Pub Date : 2025-11-25DOI: 10.1109/OJCOMS.2025.3636725
Haoshuo Zhang;Yufei Bo;Meixia Tao
Multimodal semantic communication has great potential to enhance downstream task performance by integrating complementary information across modalities. This paper introduces ProMSC-MIS, a novel Prompt-based Multimodal Semantic Communication framework for Multi-Spectral Image Segmentation. It enables efficient task-oriented transmission of spatially aligned RGB and thermal images over band-limited channels. Our framework has two main design novelties. First, by leveraging prompt learning and contrastive learning, unimodal semantic encoders are pre-trained to learn diverse and complementary semantic representations, where each modality serves as a cross-modal prompt for the other. Second, a semantic fusion module that combines cross-attention mechanism and squeeze-and-excitation (SE) networks is designed to effectively fuse cross-modal features. Experimental results demonstrate that ProMSC-MIS substantially outperforms conventional image transmission combined with state-of-the-art segmentation methods. Notably, it reduces the required communication cost by 50%–70% at the same segmentation performance, while also decreasing the storage overhead and computational complexity by 26% and 37%, respectively. Ablation studies also validate the effectiveness of the proposed pre-training and semantic fusion strategies. Our scheme is highly suitable for applications such as autonomous driving and nighttime surveillance.
{"title":"ProMSC-MIS: Prompt-Based Multimodal Semantic Communication for Multi-Spectral Image Segmentation","authors":"Haoshuo Zhang;Yufei Bo;Meixia Tao","doi":"10.1109/OJCOMS.2025.3636725","DOIUrl":"https://doi.org/10.1109/OJCOMS.2025.3636725","url":null,"abstract":"Multimodal semantic communication has great potential to enhance downstream task performance by integrating complementary information across modalities. This paper introduces ProMSC-MIS, a novel Prompt-based Multimodal Semantic Communication framework for Multi-Spectral Image Segmentation. It enables efficient task-oriented transmission of spatially aligned RGB and thermal images over band-limited channels. Our framework has two main design novelties. First, by leveraging prompt learning and contrastive learning, unimodal semantic encoders are pre-trained to learn diverse and complementary semantic representations, where each modality serves as a cross-modal prompt for the other. Second, a semantic fusion module that combines cross-attention mechanism and squeeze-and-excitation (SE) networks is designed to effectively fuse cross-modal features. Experimental results demonstrate that ProMSC-MIS substantially outperforms conventional image transmission combined with state-of-the-art segmentation methods. Notably, it reduces the required communication cost by 50%–70% at the same segmentation performance, while also decreasing the storage overhead and computational complexity by 26% and 37%, respectively. Ablation studies also validate the effectiveness of the proposed pre-training and semantic fusion strategies. Our scheme is highly suitable for applications such as autonomous driving and nighttime surveillance.","PeriodicalId":33803,"journal":{"name":"IEEE Open Journal of the Communications Society","volume":"6 ","pages":"9926-9941"},"PeriodicalIF":6.3,"publicationDate":"2025-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11268456","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145674725","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-25DOI: 10.1109/OJCOMS.2025.3636915
Haider A. H. Alobaidy;Mehran Behjati;Rosdiadee Nordin;Muhammad Aidiel Zulkifley;Nor Fadzilah Abdullah;Nur Fahimah Mat Salleh
This work proposes an empirical air-to-ground (A2G) propagation model specifically designed for cellular-connected unmanned aerial vehicles (UAVs). An in-depth aerial drive test was carried out within an operating Long-Term Evolution (LTE) network, gathering thorough measurements of key network parameters. Rigid preprocessing and statistical analysis of these data produced a strong foundation for training a new triple-layer machine learning (ML) model. The proposed ML framework employs a systematic hierarchical approach. Accordingly, the first two layers, Stepwise Linear Regression (STW) and Ensemble of Bagged Trees (EBT) generate predictions independently; meanwhile, the third layer, Gaussian Process Regression (GPR), explicitly acts as an aggregation layer, refining these predictions to accurately estimate Key Performance Indicators (KPIs) such as Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), Received Signal Strength (RSSI), and Path Loss (PL). Compared to traditional single-layer ML or computationally intensive ray-tracing approaches, the proposed triple-layer ML framework significantly improves predictive performance and robustness, achieving a coefficient of determination $(R^{2})$ values of approximately 0.99 in training and above 0.90 in testing while utilizing a minimal but effective feature set (log-transformed 3D and 2D propagation distances, azimuth, and elevation angles). This streamlined feature selection substantially reduces computing complexity, thus enhancing scalability across various operating environments. The proposed framework’s practicality and efficacy for real-world deployment in UAV-integrated cellular networks are further demonstrated by comparative analyses, which underscore its substantial improvement.
