Pub Date : 2026-01-30DOI: 10.1109/LCOMM.2026.3659461
Wufeng Kong;Lihua Li;Chenhui Li;Xingwang Li
In this letter, we propose the spectral efficiency standard deviation (SESD) as a new metric to evaluate the consistency of user communication experience in scalable cell-free massive multiple input multiple output (CF mMIMO) systems degraded by channel aging. Closed-form expressions of the downlink spectral efficiency (SE) and SESD are derived under channel aging conditions. To minimize SESD while satisfying a given minimum SE requirement associated with different services, we design a joint access point (AP) selection and power allocation scheme. This SESD minimization approach achieves approximately 6% reduction in SESD compared to the conventional sum-rate maximization and fixed allocation methods, effectively mitigating communication fluctuations. Furthermore, by mapping SE stability to the speech Mean Opinion Score (MOS), we show that SESD minimization also effectively improves the consistency of user-perceived quality of experience (QoE).
{"title":"Mitigating User Communication Inconsistency Caused by Channel Aging in Scalable Cell-Free Massive MIMO Systems","authors":"Wufeng Kong;Lihua Li;Chenhui Li;Xingwang Li","doi":"10.1109/LCOMM.2026.3659461","DOIUrl":"https://doi.org/10.1109/LCOMM.2026.3659461","url":null,"abstract":"In this letter, we propose the spectral efficiency standard deviation (SESD) as a new metric to evaluate the consistency of user communication experience in scalable cell-free massive multiple input multiple output (CF mMIMO) systems degraded by channel aging. Closed-form expressions of the downlink spectral efficiency (SE) and SESD are derived under channel aging conditions. To minimize SESD while satisfying a given minimum SE requirement associated with different services, we design a joint access point (AP) selection and power allocation scheme. This SESD minimization approach achieves approximately 6% reduction in SESD compared to the conventional sum-rate maximization and fixed allocation methods, effectively mitigating communication fluctuations. Furthermore, by mapping SE stability to the speech Mean Opinion Score (MOS), we show that SESD minimization also effectively improves the consistency of user-perceived quality of experience (QoE).","PeriodicalId":13197,"journal":{"name":"IEEE Communications Letters","volume":"30 ","pages":"1086-1090"},"PeriodicalIF":4.4,"publicationDate":"2026-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146223566","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-30DOI: 10.1109/LCOMM.2026.3659522
Suting Liu;Zhiyuan Lin;Jin Jin;Ruiqing Wen;Linling Kuang;Guowei Yang
Electromagnetic map reconstruction is a critical technology for communication satellites to be aware of the complete electromagnetic situation, serving as a fundamental basis for efficient spectrum sharing in satellite communications. The satellite electromagnetic environment exhibits significant short-term temporal correlation, with sampling points dynamically varying over time. Leveraging these characteristics, we propose a method for sparse reconstruction of satellite electromagnetic maps using past sensing information, and design a neural network, Temporal Attention UNet (TA-UNet). By incorporating a temporal attention mechanism, TA-UNet effectively integrates sensing information from multiple time slots, facilitating more efficient electromagnetic map reconstruction. Experiments demonstrate that past sensing information significantly improves the similarity between the reconstructed map and the ground truth. Moreover, TA-UNet demonstrates superior reconstruction accuracy and real-time performance, outperforming previous methods. Compared with UNet without past sensing information, the reconstruction error is reduced by 12.67% to 62.70%.
