null,"abstract":"The development of modern consumer electronics (CE), from smart home ecosystems to wearable devices, hinges on integrating a diverse array of APIs from heterogeneous hardware and cloud services. Existing recommendation methods can generate candidate API lists but lack explanatory power. As a result, developers must interpret the semantic role of each API and its relevance to high-level requirements. This manual process creates a bottleneck and delays feature deployment. To address this limitation, we propose a novel Explanatory Signal-driven API Recommendation framework for CE software development, named ESAR. ESAR leverages Large Language Models (LLMs) to generate recommendation explanations, which are clear and structured rationales that explicitly describe how each recommended API contributes to the developer’s goal. Specifically, ESAR first adopts an LLM-based data augmentation approach to synthesize explanatory signals that explicitly encode the semantic relevance between APIs and development requirements. It then incorporates these signals into the recommendation model through a three-stage progressive training strategy, enabling the model to provide not only accurate API recommendations but also actionable explanations that facilitate rapid validation and enhance developer trust. Experiments conducted on a large-scale dataset from ProgrammableWeb, whose complex API integration patterns are highly representative of the challenges in modern cloud-connected CE software, show that ESAR achieves significant improvements on NDCG metrics, while maintaining comparable performance to strong baselines on MAP, and generates explanations with greater textual diversity and higher semantic consistency with development requirements.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"72 3","pages":"7014-7025"},"PeriodicalIF":9.9,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837211","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}

{"title":"TrustHeart: Explainable and Privacy-Preserving AI Architecture for Trustworthy Cardiac Monitoring in IoMT Consumer Applications","authors":"Qingyang Liu;Jingwen Tang;Jingsong Shen;Yanhui Cai;Yanhuan Huang;Nanyu Lu","doi":"10.1109/TCE.2026.3695954","DOIUrl":"https://doi.org/10.1109/TCE.2026.3695954","url":null,"abstract":"Consumer wearable devices equipped with AI-powered cardiac monitoring systems face significant challenges including adversarial attacks, privacy vulnerabilities, and lack of interpretability, which undermine clinical trust. Existing privacy-preserving cardiac AI systems struggle to maintain accuracy above 92% under strict privacy guarantees (<inline-formula> <tex-math>$varepsilon lt 5$ </tex-math></inline-formula>), fail to provide interpretability while preserving privacy, and lack robustness for safety-critical deployments. This paper presents TrustHeart, a lightweight architecture integrating robust privacy, interpretability, and proactive threat detection for consumer wearables. The major contributions include: (i) an adaptive privacy budget allocation mechanism achieving <inline-formula> <tex-math>$varepsilon = 4.2$ </tex-math></inline-formula> with only 2.8% accuracy degradation through gradient-aware noise calibration; (ii) a privacy-aware attention mechanism achieving 0.76 Intersection over Union(IoU) alignment with clinical annotations under differential privacy guarantees; and (iii) lightweight anomaly detection algorithms detecting > 93.2% of adversarial attacks with 0.9ms overhead. Experiments on MIT-BIH and CPSC2018 datasets demonstrate that TrustHeart achieves 97.6% accuracy on MIT-BIH and 86.1% on CPSC2018, F1=90.8% under Projected Gradient Descent(PGD) attacks, with 35ms inference latency on ARM Cortex-M4, 295kB model size, and < 2.3mW inference power consumption. TrustHeart provides a comprehensive reference framework for trustworthy cardiac AI in consumer wearables.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"72 3","pages":"7121-7137"},"PeriodicalIF":9.9,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837223","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}