{"title":"Empirical 3-D Channel Modeling for Cellular-Connected UAVs: A Triple-Layer Machine Learning Approach","authors":"Haider A. H. Alobaidy;Mehran Behjati;Rosdiadee Nordin;Muhammad Aidiel Zulkifley;Nor Fadzilah Abdullah;Nur Fahimah Mat Salleh","doi":"10.1109/OJCOMS.2025.3636915","DOIUrl":"https://doi.org/10.1109/OJCOMS.2025.3636915","url":null,"abstract":"This work proposes an empirical air-to-ground (A2G) propagation model specifically designed for cellular-connected unmanned aerial vehicles (UAVs). An in-depth aerial drive test was carried out within an operating Long-Term Evolution (LTE) network, gathering thorough measurements of key network parameters. Rigid preprocessing and statistical analysis of these data produced a strong foundation for training a new triple-layer machine learning (ML) model. The proposed ML framework employs a systematic hierarchical approach. Accordingly, the first two layers, Stepwise Linear Regression (STW) and Ensemble of Bagged Trees (EBT) generate predictions independently; meanwhile, the third layer, Gaussian Process Regression (GPR), explicitly acts as an aggregation layer, refining these predictions to accurately estimate Key Performance Indicators (KPIs) such as Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), Received Signal Strength (RSSI), and Path Loss (PL). Compared to traditional single-layer ML or computationally intensive ray-tracing approaches, the proposed triple-layer ML framework significantly improves predictive performance and robustness, achieving a coefficient of determination <inline-formula> <tex-math>$(R^{2})$ </tex-math></inline-formula> values of approximately 0.99 in training and above 0.90 in testing while utilizing a minimal but effective feature set (log-transformed 3D and 2D propagation distances, azimuth, and elevation angles). This streamlined feature selection substantially reduces computing complexity, thus enhancing scalability across various operating environments. The proposed framework’s practicality and efficacy for real-world deployment in UAV-integrated cellular networks are further demonstrated by comparative analyses, which underscore its substantial improvement.","PeriodicalId":33803,"journal":{"name":"IEEE Open Journal of the Communications Society","volume":"6 ","pages":"9908-9925"},"PeriodicalIF":6.3,"publicationDate":"2025-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11269020","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145674696","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The new mid-band spectrum (6-24 GHz, including Frequency Range 3 (FR3)) has attracted significant attention from both academia and industry, which is the spectrum with continuous bandwidth that combines the coverage benefits of low frequency with the capacity advantages of high frequency. Considering the outdoor environment is the primary application scenario for mobile communications, this paper presents the first comprehensive summary of multi-scenario and multi-frequency channel characteristics based on the new mid-band channel measurements, mainly including Urban Macrocell (UMa), Urban Microcell (UMi), and Outdoor to Indoor (O2I). Specifically, the analysis of the channel characteristics is presented, such as path loss, delay spread, angular spread, channel sparsity, capacity, near-field and spatial non-stationary characteristics. Then, considering that satellite communication will be an important component of future communication systems, we examine the impact of clutter loss in air-to-ground communications. The analysis suggests that the frequency dependence of clutter loss is not significant for the mid-band. Additionally, given that penetration loss is frequency-dependent, we summarize its variation within the FR3 band. The experimental results show that the 3rd Generation Partnership Project (3GPP) TR 38.901 model is still a useful reference for the penetration loss of the wood, but there are significant deviations for the penetration loss of concrete and glass, and further improvement is needed. In summary, the findings of this paper provide both empirical data and theoretical support for the deployment of mid-band in future communication systems, as well as guidance for optimizing mid-band base station deployment in the communication environment. This paper provides a reference for the standards and research of potential spectra and technologies.