{"title":"TA-UNet: A Satellite Electromagnetic Map Sparse Reconstruction Method With Past Sensing Information","authors":"Suting Liu;Zhiyuan Lin;Jin Jin;Ruiqing Wen;Linling Kuang;Guowei Yang","doi":"10.1109/LCOMM.2026.3659522","DOIUrl":"https://doi.org/10.1109/LCOMM.2026.3659522","url":null,"abstract":"Electromagnetic map reconstruction is a critical technology for communication satellites to be aware of the complete electromagnetic situation, serving as a fundamental basis for efficient spectrum sharing in satellite communications. The satellite electromagnetic environment exhibits significant short-term temporal correlation, with sampling points dynamically varying over time. Leveraging these characteristics, we propose a method for sparse reconstruction of satellite electromagnetic maps using past sensing information, and design a neural network, Temporal Attention UNet (TA-UNet). By incorporating a temporal attention mechanism, TA-UNet effectively integrates sensing information from multiple time slots, facilitating more efficient electromagnetic map reconstruction. Experiments demonstrate that past sensing information significantly improves the similarity between the reconstructed map and the ground truth. Moreover, TA-UNet demonstrates superior reconstruction accuracy and real-time performance, outperforming previous methods. Compared with UNet without past sensing information, the reconstruction error is reduced by 12.67% to 62.70%.","PeriodicalId":13197,"journal":{"name":"IEEE Communications Letters","volume":"30 ","pages":"1021-1025"},"PeriodicalIF":4.4,"publicationDate":"2026-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146223675","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This letter investigates a dual-hop relaying system enhanced by reconfigurable intelligent surfaces (RISs) deployed at the transmitter side (RIS-T) and the relay–destination link (RIS-DH). Both amplify-and-forward (AF) and decode-and-forward (DF) relaying protocols are considered. An active RIS-assisted RIS-DH configuration is additionally studied to evaluate the effect of signal amplification. Closed-form expressions for the outage probability (OP) and average bit error probability (ABEP) are derived, along with high-SNR asymptotic analysis. The results reveal the impact of RIS placement, relaying strategy, and active RIS operation. Numerical results, supported by Monte Carlo simulations, validate the analysis and show that the joint use of RIS-T and RIS-DH provides notable performance gains over conventional dual-hop relaying. The findings offer useful insights for the design of RIS-assisted dual-hop architectures in future wireless networks.
{"title":"RIS-T and RIS-DH Enhanced Dual-Hop Relaying","authors":"Namburi Venkatalakshmi;Bijaylaxmi Das;Hemanta Kumar Sahu","doi":"10.1109/LCOMM.2026.3659715","DOIUrl":"https://doi.org/10.1109/LCOMM.2026.3659715","url":null,"abstract":"This letter investigates a dual-hop relaying system enhanced by reconfigurable intelligent surfaces (RISs) deployed at the transmitter side (RIS-T) and the relay–destination link (RIS-DH). Both amplify-and-forward (AF) and decode-and-forward (DF) relaying protocols are considered. An active RIS-assisted RIS-DH configuration is additionally studied to evaluate the effect of signal amplification. Closed-form expressions for the outage probability (OP) and average bit error probability (ABEP) are derived, along with high-SNR asymptotic analysis. The results reveal the impact of RIS placement, relaying strategy, and active RIS operation. Numerical results, supported by Monte Carlo simulations, validate the analysis and show that the joint use of RIS-T and RIS-DH provides notable performance gains over conventional dual-hop relaying. The findings offer useful insights for the design of RIS-assisted dual-hop architectures in future wireless networks.","PeriodicalId":13197,"journal":{"name":"IEEE Communications Letters","volume":"30 ","pages":"1012-1016"},"PeriodicalIF":4.4,"publicationDate":"2026-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175857","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-29DOI: 10.1109/LCOMM.2026.3659347
Mehrab Rajabi;Mohammad-Reza Sadeghi
Protograph-based Raptor-like (PBRL) LDPC codes, adopted in the 5G NR eMBB data channel, support a wide range of code rates by generating incremental redundancy through XOR operations. As the protograph structure directly influences performance, its efficient design is crucial. This letter introduces a novel framework for optimizing the IRC submatrix by minimizing the Euclidean norm of capacity gaps, a new evaluation metric that captures overall threshold behavior across all supported rates, in contrast to the conventional rate-by-rate greedy optimization. The approach formulates this as a global optimization problem solved via a genetic algorithm (GA) with simulated annealing (SA) applied for improved convergence. Numerical results show notable threshold reductions, including a 58.4% improvement in the capacity gap norm for the 5G Base Graph 1. The method is also compatible with other rate-compatible structures, such as rate-compatible spatially coupled LDPC codes, which are promising candidates for coding schemes in next-generation mobile systems.
