Edge Intelligence (EI) combines edge computing and artificial intelligence to deliver low-latency and resource-efficient services. Integrated Sensing and Communication (ISAC) further empowers EI by enhancing edge perception and accelerating intelligent model training. However, integrating ISAC into EI complicates the coordination of dynamically varying sensing, communication, and computation resources, especially under device mobility and unpredictable network conditions, leading to degraded service performance. To address these coordination challenges and sustain high-quality service under mobility and dynamics, we aim to design an adaptive service provision framework that tightly couples real-time perception with intelligent decision-making at the edge. Specifically, we propose an adaptive service provision architecture for ISAC-assisted EI, where Digital Twins (DTs) hosted on edge servers represent edge devices and their contexts to enable accurate perception and intelligent decision-making, thereby enhancing the efficiency of ISAC-enabled services. By dynamically migrating DTs across edge servers based on device mobility and resource availability, the system supports continuous decision-making and seamless service delivery. We further integrate convex optimization for efficient multi-resource coordination and a Time-Varying Contextual Bandit (TVCB) algorithm to enable adaptive, context-aware DT migration in dynamic environments. Extensive simulations demonstrate that our approach significantly improves service quality, reliability, and adaptability in ISAC-assisted EI systems, reducing migration oscillations and overhead while achieving lower latency and higher utility compared with representative baselines.
{"title":"AdaDT: Adaptive Service Provision and Digital Twin Migration for ISAC-Assisted Edge Intelligence","authors":"Wenqiang Ma;Yi Yang;Wen Sun;Peng Wang;Lei Liu;Dusit Niyato;Victor C.M. Leung","doi":"10.1109/TMC.2025.3634128","DOIUrl":"https://doi.org/10.1109/TMC.2025.3634128","url":null,"abstract":"Edge Intelligence (EI) combines edge computing and artificial intelligence to deliver low-latency and resource-efficient services. Integrated Sensing and Communication (ISAC) further empowers EI by enhancing edge perception and accelerating intelligent model training. However, integrating ISAC into EI complicates the coordination of dynamically varying sensing, communication, and computation resources, especially under device mobility and unpredictable network conditions, leading to degraded service performance. To address these coordination challenges and sustain high-quality service under mobility and dynamics, we aim to design an adaptive service provision framework that tightly couples real-time perception with intelligent decision-making at the edge. Specifically, we propose an adaptive service provision architecture for ISAC-assisted EI, where Digital Twins (DTs) hosted on edge servers represent edge devices and their contexts to enable accurate perception and intelligent decision-making, thereby enhancing the efficiency of ISAC-enabled services. By dynamically migrating DTs across edge servers based on device mobility and resource availability, the system supports continuous decision-making and seamless service delivery. We further integrate convex optimization for efficient multi-resource coordination and a Time-Varying Contextual Bandit (TVCB) algorithm to enable adaptive, context-aware DT migration in dynamic environments. Extensive simulations demonstrate that our approach significantly improves service quality, reliability, and adaptability in ISAC-assisted EI systems, reducing migration oscillations and overhead while achieving lower latency and higher utility compared with representative baselines.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"25 4","pages":"5775-5791"},"PeriodicalIF":9.2,"publicationDate":"2025-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147362428","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 : 2025-11-17DOI: 10.1109/TMC.2025.3633953
Jiahui Li;Geng Sun;Qingqing Wu;Shuang Liang;Jiacheng Wang;Dusit Niyato;Dong In Kim
The rapid development of the low-altitude economy (LAE) has significantly increased the utilization of autonomous aerial vehicles (AAVs) in various applications, necessitating efficient and secure communication methods among AAV swarms. In this work, we aim to introduce distributed collaborative beamforming (DCB) into AAV swarms and handle the eavesdropper collusion by controlling the corresponding signal distributions. Specifically, we consider a two-way DCB-enabled aerial communication between two AAV swarms and construct these swarms as two AAV virtual antenna arrays. Then, we minimize the two-way known secrecy capacity and maximum sidelobe level to avoid information leakage from the known and unknown eavesdroppers, respectively. Simultaneously, we also minimize the energy consumption of AAVs when constructing virtual antenna arrays. Due to the conflicting relationships between secure performance and energy efficiency, we consider these objectives by formulating a multi-objective optimization problem, which is NP-hard and with a large number of decision variables. Accordingly, we design a novel generative swarm intelligence (GenSI) framework to solve the problem with less overhead, which contains a conditional variational autoencoder (CVAE)-based generative method and a proposed powerful swarm intelligence algorithm. In this framework, CVAE can collect expert solutions obtained by the swarm intelligence algorithm in other environment states to explore characteristics and patterns, thereby directly generating high-quality initial solutions in new environment factors for the swarm intelligence algorithm to search solution space efficiently. Simulation results show that the proposed swarm intelligence algorithm outperforms other state-of-the-art baseline algorithms, and the GenSI can achieve similar optimization results by using far fewer iterations than the ordinary swarm intelligence algorithm. Experimental tests demonstrate that introducing the CVAE mechanism achieves a 58.7% reduction in execution time, which enables the deployment of GenSI even on AAV platforms with limited computing power.