新的中频频谱（6- 24ghz，包括FR3 (Frequency Range 3)）是一种集低频覆盖优势和高频容量优势于一体的连续带宽频谱，受到了学术界和工业界的广泛关注。考虑到户外环境是移动通信的主要应用场景，本文首次综合总结了基于新型中频信道测量的多场景多频信道特性，主要包括城市宏蜂窝（UMa）、城市微蜂窝（UMi）和室内外（O2I）。具体而言，分析了信道特性，如路径损耗、延迟扩展、角扩展、信道稀疏性、容量、近场和空间非平稳特性。然后，考虑到卫星通信将成为未来通信系统的重要组成部分，我们研究了杂波损耗对空对地通信的影响。分析表明，中频段杂波损失的频率依赖性不显著。此外，考虑到穿透损耗是频率相关的，我们总结了其在FR3波段内的变化。实验结果表明，第三代伙伴计划（3GPP） TR 38.901模型对木材的侵彻损失仍有参考价值，但对混凝土和玻璃的侵彻损失存在较大偏差，需要进一步改进。综上所述，本文的研究结果为未来通信系统中频部署提供了经验数据和理论支持，也为优化通信环境中频基站部署提供了指导。本文为势谱标准的制定和技术的研究提供了参考。
{"title":"6G New Mid-Band/FR3 (6–24 GHz): Channel Measurement, Characteristics and Modeling","authors":"Haiyang Miao;Jianhua Zhang;Pan Tang;Qi Zhen;Jie Meng;Ximan Liu;Enrui Liu;Peijie Liu;Lei Tian;Guangyi Liu","doi":"10.1109/OJCOMS.2025.3636972","DOIUrl":"https://doi.org/10.1109/OJCOMS.2025.3636972","url":null,"abstract":"The new mid-band spectrum (6-24 GHz, including Frequency Range 3 (FR3)) has attracted significant attention from both academia and industry, which is the spectrum with continuous bandwidth that combines the coverage benefits of low frequency with the capacity advantages of high frequency. Considering the outdoor environment is the primary application scenario for mobile communications, this paper presents the first comprehensive summary of multi-scenario and multi-frequency channel characteristics based on the new mid-band channel measurements, mainly including Urban Macrocell (UMa), Urban Microcell (UMi), and Outdoor to Indoor (O2I). Specifically, the analysis of the channel characteristics is presented, such as path loss, delay spread, angular spread, channel sparsity, capacity, near-field and spatial non-stationary characteristics. Then, considering that satellite communication will be an important component of future communication systems, we examine the impact of clutter loss in air-to-ground communications. The analysis suggests that the frequency dependence of clutter loss is not significant for the mid-band. Additionally, given that penetration loss is frequency-dependent, we summarize its variation within the FR3 band. The experimental results show that the 3rd Generation Partnership Project (3GPP) TR 38.901 model is still a useful reference for the penetration loss of the wood, but there are significant deviations for the penetration loss of concrete and glass, and further improvement is needed. In summary, the findings of this paper provide both empirical data and theoretical support for the deployment of mid-band in future communication systems, as well as guidance for optimizing mid-band base station deployment in the communication environment. This paper provides a reference for the standards and research of potential spectra and technologies.","PeriodicalId":33803,"journal":{"name":"IEEE Open Journal of the Communications Society","volume":"6 ","pages":"9942-9960"},"PeriodicalIF":6.3,"publicationDate":"2025-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11269017","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145674724","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-25DOI: 10.1109/OJCOMS.2025.3637110
Jiahong Ning;Aimin Li;Gary C. F. Lee;Sumei Sun;Tingting Yang
This paper presents MARHO, a Multi-Agent Reinforcement learning-based Hybrid task Offloading framework, designed for maritime mobile edge computing (MEC) environments characterized by time-varying wireless channels, heterogeneous workloads, and stringent quality of service (QoS) requirements. The considered MEC architecture integrates uncrewed surface vessels (USVs), uncrewed aerial vehicles (UAVs), and a ship platform with high-performance edge servers. USVs generate sensing and computing tasks that can be (i) executed locally, (ii) offloaded to UAVs for aerial edge processing, or (iii) relayed through UAVs to the ship under line-of-sight (LoS) links. The system model jointly captures queueing dynamics, wireless transmission latency, computation delay, and battery constraints. The hybrid offloading problem is formulated as a Decentralized Partially Observable Markov Decision Process (Dec-POMDP), where each USV acts as an agent that decides its offloading mode under partial observations. To solve this, MARHO employs a centralized training and decentralized execution (CTDE) scheme, enabling agents to learn resource-aware strategies that effectively balance communication and computation. A Gym-based simulation environment is developed, integrating realistic maritime signal propagation, queue dynamics, and mixed offloading scenarios. The experimental results under different task loads demonstrate that MARHO consistently achieves higher throughput and has a lower average latency compared to the existing benchmark.