{"title":"Recurrent Prompt Learning for Spatio-Temporal Forecasting","authors":"Changlu Chen;Yanbin Liu;Chaoxi Niu;Kaize Shi;Ling Chen;Tianqing Zhu","doi":"10.1109/TCE.2026.3675217","DOIUrl":"https://doi.org/10.1109/TCE.2026.3675217","url":null,"abstract":"Spatio-temporal forecasting holds great significance for various applications in the intelligent transportation system. Foundation models are revolutionizing spatio-temporal forecasting models due to their one-fits-all generalization capabilities. To reprogram the foundation models for the targeted downstream tasks, prompt learning has emerged as an effective approach by optimizing a small set of learnable input tokens while keeping the backbone intact. However, current prompt learning in the spatio-temporal domain typically suffers from two critical limitations: 1) time-agnostic, which cannot capture the temporal evolution during the prompt learning to deal with temporally dependent or sensitive scenarios. 2) input-agnostic, which remain static upon the end of the training, thus failing to fit different distributions at inference. To address these challenges, we propose a recurrent prompt learning framework named RePro to repurpose foundation models to downstream ST forecasting tasks. For the first challenge, we design a recurrent prompt network that is dynamically conditioned on the time-evolving prompts and recurrently optimized based on the historical context. This design injects time-awareness into the prompt to achieve progressive recalibration of the intermediate representations in the foundation model under varying temporal contexts. For the second challenge, we incorporate input data of the current time step into the update of each recurrent prompt state, leading to the input-conditioned prompt learning. This design effectively encapsulates distributional shifts into the prompt dynamics, improving generalization and robustness. Furthermore, two complementary modules are introduced to facilitate the effective application of the recurrent prompt to the foundation model, i.e., cross-prompt aggregation and layer-conditioned prompt adaptation. Specifically, the first module aims to unify the prompt representation and reduce the redundancy, while the second module distributes the recurrent prompts into diverse layers of the foundation model for hierarchical prompting. Extensive experiments on multiple spatio-temporal forecasting benchmarks demonstrate that RePro consistently outperforms strong state-of-the-art baselines across MAE, RMSE, and MAPE, achieving up to 8.3% reduction in MAE, with ablation studies validating the contribution of each proposed component.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"72 3","pages":"6488-6499"},"PeriodicalIF":9.9,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837166","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}

{"title":"Custom Design of Full Pipeline SHA-256 Algorithm Based on Dynamic Pulsed Latch","authors":"Zhen Zhang;Min Song;Chuanhang Shao;Ming Zhang;Yin Zhang;Zhangqing He;Meilin Wan","doi":"10.1109/TCE.2026.3659576","DOIUrl":"https://doi.org/10.1109/TCE.2026.3659576","url":null,"abstract":"In the full pipeline implementation of SHA-256 algorithm, registers occupy a significant portion of the area. To reduce the cost of registers, we introduce the simplest dynamic pulsed latch (DPL) composed of only one inverter and one transmission gate as the data register. As the DPL exhibits inverted storage characteristic, we also modify all the suboperators in the SHA-256 algorithm, including CSA, MAJ, CH, <inline-formula> <tex-math>$sum _{0}$ </tex-math></inline-formula>, <inline-formula> <tex-math>$sum _{1}, sigma _{0}$ </tex-math></inline-formula>, and <inline-formula> <tex-math>$sigma _{1}$ </tex-math></inline-formula>, enabling them to accommodate input data with different polarities. Using a customized approach, we have designed and implemented a 130 -stage full pipeline circuit, incorporating two SHA-256 algorithms, based on a 14 nm FinFET technology. The entire core occupies an area of approximately <inline-formula> <tex-math>$0.1~mathrm {mm}^{2}$ </tex-math></inline-formula>, resulting in a 60% area reduction compared to the traditional full pipeline circuit based on dynamic D flip-flops. Test results indicate that under <inline-formula> <tex-math>$27{}^{circ } mathrm {C}$ </tex-math></inline-formula> and 0.8 V power supply conditions, the average power-delay product of the entire 130stage full pipeline SHA-256 circuit is approximately <inline-formula> <tex-math>$345 mathrm {~mW}^{*} mathrm {~ns}$ </tex-math></inline-formula>.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"72 3","pages":"5339-5355"},"PeriodicalIF":9.9,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837269","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}

{"title":"Generative AI for Cross-Technology Communication in Consumer Electronics","authors":"Demin Gao;Xin Li;Weizheng Wang;Zhaoyang Han;Thippa Reddy Gadekallu","doi":"10.1109/TCE.2026.3684505","DOIUrl":"https://doi.org/10.1109/TCE.2026.3684505","url":null,"abstract":"Cross-Technology Communication (CTC) enables interoperability between wireless technologies such as WiFi, BLE, LoRa, and ZigBee in Consumer Electronics (CE). However, existing methods face limitations in scalability and adaptability, particularly under real-world impairments like Carrier Frequency Offset (CFO) and Adjacent-Channel Interference (ACI). To address these challenges, we propose <bold>GAI-CTC</b>, a Generative AI-driven framework for CTC that optimizes waveform synthesis, spectrum adaptation, and semantic alignment in a unified, end-to-end neural architecture. GAI-CTC utilizes a multi-domain neural network that generates waveforms for multiple protocols while complying with spectral-mask, power, and latency constraints. The framework leverages a cyclic generative learning mechanism and semantic consistency loss to ensure efficient transmitter-receiver mapping. The model is trained through joint optimization of key loss components, including <inline-formula> <tex-math>${mathcal {L}}_{text {CTC}}$ </tex-math></inline-formula>, <inline-formula> <tex-math>${mathcal {L}}_{text {mask}}$ </tex-math></inline-formula>, and <inline-formula> <tex-math>${mathcal {L}}_{text {sem}}$ </tex-math></inline-formula>. Simulation results show that GAI-CTC achieves up to <inline-formula> <tex-math>$3times $ </tex-math></inline-formula> higher goodput and 20-40% lower energy-per-bit compared to Baseline CTC. Additionally, GAI-CTC improves spectral mask compliance by 30-50 percentage points and enhances CFO tolerance by 20 ppm, demonstrating its potential to enable efficient and adaptive communication in multi-protocol wireless systems.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"72 3","pages":"7026-7038"},"PeriodicalIF":9.9,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837295","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}