{"title":"Aerial Secure Collaborative Communications Under Eavesdropper Collusion in Low-Altitude Economy: A Generative Swarm Intelligent Approach","authors":"Jiahui Li;Geng Sun;Qingqing Wu;Shuang Liang;Jiacheng Wang;Dusit Niyato;Dong In Kim","doi":"10.1109/TMC.2025.3633953","DOIUrl":"https://doi.org/10.1109/TMC.2025.3633953","url":null,"abstract":"The rapid development of the low-altitude economy (LAE) has significantly increased the utilization of autonomous aerial vehicles (AAVs) in various applications, necessitating efficient and secure communication methods among AAV swarms. In this work, we aim to introduce distributed collaborative beamforming (DCB) into AAV swarms and handle the eavesdropper collusion by controlling the corresponding signal distributions. Specifically, we consider a two-way DCB-enabled aerial communication between two AAV swarms and construct these swarms as two AAV virtual antenna arrays. Then, we minimize the two-way known secrecy capacity and maximum sidelobe level to avoid information leakage from the known and unknown eavesdroppers, respectively. Simultaneously, we also minimize the energy consumption of AAVs when constructing virtual antenna arrays. Due to the conflicting relationships between secure performance and energy efficiency, we consider these objectives by formulating a multi-objective optimization problem, which is NP-hard and with a large number of decision variables. Accordingly, we design a novel generative swarm intelligence (GenSI) framework to solve the problem with less overhead, which contains a conditional variational autoencoder (CVAE)-based generative method and a proposed powerful swarm intelligence algorithm. In this framework, CVAE can collect expert solutions obtained by the swarm intelligence algorithm in other environment states to explore characteristics and patterns, thereby directly generating high-quality initial solutions in new environment factors for the swarm intelligence algorithm to search solution space efficiently. Simulation results show that the proposed swarm intelligence algorithm outperforms other state-of-the-art baseline algorithms, and the GenSI can achieve similar optimization results by using far fewer iterations than the ordinary swarm intelligence algorithm. Experimental tests demonstrate that introducing the CVAE mechanism achieves a 58.7% reduction in execution time, which enables the deployment of GenSI even on AAV platforms with limited computing power.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"25 4","pages":"5758-5774"},"PeriodicalIF":9.2,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147362473","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 : 2025-11-17DOI: 10.1109/TMC.2025.3633263
Jie Su;Yuting Jiang;Yuheng Ye;Zhenyu Wen;Taotao Li;Shibo He;Xiaoqin Zhang;Rajiv Ranjan
The emergence of abundant non-stationary radio signal (NSRS) data presents significant opportunities for applications in wireless communications, radar systems, remote sensing, and healthcare. While deep learning models have shown promise in capturing sequence dependencies, deriving generic and fine-grained representations of NSRS data remains challenging due to its complex, dynamic nature and the scarcity of labeled data. The NSRS data are often frequency-sensitive and exhibit minuscule inter-class distances, posing significant challenges for precise classification. To address these issues, we propose a novel Dual Modality Patch Contrastive (DMPC) framework. This framework leverages a stochastic patching paradigm for diverse local pattern extraction and a time-frequency cross-view optimization for frequency-sensitive feature mining. Furthermore, an Attentive Patch Aggregation (APA) mechanism enhances fine-grained inference under few-shot conditions through patch-level feature voting. Extensive experiments demonstrate the effectiveness of our approach in addressing the unique challenges of NSRS data.