{"title":"MARHO: Hybrid Task Offloading in Maritime MEC via Multi-Agent Reinforcement Learning","authors":"Jiahong Ning;Aimin Li;Gary C. F. Lee;Sumei Sun;Tingting Yang","doi":"10.1109/OJCOMS.2025.3637110","DOIUrl":"https://doi.org/10.1109/OJCOMS.2025.3637110","url":null,"abstract":"This paper presents MARHO, a Multi-Agent Reinforcement learning-based Hybrid task Offloading framework, designed for maritime mobile edge computing (MEC) environments characterized by time-varying wireless channels, heterogeneous workloads, and stringent quality of service (QoS) requirements. The considered MEC architecture integrates uncrewed surface vessels (USVs), uncrewed aerial vehicles (UAVs), and a ship platform with high-performance edge servers. USVs generate sensing and computing tasks that can be (i) executed locally, (ii) offloaded to UAVs for aerial edge processing, or (iii) relayed through UAVs to the ship under line-of-sight (LoS) links. The system model jointly captures queueing dynamics, wireless transmission latency, computation delay, and battery constraints. The hybrid offloading problem is formulated as a Decentralized Partially Observable Markov Decision Process (Dec-POMDP), where each USV acts as an agent that decides its offloading mode under partial observations. To solve this, MARHO employs a centralized training and decentralized execution (CTDE) scheme, enabling agents to learn resource-aware strategies that effectively balance communication and computation. A Gym-based simulation environment is developed, integrating realistic maritime signal propagation, queue dynamics, and mixed offloading scenarios. The experimental results under different task loads demonstrate that MARHO consistently achieves higher throughput and has a lower average latency compared to the existing benchmark.","PeriodicalId":33803,"journal":{"name":"IEEE Open Journal of the Communications Society","volume":"6 ","pages":"10322-10337"},"PeriodicalIF":6.3,"publicationDate":"2025-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11269014","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145778105","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper presents a novel wireless-optical interconnection scheme employing cascaded transmissive and reflective metasurfaces to overcome the switching capacity limitations of conventional spatial light modulators and micro-electro-mechanical systems in data centers. We design a passive transmissive metasurface that splits an incident beam into N transmitted beams, which are subsequently reflected by a reflective metasurface to generate $2times N$ output beams, substantially enhancing switching capacity. To dynamically optimize resource utilization and prevent service disruptions due to congestion or underutilization, we develop an AI-driven traffic prediction algorithm for intelligent topology reconfiguration. Extensive simulations validate the system’s 1-to-4 beam-splitting capability with remarkably low insertion loss of 0.5dB, while achieving 91% traffic prediction accuracy-representing a 34% improvement over conventional long short-term memory (LSTM) models. The proposed data center architecture establishes a new paradigm for next-generation data center interconnects, offering superior capacity, minimal loss, and intelligent reconfigurability.
{"title":"Resilient Wireless-Optical Interconnection Scheme for Data Centers: Cascaded Reflective and Transmissive Meta-Surfaces","authors":"Weigang Hou;Weijie Qiu;Xiaoxue Gong;Yuxin Xu;Lei Guo","doi":"10.1109/OJCOMS.2025.3637098","DOIUrl":"https://doi.org/10.1109/OJCOMS.2025.3637098","url":null,"abstract":"This paper presents a novel wireless-optical interconnection scheme employing cascaded transmissive and reflective metasurfaces to overcome the switching capacity limitations of conventional spatial light modulators and micro-electro-mechanical systems in data centers. We design a passive transmissive metasurface that splits an incident beam into N transmitted beams, which are subsequently reflected by a reflective metasurface to generate <inline-formula> <tex-math>$2times N$ </tex-math></inline-formula> output beams, substantially enhancing switching capacity. To dynamically optimize resource utilization and prevent service disruptions due to congestion or underutilization, we develop an AI-driven traffic prediction algorithm for intelligent topology reconfiguration. Extensive simulations validate the system’s 1-to-4 beam-splitting capability with remarkably low insertion loss of 0.5dB, while achieving 91% traffic prediction accuracy-representing a 34% improvement over conventional long short-term memory (LSTM) models. The proposed data center architecture establishes a new paradigm for next-generation data center interconnects, offering superior capacity, minimal loss, and intelligent reconfigurability.","PeriodicalId":33803,"journal":{"name":"IEEE Open Journal of the Communications Society","volume":"6 ","pages":"10244-10253"},"PeriodicalIF":6.3,"publicationDate":"2025-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11269009","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145729276","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-21DOI: 10.1109/OJCOMS.2025.3635689
Ehsan Eslami;Walaa Hamouda