{"title":"An Interpretable Robust Cyber Defense Framework for Consumer PPG-Based Healthcare Monitoring","authors":"Faisal Saeed;Sanjar Karshiev;Abdul Rehman;Inam Ullah;Yonis Gulzar;Mahran Al-Zyoud","doi":"10.1109/TCE.2026.3689203","DOIUrl":"https://doi.org/10.1109/TCE.2026.3689203","url":null,"abstract":"Consumer health wearables and IoT-enabled devices increasingly rely on Photoplethysmography (PPG) signals for real-time health monitoring and connected healthcare services. However, AI models deployed on these consumer electronics are vulnerable to cyber-physical adversarial manipulation, undermining reliability, transparency, and clinical trust. Subtle digital perturbations or realistic physical distortions, such as motion-induced drift or ambient light interference, can mislead downstream models and result in unsafe or non-interpretable health assessments. To address this, we propose HP-PPG, a cyber–physical robustness framework that formalizes and trains against sequential hybrid perturbations via a worst-case objective over both digital and physical acquisition distortions. We instantiate the inner maximization using projected gradient-based perturbations for digital manipulation and an efficient query-based search over physically plausible distortion parameters under wearable constraints. To prevent clean-signal degradation caused by a mismatch in feature statistics between clean and attacked PPG distributions, we introduce Mixed Batch Normalization (MixBN), which maintains separate normalization statistics for clean and hybrid-attacked features. All experiments are conducted exclusively on the WESAD dataset for heart-rate estimation and heart-rate zone classification tasks. The results demonstrate that HP-PPG significantly enhances robustness, maintaining <inline-formula> <tex-math>$mathrm {ROI}_{mathrm {reg}} gt 0.70$ </tex-math></inline-formula> under strong hybrid attacks where standard models fall below 0.30, while preserving high accuracy on clean data (3.90 bpm MAE compared to 3.98 bpm). These results demonstrate the potential of explainability-aware defense mechanisms to enhance the trustworthiness of AI-driven consumer electronics in connected healthcare systems.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"72 3","pages":"6976-6987"},"PeriodicalIF":9.9,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837292","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}

{"title":"CardioGPT-ViT: Explainable Deep Learning Models With Multimodal Data for Coronary Artery Disease and Heart Failure on Edge Consumer Devices","authors":"Zhiwen Zhang;Vijay Govindarajan;Xin Ouyang;Jing Yang;Bayan Alabdullah;Saad Alammar;Lip Yee Por;Quan Guo;Muwei Li","doi":"10.1109/TCE.2026.3667507","DOIUrl":"https://doi.org/10.1109/TCE.2026.3667507","url":null,"abstract":"Cardiovascular diseases (CVDs) are the most common causes of death in the world, especially coronary artery disease (CAD) and heart failure (HF). Prevention and sustained monitoring in the early stages are essential, yet conventional diagnostic methods are expensive, invasive, and limited to the clinical environment. Recent developments in consumer electronics (CE), including smartwatches and smartphones, are providing new opportunities for real-time, non-invasive cardiovascular monitoring. Nevertheless, the implementation of deep learning (DL) on such devices poses challenges for explainability, computational efficiency, and privacy. This paper introduces CardioGPT-ViT, a novel multimodal, explainable DL model for detecting CAD and HF on edge devices. The model combines Vision Transformers (ViT), GPT-style attention, Mamba state-space modeling, and a small You Only Look Once (YOLO) detection head and incorporates the explainability functionality of clinical trust. CardioGPT-ViT was trained and tested on three benchmark datasets: PulseDB, UCI Heart Disease, and an external electrocardiography (ECG)/photoplethysmography (PPG) cohort. The framework includes SHapley Additive exPlanations (SHAP), Local explainable Model-Agnostic Explanation (LIME), attention maps, and Gradient-weighted Class Activation Mapping (Grad-CAM) to explain results. Quantization, pruning, and latency/energy profiling were used to verify the feasibility of edge deployment on consumer devices. It was compared to performance with recent baseline models across convolutional neural networks (CNNs), ViTs, and state-space architectures. CardioGPT-ViT also scored over 0.90 across all datasets, reaching a maximum of 0.962 on PulseDB. The model was found to be more accurate (92.8%) and robust and had real-time feasibility (latency of less than 50 ms) and low energy consumption (less than 80 mJ per inference). Explainability results correlated with clinical biomarkers, increasing trust. CardioGPT-ViT provides a clinically explainable, resource-efficient, and edge-deployable solution to next-generation cardiovascular care, connecting the gap between consumer electronics and long-term preventive healthcare.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"72 3","pages":"6951-6963"},"PeriodicalIF":9.9,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837299","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}