5G NR eMBB数据通道采用基于原型的类猛禽（PBRL） LDPC码，通过异或操作产生增量冗余，支持大范围的码率。由于原型机的结构直接影响其性能，因此其高效设计至关重要。这封信介绍了一个新的框架，通过最小化容量缺口的欧几里得范数来优化IRC子矩阵，这是一个新的评估指标，可以捕获所有支持速率的总体阈值行为，与传统的逐速率贪婪优化相反。该方法将其表述为通过遗传算法（GA）和模拟退火（SA）来解决全局优化问题，以提高收敛性。数值结果显示了显著的阈值降低，包括5G基本图1的容量缺口规范提高了58.4%。该方法还与其他速率兼容的结构兼容，例如速率兼容的空间耦合LDPC码，这是下一代移动系统中编码方案的有希望的候选者。
{"title":"Simultaneous Multi-Rate Threshold Optimization of Protograph-Based Raptor-Like LDPC Codes via Genetic Algorithm","authors":"Mehrab Rajabi;Mohammad-Reza Sadeghi","doi":"10.1109/LCOMM.2026.3659347","DOIUrl":"https://doi.org/10.1109/LCOMM.2026.3659347","url":null,"abstract":"Protograph-based Raptor-like (PBRL) LDPC codes, adopted in the 5G NR eMBB data channel, support a wide range of code rates by generating incremental redundancy through XOR operations. As the protograph structure directly influences performance, its efficient design is crucial. This letter introduces a novel framework for optimizing the IRC submatrix by minimizing the Euclidean norm of capacity gaps, a new evaluation metric that captures overall threshold behavior across all supported rates, in contrast to the conventional rate-by-rate greedy optimization. The approach formulates this as a global optimization problem solved via a genetic algorithm (GA) with simulated annealing (SA) applied for improved convergence. Numerical results show notable threshold reductions, including a 58.4% improvement in the capacity gap norm for the 5G Base Graph 1. The method is also compatible with other rate-compatible structures, such as rate-compatible spatially coupled LDPC codes, which are promising candidates for coding schemes in next-generation mobile systems.","PeriodicalId":13197,"journal":{"name":"IEEE Communications Letters","volume":"30 ","pages":"1017-1020"},"PeriodicalIF":4.4,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175672","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-28DOI: 10.1109/LCOMM.2026.3658407
Mayank Gupta;Pratik Chakraborty;Shankar Prakriya
This letter proposes a novel uplink multiple access scheme that accommodates up to $K$ low data-rate non-priority users (NPUs) within a selected priority user’s (PU) spectrum, without compromising the PU performance. The uplink base station decodes the PU symbol first, followed by the symbols of $K$ NPUs, employing successive interference cancellation. A dynamic interference threshold limit, set after PU selection, determines $K$ and NPU powers in closed-form using a novel instantaneous power allocation scheme (IPAS) inspired by underlay cognitive radio. Spectrum utilization is maximized with precise control over NPU target rate and $K$ . Closed-form throughput analysis, validated via simulations, demonstrates superiority of the suggested scheme over a traditional multiple access employing static optimal power allocation.