{"title":"Patch Matter: Dual Modality Patch Contrastive for Non-Stationary Radio Signals","authors":"Jie Su;Yuting Jiang;Yuheng Ye;Zhenyu Wen;Taotao Li;Shibo He;Xiaoqin Zhang;Rajiv Ranjan","doi":"10.1109/TMC.2025.3633263","DOIUrl":"https://doi.org/10.1109/TMC.2025.3633263","url":null,"abstract":"The emergence of abundant non-stationary radio signal (NSRS) data presents significant opportunities for applications in wireless communications, radar systems, remote sensing, and healthcare. While deep learning models have shown promise in capturing sequence dependencies, deriving generic and fine-grained representations of NSRS data remains challenging due to its complex, dynamic nature and the scarcity of labeled data. The NSRS data are often frequency-sensitive and exhibit minuscule inter-class distances, posing significant challenges for precise classification. To address these issues, we propose a novel <bold>D</b>ual <bold>M</b>odality <bold>P</b>atch <bold>C</b>ontrastive (<bold>DMPC</b>) framework. This framework leverages a stochastic patching paradigm for diverse local pattern extraction and a time-frequency cross-view optimization for frequency-sensitive feature mining. Furthermore, an Attentive Patch Aggregation (APA) mechanism enhances fine-grained inference under few-shot conditions through patch-level feature voting. Extensive experiments demonstrate the effectiveness of our approach in addressing the unique challenges of NSRS data.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"25 4","pages":"5936-5951"},"PeriodicalIF":9.2,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147362579","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}
Radio-frequency (RF) signal processing has seen significant advancements with the advent of deep learning, providing more accurate and efficient solutions for tasks such as signal classification and generation. However, most existing methods are heavily dependent on large labeled datasets, which are often scarce and costly to obtain in real-world RF environments. Furthermore, these approaches tend to be task-specific, limiting their ability to generalize across various RF applications. To address these challenges, this paper proposes RF-MAE, a self-supervised adaptive frequency masked autoencoder. RF-MAE leverages self-supervised learning (SSL) to capture intrinsic patterns from large-scale unlabeled RF data. Central to RF-MAE is a novel Adaptive Frequency Masked (AFM) strategy, which dynamically masks frequency components based on their energy distribution. Supported by a robust theoretical foundation, AFM ensures the model focuses on the most informative signal components, thereby enhancing generalization across RF tasks. By pretraining on unlabeled data and fine-tuning on specific tasks, RF-MAE significantly reduces the reliance on labeled datasets while improving adaptability across diverse RF signal processing tasks. Experimental results demonstrate that RF-MAE consistently outperforms traditional models, underscoring its potential to generalize across tasks and deliver superior performance in a wide range of RF signal applications.