Network traffic classification (NTC) plays an essential role in managing, securing, and optimizing networks. Supervised learning methods face challenges such as label scarcity. Given that network traffic contains sensitive data and is distributed across multiple nodes, privacy-aware and scalable approaches are necessary for real-world deployment. In this paper, we introduce FedSSL-NTC, a privacy-enhancing federated framework that integrates self-supervised learning (SSL) and a traffic-adapted confident learning (CL) approach. In FedSSL-NTC, clients locally pretrain SSL models (Autoencoders or Tabular Contrastive Learning) and generate pseudo-labels. CL is then applied on the client side to reduce pseudo-label noise before federated classifier training. Robustness to non-independently and non-identically distributed (non-IID) data and class imbalance is achieved via FedProx, class-weighted loss, and a sample-size weighted FedAvg aggregation method. This framework uses Secure Aggregation to protect individual updates. On a self-generated + ISCX VPN-nonVPN dataset and the UCDavis–QUIC dataset, FedSSL-NTC achieves 95.88% and 98.24% accuracy (vs. centralized 96.29% and 98.76%), while reducing training time by approximately 4–$5times $ through parallel client updates. The method outperforms recent federated/self-supervised baselines on the same evaluation protocol (e.g., 6% improvement compared to FS-GAN). Therefore, FedSSL-NTC offers a practical path to high-accuracy NTC under privacy constraints, non-IID distributions, and label scarcity.
{"title":"FedSSL-NTC: A Robust Federated Self-Supervised Learning Framework for Network Traffic Classification Under Privacy Constraints","authors":"Ehsan Eslami;Walaa Hamouda","doi":"10.1109/OJCOMS.2025.3635689","DOIUrl":"https://doi.org/10.1109/OJCOMS.2025.3635689","url":null,"abstract":"Network traffic classification (NTC) plays an essential role in managing, securing, and optimizing networks. Supervised learning methods face challenges such as label scarcity. Given that network traffic contains sensitive data and is distributed across multiple nodes, privacy-aware and scalable approaches are necessary for real-world deployment. In this paper, we introduce FedSSL-NTC, a privacy-enhancing federated framework that integrates self-supervised learning (SSL) and a traffic-adapted confident learning (CL) approach. In FedSSL-NTC, clients locally pretrain SSL models (Autoencoders or Tabular Contrastive Learning) and generate pseudo-labels. CL is then applied on the client side to reduce pseudo-label noise before federated classifier training. Robustness to non-independently and non-identically distributed (non-IID) data and class imbalance is achieved via FedProx, class-weighted loss, and a sample-size weighted FedAvg aggregation method. This framework uses Secure Aggregation to protect individual updates. On a self-generated + ISCX VPN-nonVPN dataset and the UCDavis–QUIC dataset, FedSSL-NTC achieves 95.88% and 98.24% accuracy (vs. centralized 96.29% and 98.76%), while reducing training time by approximately 4–<inline-formula> <tex-math>$5times $ </tex-math></inline-formula> through parallel client updates. The method outperforms recent federated/self-supervised baselines on the same evaluation protocol (e.g., 6% improvement compared to FS-GAN). Therefore, FedSSL-NTC offers a practical path to high-accuracy NTC under privacy constraints, non-IID distributions, and label scarcity.","PeriodicalId":33803,"journal":{"name":"IEEE Open Journal of the Communications Society","volume":"6 ","pages":"10023-10041"},"PeriodicalIF":6.3,"publicationDate":"2025-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11263867","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145729419","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-21DOI: 10.1109/OJCOMS.2025.3636436
Mohammad Cheraghinia;Eli De Poorter;Jaron Fontaine;Merouane Debbah;Adnan Shahid
Wireless Technology Recognition (WTR) and localization are essential in modern communication systems, enabling efficient spectrum usage, coexistence across diverse technologies, and accurate positioning in dynamic environments. Real-world deployments must handle signals from different sampling rates, capturing devices, frequency bands, and propagation conditions. Traditional methods, such as energy detection and conventional Deep Learning (DL) models like Convolutional Neural Networks (CNNs), often fail to generalize across unseen technologies, environments, or tasks. In this work, we introduce a Transformer-based foundation model for both WTR and localization, pre-trained in a self-supervised manner on large-scale unlabeled aciq and Channel Impulse Response (CIR) timeseries data. The model aims for reusability and generalizability compared to single-task architectures. It leverages input patching for computational efficiency and employs a two-stage pipeline: self-supervised pre-training to learn general-purpose representations, followed by lightweight fine-tuning for task-specific adaptation. This enables the model to generalize to new wireless technologies and unseen environments using minimal labeled samples. Evaluations across short-range and long-range datasets show superior accuracy in WTR (up to 99.99%), Line-Of-Sight (LOS) detection (up to 100%), and ranging error correction (reducing Mean Absolute Error (MAE) by up to 50%), all while maintaining low computational complexity. These results underscore the potential of a reusable wireless foundation model for multi-task applications with minimal retraining.