{"title":"Optimized Uplink Spectrum Access With Adaptive Interference-Aware User Admission","authors":"Mayank Gupta;Pratik Chakraborty;Shankar Prakriya","doi":"10.1109/LCOMM.2026.3658407","DOIUrl":"https://doi.org/10.1109/LCOMM.2026.3658407","url":null,"abstract":"This letter proposes a novel uplink multiple access scheme that accommodates up to <inline-formula> <tex-math>$K$ </tex-math></inline-formula> low data-rate non-priority users (NPUs) within a selected priority user’s (PU) spectrum, without compromising the PU performance. The uplink base station decodes the PU symbol first, followed by the symbols of <inline-formula> <tex-math>$K$ </tex-math></inline-formula> NPUs, employing successive interference cancellation. A dynamic interference threshold limit, set after PU selection, determines <inline-formula> <tex-math>$K$ </tex-math></inline-formula> and NPU powers in closed-form using a novel instantaneous power allocation scheme (IPAS) inspired by underlay cognitive radio. Spectrum utilization is maximized with precise control over NPU target rate and <inline-formula> <tex-math>$K$ </tex-math></inline-formula>. Closed-form throughput analysis, validated via simulations, demonstrates superiority of the suggested scheme over a traditional multiple access employing static optimal power allocation.","PeriodicalId":13197,"journal":{"name":"IEEE Communications Letters","volume":"30 ","pages":"1007-1011"},"PeriodicalIF":4.4,"publicationDate":"2026-01-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175698","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-26DOI: 10.1109/LCOMM.2026.3658190
Daniel Dinis;Risto Wichman
Fluid antenna system (FAS) enhance wireless performance by dynamically selecting the best antenna port among many closely spaced ports. However, identifying the optimal port at each time step typically requires measuring all ports, which is costly due to pilot overhead and switching latency that may become a limiting factor in time-varying channels. In this work, we propose a spatio-temporal interpolation framework based on Gaussian processes (GP) that leverages both spatial and temporal channel correlations to estimate the channel response from a small subset of measured ports. Simulation results demonstrate that our approach achieves impressive performance, even outperforming full-port scanning in time-varying channels by effectively denoising through temporal smoothing.
{"title":"Spatio–Temporal Port Selection in Fluid Antennas Under Switching Delays","authors":"Daniel Dinis;Risto Wichman","doi":"10.1109/LCOMM.2026.3658190","DOIUrl":"https://doi.org/10.1109/LCOMM.2026.3658190","url":null,"abstract":"Fluid antenna system (FAS) enhance wireless performance by dynamically selecting the best antenna port among many closely spaced ports. However, identifying the optimal port at each time step typically requires measuring all ports, which is costly due to pilot overhead and switching latency that may become a limiting factor in time-varying channels. In this work, we propose a spatio-temporal interpolation framework based on Gaussian processes (GP) that leverages both spatial and temporal channel correlations to estimate the channel response from a small subset of measured ports. Simulation results demonstrate that our approach achieves impressive performance, even outperforming full-port scanning in time-varying channels by effectively denoising through temporal smoothing.","PeriodicalId":13197,"journal":{"name":"IEEE Communications Letters","volume":"30 ","pages":"992-996"},"PeriodicalIF":4.4,"publicationDate":"2026-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175601","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-26DOI: 10.1109/LCOMM.2026.3658128
Peng Jin;Xi Wang;Jing Xu;Ruikun Luo;Wei Liu
The metaverse, offering immersive and interactive experiences, has become a key focus resently. In this letter, we investigate a metaverse wireless communication system incorporating Non-Orthogonal Multiple Access (NOMA). To strike an effective balance between user Quality of Experience (QoE) and system cost, we propose a comprehensive algorithmic framework. Specifically, the problem is decomposed into two subproblems: subchannel allocation and user scheduling. For these, we design a NOMA-based Subchannel Allocation algorithm (NSA) and a Genetic Algorithm-based Joint User and Subchannel Allocation scheme (GA-JUSA), which are executed iteratively to enhance overall performance. Experimental results show that the proposed approach achieves up to a 22.98% improvement over existing methods, highlighting its potential for future applications.