{"title":"RF-MAE: A Self-Supervised Adaptive Frequency Masked Autoencoder With Radio-Frequency Signal Processing Applications","authors":"Zhongyi Wen;Zhikai Zhai;Yatong Wang;Qiang Li;Wei Zhang;Huaizong Shao","doi":"10.1109/TMC.2025.3633287","DOIUrl":"https://doi.org/10.1109/TMC.2025.3633287","url":null,"abstract":"Radio-frequency (RF) signal processing has seen significant advancements with the advent of deep learning, providing more accurate and efficient solutions for tasks such as signal classification and generation. However, most existing methods are heavily dependent on large labeled datasets, which are often scarce and costly to obtain in real-world RF environments. Furthermore, these approaches tend to be task-specific, limiting their ability to generalize across various RF applications. To address these challenges, this paper proposes RF-MAE, a self-supervised adaptive frequency masked autoencoder. RF-MAE leverages self-supervised learning (SSL) to capture intrinsic patterns from large-scale unlabeled RF data. Central to RF-MAE is a novel Adaptive Frequency Masked (AFM) strategy, which dynamically masks frequency components based on their energy distribution. Supported by a robust theoretical foundation, AFM ensures the model focuses on the most informative signal components, thereby enhancing generalization across RF tasks. By pretraining on unlabeled data and fine-tuning on specific tasks, RF-MAE significantly reduces the reliance on labeled datasets while improving adaptability across diverse RF signal processing tasks. Experimental results demonstrate that RF-MAE consistently outperforms traditional models, underscoring its potential to generalize across tasks and deliver superior performance in a wide range of RF signal applications.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"25 4","pages":"5920-5935"},"PeriodicalIF":9.2,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147362580","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 : 2025-11-17DOI: 10.1109/TMC.2025.3633250
Xiaoyan Liu;Peng Yang;Xingyu Liu;Zhenqi Wang;Zijian Bai
Text representation models, such as Transformers and RNNs, are foundational to natural language processing research and play a crucial role in a wide range of downstream tasks. However, most existing large models rely on the self-attention mechanism, which involves frequent token interactions, a large parameter scale, and substantial hardware and data resource requirements—posing significant challenges for deployment on mobile and edge devices with limited computing power, high communication latency, or intermittent connectivity. To address these practical constraints, this study proposes an ultra-lightweight text representation model based on Causal Recurrent Convolutional Networks (CausalRCN), specifically engineered for efficient inference in resource-constrained mobile environments. Rather than introducing new atomic modules, our innovation lies in the systematic integration of causal convolution and self-enhancement mechanisms into a compact recurrent architecture that eliminates attention-driven computation entirely. By replacing self-attention with parallelized causal convolutions and recurrent feature propagation, the model achieves dramatically lower computational complexity and memory footprint, enabling real-time inference on commodity edge hardware. The design leverages local causality to approximate global contextual dependencies and employs feature self-enhancement to strengthen nonlinear expressiveness, effectively balancing accuracy and efficiency. Experimental results show that with only 666 K parameters, the proposed model achieves over 10× higher throughput and 50% less memory usage compared to competitive baselines, while improving accuracy by more than 1%. Validated on real-world platforms including Raspberry Pi 4B and Qualcomm Snapdragon Gen 1, the system demonstrates up to a 90× speedup over ALBERT in end-to-end latency, underscoring its suitability for mobile intelligent computing.
文本表示模型，如Transformers和rnn，是自然语言处理研究的基础，在广泛的下游任务中起着至关重要的作用。然而，大多数现有的大型模型依赖于自关注机制，这涉及频繁的令牌交互、大的参数规模以及大量的硬件和数据资源需求，这对在计算能力有限、通信延迟高或间歇性连接的移动和边缘设备上的部署构成了重大挑战。为了解决这些实际限制，本研究提出了一种基于因果循环卷积网络（CausalRCN）的超轻量级文本表示模型，该模型专门用于在资源受限的移动环境中进行高效推理。我们的创新不是引入新的原子模块，而是将因果卷积和自我增强机制系统地集成到一个紧凑的循环架构中，从而完全消除了注意力驱动的计算。通过用并行因果卷积和循环特征传播取代自关注，该模型显著降低了计算复杂度和内存占用，实现了在商品边缘硬件上的实时推理。设计利用局部因果关系来近似全局上下文依赖关系，并采用特征自增强来加强非线性表达，有效地平衡了准确性和效率。实验结果表明，与竞争基准相比，仅使用666 K个参数，该模型的吞吐量提高了10倍以上，内存使用减少了50%，准确率提高了1%以上。在包括Raspberry Pi 4B和Qualcomm Snapdragon Gen 1在内的实际平台上进行了验证，该系统在端到端延迟方面的速度比ALBERT提高了90倍，强调了其对移动智能计算的适用性。
{"title":"Toward an Ultra-Lightweight Text Representation: Causal Convolutional Networks With Feature Self-Enhancement Mechanisms","authors":"Xiaoyan Liu;Peng Yang;Xingyu Liu;Zhenqi Wang;Zijian Bai","doi":"10.1109/TMC.2025.3633250","DOIUrl":"https://doi.org/10.1109/TMC.2025.3633250","url":null,"abstract":"Text representation models, such as Transformers and RNNs, are foundational to natural language processing research and play a crucial role in a wide range of downstream tasks. However, most existing large models rely on the self-attention mechanism, which involves frequent token interactions, a large parameter scale, and substantial hardware and data resource requirements—posing significant challenges for deployment on mobile and edge devices with limited computing power, high communication latency, or intermittent connectivity. To address these practical constraints, this study proposes an ultra-lightweight text representation model based on Causal Recurrent Convolutional Networks (CausalRCN), specifically engineered for efficient inference in resource-constrained