{"title":"A Foundation Model for Wireless Technology Recognition and Localization Tasks","authors":"Mohammad Cheraghinia;Eli De Poorter;Jaron Fontaine;Merouane Debbah;Adnan Shahid","doi":"10.1109/OJCOMS.2025.3636436","DOIUrl":"https://doi.org/10.1109/OJCOMS.2025.3636436","url":null,"abstract":"Wireless Technology Recognition (WTR) and localization are essential in modern communication systems, enabling efficient spectrum usage, coexistence across diverse technologies, and accurate positioning in dynamic environments. Real-world deployments must handle signals from different sampling rates, capturing devices, frequency bands, and propagation conditions. Traditional methods, such as energy detection and conventional Deep Learning (DL) models like Convolutional Neural Networks (CNNs), often fail to generalize across unseen technologies, environments, or tasks. In this work, we introduce a Transformer-based foundation model for both WTR and localization, pre-trained in a self-supervised manner on large-scale unlabeled aciq and Channel Impulse Response (CIR) timeseries data. The model aims for reusability and generalizability compared to single-task architectures. It leverages input patching for computational efficiency and employs a two-stage pipeline: self-supervised pre-training to learn general-purpose representations, followed by lightweight fine-tuning for task-specific adaptation. This enables the model to generalize to new wireless technologies and unseen environments using minimal labeled samples. Evaluations across short-range and long-range datasets show superior accuracy in WTR (up to 99.99%), Line-Of-Sight (LOS) detection (up to 100%), and ranging error correction (reducing Mean Absolute Error (MAE) by up to 50%), all while maintaining low computational complexity. These results underscore the potential of a reusable wireless foundation model for multi-task applications with minimal retraining.","PeriodicalId":33803,"journal":{"name":"IEEE Open Journal of the Communications Society","volume":"6 ","pages":"9879-9896"},"PeriodicalIF":6.3,"publicationDate":"2025-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11264506","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145674732","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-19DOI: 10.1109/OJCOMS.2025.3635533
Sandra Aladin;Lena Wosinska;Christine Tremblay
Fast and accurate estimation of lightpaths’ quality of transmission (QoT) is crucial for ensuring quality of service (QoS) and seamless operation in real-world optical networks. Machine learning (ML) algorithms are promising tools for QoT estimation of lightpaths before their establishment. In multi-domain optical networks, where learned QoT estimation models must be transferred between heterogeneous environments with limited target data, deep neural networks (DNNs) have shown promising results. However, DNN-based transfer learning (TL) approaches using fine-tuned artificial neural networks (ANNs) and convolutional neural networks (CNNs), offer limited interpretability. Consequently, little insight into the decision-making process is provided, and large labeled datasets as well as high computational resources are required, limiting their suitability for real-time, large-scale deployment in production networks. To address these challenges, we propose a novel lightweight and interpretable TL framework that integrates the Boruta-SHAP algorithm for automated feature selection (FS) together with two domain adaptation (DA) techniques: Feature Augmentation and Correlation Alignment. In contrast to the existing approaches based on DNN, our strategy leverages interpretable and efficient ML models to enhance the adaptability across diverse datasets in real-world network environments. We show that our random forest (RF)-based models achieve better performance than the ANN-based models, without sacrificing the classification accuracy. The FS via Boruta-SHAP allows for reducing dimensionality as well as training and inference times up to 70.68%, and 41.64%, respectively. Our proposed framework outperforms DA baseline models achieving 99.35% accuracy improvement in domain shift. Moreover, it offers 86% accuracy with a 99.83% reduction in the size of the target domain.