{"title":"Joint Subchannel and User Scheduling for Wireless Metaverse Services: A Genetic Algorithm-Based Approach","authors":"Peng Jin;Xi Wang;Jing Xu;Ruikun Luo;Wei Liu","doi":"10.1109/LCOMM.2026.3658128","DOIUrl":"https://doi.org/10.1109/LCOMM.2026.3658128","url":null,"abstract":"The metaverse, offering immersive and interactive experiences, has become a key focus resently. In this letter, we investigate a metaverse wireless communication system incorporating Non-Orthogonal Multiple Access (NOMA). To strike an effective balance between user Quality of Experience (QoE) and system cost, we propose a comprehensive algorithmic framework. Specifically, the problem is decomposed into two subproblems: subchannel allocation and user scheduling. For these, we design a NOMA-based Subchannel Allocation algorithm (NSA) and a Genetic Algorithm-based Joint User and Subchannel Allocation scheme (GA-JUSA), which are executed iteratively to enhance overall performance. Experimental results show that the proposed approach achieves up to a 22.98% improvement over existing methods, highlighting its potential for future applications.","PeriodicalId":13197,"journal":{"name":"IEEE Communications Letters","volume":"30 ","pages":"1056-1060"},"PeriodicalIF":4.4,"publicationDate":"2026-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146223621","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-23DOI: 10.1109/LCOMM.2026.3657730
Seunghun Yu;Youngjoon Lee;Jinu Gong;Joonhyuk Kang
Federated learning (FL), widely used in privacy-critical applications, suffers from limited interpretability, whereas Kolmogorov–Arnold Networks (KAN) address this limitation via learnable spline functions. However, existing FL studies applying KAN overlook the communication overhead introduced by grid extension, which is essential for modeling complex functions. In this letter, we propose CG-FKAN, which compresses extended grids by sparsifying and transmitting only essential coefficients under a communication budget. Experiments show that CG-FKAN achieves up to 13.6% lower RMSE than fixed-grid KAN in communication-constrained settings. In addition, we derive a theoretical upper bound on its approximation error.
{"title":"CG-FKAN: Compressed-Grid Federated Kolmogorov–Arnold Networks for Communication Constrained Environment","authors":"Seunghun Yu;Youngjoon Lee;Jinu Gong;Joonhyuk Kang","doi":"10.1109/LCOMM.2026.3657730","DOIUrl":"https://doi.org/10.1109/LCOMM.2026.3657730","url":null,"abstract":"Federated learning (FL), widely used in privacy-critical applications, suffers from limited interpretability, whereas Kolmogorov–Arnold Networks (KAN) address this limitation via learnable spline functions. However, existing FL studies applying KAN overlook the communication overhead introduced by grid extension, which is essential for modeling complex functions. In this letter, we propose CG-FKAN, which compresses extended grids by sparsifying and transmitting only essential coefficients under a communication budget. Experiments show that CG-FKAN achieves up to 13.6% lower RMSE than fixed-grid KAN in communication-constrained settings. In addition, we derive a theoretical upper bound on its approximation error.","PeriodicalId":13197,"journal":{"name":"IEEE Communications Letters","volume":"30 ","pages":"1101-1105"},"PeriodicalIF":4.4,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146223617","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-23DOI: 10.1109/LCOMM.2026.3657403
Guanxing Zhang;He Wen;Jie Zhang
To address signal blockage challenges associated with the millimeter-wave (mmWave) band, we explore the potential of reconfigurable intelligent surfaces (RIS) to enhance performance of integrated sensing and communication (ISAC) systems. Unlike most studies assuming continuous phase control, we consider the more practical scenario of discrete phase-shifted RIS. Simulation results show that RIS in mmWave ISAC systems can mitigate blockage by building a virtual line-of-sight (LOS). Besides, we present the trade-off analysis of communication and radar performance using discrete phase-shifted RIS, comparing the continuous and discrete phase configurations in terms of mean square error (MSE) and normalized sideband power (NSP).