mobile environments. Rather than introducing new atomic modules, our innovation lies in the systematic integration of causal convolution and self-enhancement mechanisms into a compact recurrent architecture that eliminates attention-driven computation entirely. By replacing self-attention with parallelized causal convolutions and recurrent feature propagation, the model achieves dramatically lower computational complexity and memory footprint, enabling real-time inference on commodity edge hardware. The design leverages local causality to approximate global contextual dependencies and employs feature self-enhancement to strengthen nonlinear expressiveness, effectively balancing accuracy and efficiency. Experimental results show that with only 666 K parameters, the proposed model achieves over 10× higher throughput and 50% less memory usage compared to competitive baselines, while improving accuracy by more than 1%. Validated on real-world platforms including Raspberry Pi 4B and Qualcomm Snapdragon Gen 1, the system demonstrates up to a 90× speedup over ALBERT in end-to-end latency, underscoring its suitability for mobile intelligent computing.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"25 4","pages":"5890-5904"},"PeriodicalIF":9.2,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147362441","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}
Semantic communication emphasizes the transmission of meaning rather than raw symbols. It offers a promising solution to alleviate network congestion and improve transmission efficiency. In this paper, we propose a wireless image communication framework that employs probability graphs as shared semantic knowledge base among distributed users. High-level image semantics are represented via scene graphs, and a two-stage compression algorithm is devised to remove predictable components based on learned conditional and co-occurrence probabilities. At the transmitter, the algorithm filters redundant relations and entity pairs, while at the receiver, semantic recovery leverages the same probability graphs to reconstruct omitted information. For further research, we also put forward a multi-round semantic compression algorithm with its theoretical performance analysis. Simulation results demonstrate that our semantic-aware scheme achieves superior transmission throughput and satiable semantic alignment, validating the efficacy of leveraging high-level semantics for image communication.
{"title":"Scene Graph-Aided Probabilistic Semantic Communication for Image Transmission","authors":"Chen Zhu;Siyun Liang;Zhouxiang Zhao;Jianrong Bao;Zhaohui Yang;Zhaoyang Zhang;Dusit Niyato","doi":"10.1109/TMC.2025.3633495","DOIUrl":"https://doi.org/10.1109/TMC.2025.3633495","url":null,"abstract":"Semantic communication emphasizes the transmission of meaning rather than raw symbols. It offers a promising solution to alleviate network congestion and improve transmission efficiency. In this paper, we propose a wireless image communication framework that employs probability graphs as shared semantic knowledge base among distributed users. High-level image semantics are represented via scene graphs, and a two-stage compression algorithm is devised to remove predictable components based on learned conditional and co-occurrence probabilities. At the transmitter, the algorithm filters redundant relations and entity pairs, while at the receiver, semantic recovery leverages the same probability graphs to reconstruct omitted information. For further research, we also put forward a multi-round semantic compression algorithm with its theoretical performance analysis. Simulation results demonstrate that our semantic-aware scheme achieves superior transmission throughput and satiable semantic alignment, validating the efficacy of leveraging high-level semantics for image communication.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"25 4","pages":"5905-5919"},"PeriodicalIF":9.2,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147362474","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 : 2025-11-17DOI: 10.1109/TMC.2025.3633266
Simin Dai;Xuechen Chen;Xiaoheng Deng;Siyu Lin
Deep learning-based joint source-channel coding (DeepJSCC) has shown significant benefits in emerging semantic and task-oriented communications, providing a promising solution for reducing latency and bandwidth requirements in next-generation mobile networks. However, its deployment on resource-constrained devices is limited by model complexity. Devices with varying computational capacities require models of distinct architectures and complexity levels, motivating the design of a cross-architecture model compression scheme for DeepJSCC. In this paper, we propose a cross-architecture knowledge distillation framework called CAKDJSCC for heterogeneous DeepJSCC models. Specifically, we design a teaching assistant network with feature fusion modules (FFMs) that dynamically perceive architecture gaps between teacher and student models, thereby generating student-adaptive feature representations to alleviate feature space misalignment caused by architectural inconsistencies. In addition, we introduce a conditional information bottleneck (CIB) loss to optimize the distillation process, which prevents students from overfitting to teacher-specific inductive biases while enhancing knowledge transfer efficiency in cross-architecture scenarios. Extensive experiments demonstrate that our approach significantly improves the student model’s reconstruction accuracy and perceptual quality without increasing the inference latency while minimizing the performance degradation during model compression.