{"title":"Automated, Interpretable and Efficient ML Models for Real-World Lightpaths’ Quality of Transmission Estimation","authors":"Sandra Aladin;Lena Wosinska;Christine Tremblay","doi":"10.1109/OJCOMS.2025.3635533","DOIUrl":"https://doi.org/10.1109/OJCOMS.2025.3635533","url":null,"abstract":"Fast and accurate estimation of lightpaths’ quality of transmission (QoT) is crucial for ensuring quality of service (QoS) and seamless operation in real-world optical networks. Machine learning (ML) algorithms are promising tools for QoT estimation of lightpaths before their establishment. In multi-domain optical networks, where learned QoT estimation models must be transferred between heterogeneous environments with limited target data, deep neural networks (DNNs) have shown promising results. However, DNN-based transfer learning (TL) approaches using fine-tuned artificial neural networks (ANNs) and convolutional neural networks (CNNs), offer limited interpretability. Consequently, little insight into the decision-making process is provided, and large labeled datasets as well as high computational resources are required, limiting their suitability for real-time, large-scale deployment in production networks. To address these challenges, we propose a novel lightweight and interpretable TL framework that integrates the Boruta-SHAP algorithm for automated feature selection (FS) together with two domain adaptation (DA) techniques: Feature Augmentation and Correlation Alignment. In contrast to the existing approaches based on DNN, our strategy leverages interpretable and efficient ML models to enhance the adaptability across diverse datasets in real-world network environments. We show that our random forest (RF)-based models achieve better performance than the ANN-based models, without sacrificing the classification accuracy. The FS via Boruta-SHAP allows for reducing dimensionality as well as training and inference times up to 70.68%, and 41.64%, respectively. Our proposed framework outperforms DA baseline models achieving 99.35% accuracy improvement in domain shift. Moreover, it offers 86% accuracy with a 99.83% reduction in the size of the target domain.","PeriodicalId":33803,"journal":{"name":"IEEE Open Journal of the Communications Society","volume":"6 ","pages":"9785-9801"},"PeriodicalIF":6.3,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11261392","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145612078","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper proposes a multi-agent deep reinforcement learning (DRL) framework based on proximal policy optimization (PPO) for joint resource optimization in full-duplex (FD) reconfigurable intelligent surface (RIS)-aided non-orthogonal multiple access (NOMA) integrated sensing and communication (ISAC) systems. The goal is to maximize the minimum beampattern gain under quality-of-service (QoS) constraints for both uplink (UL) and downlink (DL) users. The optimization jointly controls transmit beamforming, RIS phase shift, DL power allocation, and UL transmit power. A centralized training with decentralized execution approach is adopted, where two agents are defined: a DL agent responsible for DL beamforming, RIS configuration, and power allocation, and a UL agent responsible for uplink power control. Each agent interacts with the shared environment, which comprises the base station (BS), RIS, and users, and learns its optimal policy under time-varying channels and mutual interference. Simulation results demonstrate that the proposed multi-agent PPO (MA-PPO) significantly outperforms baseline methods, including single-agent PPO and heuristic schemes, in terms of convergence speed, sum-rate, and beampattern gain. Moreover, the MA-PPO method exhibits superior scalability and performance in FD mode over half-duplex (HD) counterparts under various user densities and RIS configurations, showcasing its effectiveness for real-time joint communication and sensing in next-generation wireless networks.