{"title":"Beamforming Design and Performance Analysis of Discrete Phase-Controlled RIS-Assisted mmWave ISAC Systems","authors":"Guanxing Zhang;He Wen;Jie Zhang","doi":"10.1109/LCOMM.2026.3657403","DOIUrl":"https://doi.org/10.1109/LCOMM.2026.3657403","url":null,"abstract":"To address signal blockage challenges associated with the millimeter-wave (mmWave) band, we explore the potential of reconfigurable intelligent surfaces (RIS) to enhance performance of integrated sensing and communication (ISAC) systems. Unlike most studies assuming continuous phase control, we consider the more practical scenario of discrete phase-shifted RIS. Simulation results show that RIS in mmWave ISAC systems can mitigate blockage by building a virtual line-of-sight (LOS). Besides, we present the trade-off analysis of communication and radar performance using discrete phase-shifted RIS, comparing the continuous and discrete phase configurations in terms of mean square error (MSE) and normalized sideband power (NSP).","PeriodicalId":13197,"journal":{"name":"IEEE Communications Letters","volume":"30 ","pages":"972-976"},"PeriodicalIF":4.4,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146082058","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Deep-learning-based Radio Frequency Fingerprint Identification (RFFI) leverages hardware-induced imperfections for device authentication. However, cross-receiver shifts induce category-dependent errors, and practical deployment is constrained by the trade-off between model complexity and latency. To address this, we propose a receiver-agnostic, multi-modal RFFI method that integrates Convolutional Neural Network (CNN)-derived spatial cues, Bidirectional Encoder Representations from Transformers (BERT)-based temporal modeling, and Carrier Frequency Offset (CFO) features. A two-stage knowledge distillation strategy transfers knowledge from a high-capacity Teacher (utilizing all modalities with 7.15M parameters) to a lightweight Student (utilizing only spatial and CFO features with 2.38M parameters). The Student operates without the computationally intensive BERT module during inference, achieving 66.7% parameter reduction. Under challenging cross-receiver conditions with a 9:3 receiver split, the Teacher achieves 88.77% accuracy, while the Distilled Student achieves 89.82% accuracy. The standalone Student without distillation achieves only 85.40%, demonstrating a + 4.42% improvement from cross-modal knowledge transfer. With 2.5 ms inference latency, the distilled model enables practical, receiver-agnostic RFFI deployment.
{"title":"Receiver-Agnostic Radio Frequency Fingerprint Identification Using BERT and Two-Stage Knowledge Distillation","authors":"Jiawen Shao;Zijiang Yang;Tiantian Tang;Chengcheng Liu;Yun Lin;Guan Gui","doi":"10.1109/LCOMM.2026.3656814","DOIUrl":"https://doi.org/10.1109/LCOMM.2026.3656814","url":null,"abstract":"Deep-learning-based Radio Frequency Fingerprint Identification (RFFI) leverages hardware-induced imperfections for device authentication. However, cross-receiver shifts induce category-dependent errors, and practical deployment is constrained by the trade-off between model complexity and latency. To address this, we propose a receiver-agnostic, multi-modal RFFI method that integrates Convolutional Neural Network (CNN)-derived spatial cues, Bidirectional Encoder Representations from Transformers (BERT)-based temporal modeling, and Carrier Frequency Offset (CFO) features. A two-stage knowledge distillation strategy transfers knowledge from a high-capacity Teacher (utilizing all modalities with 7.15M parameters) to a lightweight Student (utilizing only spatial and CFO features with 2.38M parameters). The Student operates without the computationally intensive BERT module during inference, achieving 66.7% parameter reduction. Under challenging cross-receiver conditions with a 9:3 receiver split, the Teacher achieves 88.77% accuracy, while the Distilled Student achieves 89.82% accuracy. The standalone Student without distillation achieves only 85.40%, demonstrating a + 4.42% improvement from cross-modal knowledge transfer. With 2.5 ms inference latency, the distilled model enables practical, receiver-agnostic RFFI deployment.","PeriodicalId":13197,"journal":{"name":"IEEE Communications Letters","volume":"30 ","pages":"942-946"},"PeriodicalIF":4.4,"publicationDate":"2026-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146082073","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
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：