{"title":"Cross-Architecture Knowledge Distillation for Deep Joint Source-Channel Coding","authors":"Simin Dai;Xuechen Chen;Xiaoheng Deng;Siyu Lin","doi":"10.1109/TMC.2025.3633266","DOIUrl":"https://doi.org/10.1109/TMC.2025.3633266","url":null,"abstract":"Deep learning-based joint source-channel coding (DeepJSCC) has shown significant benefits in emerging semantic and task-oriented communications, providing a promising solution for reducing latency and bandwidth requirements in next-generation mobile networks. However, its deployment on resource-constrained devices is limited by model complexity. Devices with varying computational capacities require models of distinct architectures and complexity levels, motivating the design of a cross-architecture model compression scheme for DeepJSCC. In this paper, we propose a cross-architecture knowledge distillation framework called CAKDJSCC for heterogeneous DeepJSCC models. Specifically, we design a teaching assistant network with feature fusion modules (FFMs) that dynamically perceive architecture gaps between teacher and student models, thereby generating student-adaptive feature representations to alleviate feature space misalignment caused by architectural inconsistencies. In addition, we introduce a conditional information bottleneck (CIB) loss to optimize the distillation process, which prevents students from overfitting to teacher-specific inductive biases while enhancing knowledge transfer efficiency in cross-architecture scenarios. Extensive experiments demonstrate that our approach significantly improves the student model’s reconstruction accuracy and perceptual quality without increasing the inference latency while minimizing the performance degradation during model compression.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"25 4","pages":"5683-5699"},"PeriodicalIF":9.2,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147362443","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 : 2025-11-14DOI: 10.1109/TMC.2025.3632794
Lei Wang;Jiyuan Liu;Qiang He;Feifei Chen;Xiaoyu Xia
Multi-access Edge Computing (MEC) deploys computation and storage resources at the network edge, enabling devices to process data and requests on nearby edge services. This reduces data transmission latency and network congestion. However, due to edge servers’ volatile running status and limited resources, the reliability of edge services deployed on them fluctuates over time. This may lead to concept drifts in edge services’ real-time reliability streaming data. A severe negative drift may indicate a runtime reliability anomaly in an edge service, which often impacts users’ Quality of Experience (QoE). To ensure edge services’ reliability, this paper proposes CS-Detection, a hierarchical approach for detecting runtime reliability anomalies based on concept drift. CS-Detection employs the compressed sensing technique to sample complex and large-scale reliability streaming data. It employs a new technique that combines Variational AutoEncoder and Energy-Based Generative Adversarial Network (E2BGAN) to estimate the anomaly level of edge services by calculating the reconstruction error and discriminant error of compressed real-time reliability streaming data. To demonstrate the usefulness of CS-Detection in ensuring the QoE of MEC systems, we present CPRest, a coordinated checkpoint-based effective rejuvenation approach for restoring the normal operation of edge services affected by runtime reliability anomalies. CPRest classifies detection results into four levels and adjusts the edge services’ restart trigger time accordingly. Comprehensive experiments conducted on real-world datasets demonstrate the effectiveness and efficiency of CS-Detection compared to state-of-the-art approaches.