{"title":"Multi-Agent PPO-Based Resource Optimization for Full-Duplex RIS-Aided NOMA-ISAC Systems","authors":"Nonis Wara;Anal Paul;Keshav Singh;Aryan Kaushik;Wonjae Shin","doi":"10.1109/OJCOMS.2025.3635274","DOIUrl":"https://doi.org/10.1109/OJCOMS.2025.3635274","url":null,"abstract":"This paper proposes a multi-agent deep reinforcement learning (DRL) framework based on proximal policy optimization (PPO) for joint resource optimization in full-duplex (FD) reconfigurable intelligent surface (RIS)-aided non-orthogonal multiple access (NOMA) integrated sensing and communication (ISAC) systems. The goal is to maximize the minimum beampattern gain under quality-of-service (QoS) constraints for both uplink (UL) and downlink (DL) users. The optimization jointly controls transmit beamforming, RIS phase shift, DL power allocation, and UL transmit power. A centralized training with decentralized execution approach is adopted, where two agents are defined: a DL agent responsible for DL beamforming, RIS configuration, and power allocation, and a UL agent responsible for uplink power control. Each agent interacts with the shared environment, which comprises the base station (BS), RIS, and users, and learns its optimal policy under time-varying channels and mutual interference. Simulation results demonstrate that the proposed multi-agent PPO (MA-PPO) significantly outperforms baseline methods, including single-agent PPO and heuristic schemes, in terms of convergence speed, sum-rate, and beampattern gain. Moreover, the MA-PPO method exhibits superior scalability and performance in FD mode over half-duplex (HD) counterparts under various user densities and RIS configurations, showcasing its effectiveness for real-time joint communication and sensing in next-generation wireless networks.","PeriodicalId":33803,"journal":{"name":"IEEE Open Journal of the Communications Society","volume":"6 ","pages":"9802-9820"},"PeriodicalIF":6.3,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11261334","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145612060","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-19DOI: 10.1109/OJCOMS.2025.3635226
Boqing Zhou;Sujun Li;Decheng Miao
Replication nodes can compromise a network by not only stealing confidential data but also by selectively forwarding packets and injecting false data to the base station. This undermines the base station’s decision-making, ultimately allowing attackers to control the network. Scholars have proposed various solutions to deal with this attack. However, network performance is still susceptible to its impact because of their inherent limitations. In this paper, we introduce a novel authentication scheme. In this scheme, the network contains high-energy nodes, mobile sensor nodes (MSNs), and a base station. High energy nodes act as cluster heads. MSNs must be authenticated by a cluster head before they can obtain or provide data to the cluster head. Authentication between intra-cluster nodes relies on key information pre-distributed by the cluster head, leveraging a Bloom filter for efficient verification. Within a cluster, node $a$ will only forward node $b$ ’s data after node $b$ is authenticated by node $a$ . The analysis and simulation validate that the proposed scheme significantly enhances the network’s resilience against replication attacks. The communication probability between high-energy nodes is about 90%, and the probability that high-energy nodes can complete MSNs’ authentication within 1 hop is about 1; after introducing the bloom filter scheme within the cluster, the storage overhead of MSNs can be reduced by 90%, and the impact on the transmission of information from MSNs within the cluster to the cluster head can be ignored.
{"title":"A Two-Layer Authentication Scheme Against Node Replication Attacks in Mobile Heterogeneous Sensor Networks","authors":"Boqing Zhou;Sujun Li;Decheng Miao","doi":"10.1109/OJCOMS.2025.3635226","DOIUrl":"https://doi.org/10.1109/OJCOMS.2025.3635226","url":null,"abstract":"Replication nodes can compromise a network by not only stealing confidential data but also by selectively forwarding packets and injecting false data to the base station. This undermines the base station’s decision-making, ultimately allowing attackers to control the network. Scholars have proposed various solutions to deal with this attack. However, network performance is still susceptible to its impact because of their inherent limitations. In this paper, we introduce a novel authentication scheme. In this scheme, the network contains high-energy nodes, mobile sensor nodes (MSNs), and a base station. High energy nodes act as cluster heads. MSNs must be authenticated by a cluster head before they can obtain or provide data to the cluster head. Authentication between intra-cluster nodes relies on key information pre-distributed by the cluster head, leveraging a Bloom filter for efficient verification. Within a cluster, node <inline-formula> <tex-math>$a$ </tex-math></inline-formula> will only forward node <inline-formula> <tex-math>$b$ </tex-math></inline-formula>’s data after node <inline-formula> <tex-math>$b$ </tex-math></inline-formula> is authenticated by node <inline-formula> <tex-math>$a$ </tex-math></inline-formula>. The analysis and simulation validate that the proposed scheme significantly enhances the network’s resilience against replication attacks. The communication probability between high-energy nodes is about 90%, and the probability that high-energy nodes can complete MSNs’ authentication within 1 hop is about 1; after introducing the bloom filter scheme within the cluster, the storage overhead of MSNs can be reduced by 90%, and the impact on the transmission of information from MSNs within the cluster to the cluster head can be ignored.","PeriodicalId":33803,"journal":{"name":"IEEE Open Journal of the Communications Society","volume":"6 ","pages":"9897-9907"},"PeriodicalIF":6.3,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11261337","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145674703","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","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：