{"title":"Prescribed Performance Fault-Tolerant Trajectory Tracking Control for Gripper-Based Underwater Salvage Robot Based on a Sliding Mode Disturbance Observer","authors":"YuAn Jiang;Qiang Zhang;WeiDong Zhang;ZeWen Tao","doi":"10.1109/TCE.2026.3677247","DOIUrl":"https://doi.org/10.1109/TCE.2026.3677247","url":null,"abstract":"This research investigates actuator failures in gripper-based underwater salvage robot (GSR) and proposes a prescribed-performance fault-tolerant sliding mode control scheme that ensures tracking accuracy. First, a fault-tolerant dynamic model of the GSR is developed to characterize multiple fault states, including partial effectiveness loss and complete failure, thereby enhancing system robustness. Second, a prescribed-performance control (PPC) strategy is designed to strictly confine trajectory errors within predefined bounds, even under fault conditions, thereby improving operational reliability. Then, a sliding mode observer is developed to estimate and compensate for external disturbances. Meanwhile, projection analysis techniques are employed to exploit decoupling properties, enabling the adaptive separation of strongly coupled fault characteristic states, thereby achieving accurate fault compensation and rational power allocation in the control system. Furthermore, event-triggered control is adopted to mitigate structural wear caused by frequent variations in thruster output during practical operations, thereby protecting the thrusters. Finally, simulations and comparative studies are carried out using a physical GSR model. The results demonstrate that, under thruster failure conditions, the proposed scheme limits the maximum trajectory error to 0.9 meters, achieves a 40% improvement in convergence speed, reduces thrust output variations to less than 50%, and decreases high-frequency similar thrust signals by 70%. Moreover, it provides more reasonable transient speed variations compared to other approaches, thereby validating its superior effectiveness in underwater salvage operations.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"72 3","pages":"5617-5627"},"PeriodicalIF":9.9,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837374","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}

{"title":"Proof-Based Secure Semantic Communication for Privacy-Critical Consumer Internet of Things","authors":"Renwen Ma;Zhaoqian Liu;Chenxin Yuan;Yiming Miao;Marios Kountouris;Pin-Han Ho","doi":"10.1109/TCE.2026.3687418","DOIUrl":"https://doi.org/10.1109/TCE.2026.3687418","url":null,"abstract":"Semantic communication (SemCom) has emerged as a promising paradigm for consumer IoT (CIoT) systems by transmitting task-relevant semantic representations instead of raw data, significantly reducing communication overhead. However, this paradigm shift introduces a new security challenge: adversaries may tamper with semantic computation to induce incorrect task decisions even when the channel remains bitwise correct. Conventional bit-level integrity mechanisms provide no guarantees of semantic execution correctness. While full zero-knowledge proofs (ZKPs) could enforce such guarantees, their overhead is impractical for latency-sensitive CIoT deployments. Accordingly, in this paper, we propose a hybrid, proof-based secure SemCom framework for privacy-critical CIoT, where a ZKP attests to correct semantic encoding without exposing raw inputs, and an optimistic fraud-proof mechanism verifies decoding and task inference with low overhead in the common case. To prevent semantic privacy leakage during disputes, the protocol binds intermediate states with lightweight commitments and resolves challenges via privacy-preserving re-execution with secure multiparty computation. Experiments on consumer-oriented semantic workloads, including image classification, object detection, semantic autoencoding, and language modeling, show that the proposed design reduces end-to-end verification overhead by up to three orders of magnitude compared with full ZKP baselines, while maintaining millisecond-level verification latency and negligible semantic utility loss. This work provides a principled, deployable approach for trustworthy and secure semantic communication in privacy-critical consumer IoT systems.","PeriodicalId":13208,"journal":{"name":"IEEE Transactions on Consumer Electronics","volume":"72 3","pages":"7185-7195"},"PeriodicalIF":9.9,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148836452","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}