{"title":"Hierarchical Runtime Reliability Anomaly Detection for Edge Services Rejuvenation","authors":"Lei Wang;Jiyuan Liu;Qiang He;Feifei Chen;Xiaoyu Xia","doi":"10.1109/TMC.2025.3632794","DOIUrl":"https://doi.org/10.1109/TMC.2025.3632794","url":null,"abstract":"Multi-access Edge Computing (MEC) deploys computation and storage resources at the network edge, enabling devices to process data and requests on nearby edge services. This reduces data transmission latency and network congestion. However, due to edge servers’ volatile running status and limited resources, the reliability of edge services deployed on them fluctuates over time. This may lead to concept drifts in edge services’ real-time reliability streaming data. A severe negative drift may indicate a runtime reliability anomaly in an edge service, which often impacts users’ Quality of Experience (QoE). To ensure edge services’ reliability, this paper proposes CS-Detection, a hierarchical approach for detecting runtime reliability anomalies based on concept drift. CS-Detection employs the compressed sensing technique to sample complex and large-scale reliability streaming data. It employs a new technique that combines Variational Auto<underline>E</u>ncoder and <underline>E</u>nergy-<underline>B</u>ased <underline>G</u>enerative <underline>A</u>dversarial <underline>N</u>etwork (E2BGAN) to estimate the anomaly level of edge services by calculating the reconstruction error and discriminant error of compressed real-time reliability streaming data. To demonstrate the usefulness of CS-Detection in ensuring the QoE of MEC systems, we present CPRest, a coordinated checkpoint-based effective rejuvenation approach for restoring the normal operation of edge services affected by runtime reliability anomalies. CPRest classifies detection results into four levels and adjusts the edge services’ restart trigger time accordingly. Comprehensive experiments conducted on real-world datasets demonstrate the effectiveness and efficiency of CS-Detection compared to state-of-the-art approaches.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"25 4","pages":"5807-5823"},"PeriodicalIF":9.2,"publicationDate":"2025-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147362475","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 rise of pre-trained foundation models (FMs) has popularized the trend of fine-tuning FMs to fit downstream tasks, while Federated Learning (FL) has become the de-facto approach for training distributed data with privacy-preservation. However, fine-tuning FMs in FL faces overwhelming overheads due to its bulky nature. While freezing parameters in FM have the potential to accelerate FL training, existing freezing strategies statically freeze parameters on specified or already converged layers, incur severe accuracy degradation, and resource-inefficiency in heterogeneous environments. In this paper, we propose AFedLF, an adaptive freezing framework for FM in FL, to accelerate its wall-clock time for convergence without losing its final accuracy. However, this poses great challenges, as different freezing strategies lead to different accuracy gains and time overheads, while unfreezing more layers may bring marginal accuracy gains but significant time overheads. To address this challenge, AFedLF mathematically establishes a correlation between the freezing strategy and the accuracy gain and time overhead, and allocates adaptive freezing strategies to clients, based on our insight that unfreezing more layers on devices with strong computation and communication capabilities helps improve resource efficiency. Besides, AFedLF incorporates our well-designed intermediate result caching scheme with constant approximation ratios utilizing the limited storage capacity on mobile devices to cache intermediate results to skip forward propagation, further saving wall-clock time. Finally, we implemented AFedLF using an open-source FL benchmark, and extensive trace-driven experimental results showed that AFedLF accelerates wall-clock time by up to 6.1× compared to state-of-the-art solutions, without sacrificing accuracy.
{"title":"AFedLF: Adaptive Layer Freezing of Foundation Models in Heterogeneous Federated Learning","authors":"Yue Zeng;Jie Zhang;Song Guo;Baoliu Ye;Zhihao Qu;Zicong Hong;Bin Tang;Jinyu Chen;Junlong Zhou;Jiaying Yu","doi":"10.1109/TMC.2025.3632655","DOIUrl":"https://doi.org/10.1109/TMC.2025.3632655","url":null,"abstract":"The rise of pre-trained foundation models (FMs) has popularized the trend of fine-tuning FMs to fit downstream tasks, while Federated Learning (FL) has become the de-facto approach for training distributed data with privacy-preservation. However, fine-tuning FMs in FL faces overwhelming overheads due to its bulky nature. While freezing parameters in FM have the potential to accelerate FL training, existing freezing strategies statically freeze parameters on specified or already converged layers, incur severe accuracy degradation, and resource-inefficiency in heterogeneous environments. In this paper, we propose AFedLF, an adaptive freezing framework for FM in FL, to accelerate its wall-clock time for convergence without losing its final accuracy. However, this poses great challenges, as different freezing strategies lead to different accuracy gains and time overheads, while unfreezing more layers may bring marginal accuracy gains but significant time overheads. To address this challenge, AFedLF mathematically establishes a correlation between the freezing strategy and the accuracy gain and time overhead, and allocates adaptive freezing strategies to clients, based on our insight that unfreezing more layers on devices with strong computation and communication capabilities helps improve resource efficiency. Besides, AFedLF incorporates our well-designed intermediate result caching scheme with constant approximation ratios utilizing the limited storage capacity on mobile devices to cache intermediate results to skip forward propagation, further saving wall-clock time. Finally, we implemented AFedLF using an open-source FL benchmark, and extensive trace-driven experimental results showed that AFedLF accelerates wall-clock time by up to 6.1× compared to state-of-the-art solutions, without sacrificing accuracy.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"25 4","pages":"5666-5682"},"PeriodicalIF":9.2,"publicationDate":"2025-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147362566","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 : 2025-11-14DOI: 10.1109/TMC.2025.3632752
Yeqin Li;David Chieng;Boon Giin Lee;Chiew Foong Kwong;Kian Ming Lim;Shuyu Li
The learning-based, passive, device-free wireless human activity recognition (WHAR) systems still face significant challenges, especially in real-world deployments. Environmental differences and domain diversities cause signals collected in the source domain to have a different distribution from those in the target domain, and this affects the accuracy. To achieve domain generalization (DG), a multi-scale linear attention vision transformer (ViT) based feature extractor and domain adversarial learning with Wasserstein distance are proposed. By aligning both marginal and conditional distributions across different source domains, the adversarial learning reduces the differences between trained and unseen domains. As a result, the extracted features become domain-invariant in the latent space, ensuring accuracy is preserved in new or unseen domains. Extensive evaluations using commercial IEEE 802.11ac routers with human activity data collected over different days, environments, human subjects, and obstacle configurations show that the proposed Wi-ViTAL achieves 97.57% average accuracy for five-label classification and more than 76% for eight-label classification in unseen domains. Wi-ViTAL also demonstrates an overall DG improvement compared to other recent benchmarks.
{"title":"Wi-ViTAL: Domain Generalization of Wireless Human Activity Recognition Using Linear Attention Vision Transformer With Adversarial Learning","authors":"Yeqin Li;David Chieng;Boon Giin Lee;Chiew Foong Kwong;Kian Ming Lim;Shuyu Li","doi":"10.1109/TMC.2025.3632752","DOIUrl":"https://doi.org/10.1109/TMC.2025.3632752","url":null,"abstract":"The learning-based, passive, device-free wireless human activity recognition (WHAR) systems still face significant challenges, especially in real-world deployments. Environmental differences and domain diversities cause signals collected in the source domain to have a different distribution from those in the target domain, and this affects the accuracy. To achieve domain generalization (DG), a multi-scale linear attention vision transformer (ViT) based feature extractor and domain adversarial learning with Wasserstein distance are proposed. By aligning both marginal and conditional distributions across different source domains, the adversarial learning reduces the differences between trained and unseen domains. As a result, the extracted features become domain-invariant in the latent space, ensuring accuracy is preserved in new or unseen domains. Extensive evaluations using commercial IEEE 802.11ac routers with human activity data collected over different days, environments, human subjects, and obstacle configurations show that the proposed Wi-ViTAL achieves 97.57% average accuracy for five-label classification and more than 76% for eight-label classification in unseen domains. Wi-ViTAL also demonstrates an overall DG improvement compared to other recent benchmarks.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"25 4","pages":"5792-5806"},"PeriodicalIF":9.2,"publicationDate":"2025-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147362429","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：