Intelligent spectrum management is crucial for improving spectrum efficiency and achieving secure utilization of spectrum resources. However, existing intelligent spectrum management methods, typically based on small-scale models, suffer from notable limitations in recognition accuracy, convergence speed, and generalization, particularly in the complex and dynamic spectrum environments. To address these challenges, this paper proposes a novel spectrum foundation model, termed SpectrumFM, establishing a new paradigm for spectrum management. SpectrumFM features an innovative encoder architecture that synergistically exploits the convolutional neural networks and the multi-head self-attention mechanisms to enhance feature extraction and enable robust representation learning. The model is pre-trained via two novel self-supervised learning tasks, namely masked reconstruction and next-slot signal prediction, which leverage large-scale in-phase and quadrature (IQ) data to achieve comprehensive and transferable spectrum representations. Furthermore, a parameter-efficient fine-tuning strategy is proposed to enable SpectrumFM to adapt to various downstream spectrum management tasks, including automatic modulation classification (AMC), wireless technology classification (WTC), spectrum sensing (SS), and anomaly detection (AD). Extensive experiments demonstrate that SpectrumFM achieves superior performance in terms of accuracy, robustness, adaptability, few-shot learning efficiency, and convergence speed, consistently outperforming conventional methods across multiple benchmarks. Specifically, SpectrumFM improves AMC accuracy by up to 12.1% and WTC accuracy by 9.3%, achieves an area under the curve (AUC) of 0.97 in SS at -4 dB signal-to-noise ratio (SNR), and enhances AD performance by over 10%.

The enhancement of spectrum efficiency and the realization of secure spectrum utilization are critically dependent on spectrum cognition. However, existing spectrum cognition methods often exhibit limited generalization and suboptimal accuracy when deployed across diverse spectrum environments and tasks. To overcome these challenges, we propose a spectrum foundation model, termed SpectrumFM, which provides a new paradigm for spectrum cognition. An innovative spectrum encoder that exploits the convolutional neural networks and the multi-head self attention mechanisms is proposed to effectively capture both fine-grained local signal structures and high-level global dependencies in the spectrum data. To enhance its adaptability, two novel self-supervised learning tasks, namely masked reconstruction and next-slot signal prediction, are developed for pre-training SpectrumFM, enabling the model to learn rich and transferable representations. Furthermore, low-rank adaptation (LoRA) parameter-efficient fine-tuning is exploited to enable SpectrumFM to seamlessly adapt to various downstream spectrum cognition tasks, including spectrum sensing (SS), anomaly detection (AD), and wireless technology classification (WTC). Extensive experiments demonstrate the superiority of SpectrumFM over state-of-the-art methods. Specifically, it improves detection probability in the SS task by 30% at -4 dB signal-to-noise ratio (SNR), boosts the area under the curve (AUC) in the AD task by over 10%, and enhances WTC accuracy by 9.6%.1

Spectrum map construction is a key technology for enhancing dynamic spectrum management and spectrum efficiency in the sixth-generation wireless communication networks. However, traditional spectrum map construction methods face a dual bottleneck in model generalizability and data dependence. Specifically, the model-driven methods struggle to adapt to dynamic and complex electromagnetic environments, whereas data-driven methods depend on the quality of training data, including sampling density and spatial correlation complexity. To address these challenges, a vision transformer-based large model for spectrum map construction is proposed, namely GenSpectraLM. Inspired by bidirectional encoder representations from transformers masked semantic inference and masked autoencoders local-global construction mechanism, GenSpectraLM employs self-supervised masked pretraining to implicitly learn electromagnetic propagation patterns from diverse datasets. Then, fine-tune is performed to achieve cross-scenario generalization. Simulation results demonstrate that GenSpectraLM achieves accurate spectrum map construction with the root mean squared error of 1.3286 at a sampling rate of 25%. It consistently outperforms benchmark methods by approximately 40%, effectively addressing data efficiency challenges in complex environments.

Spectrum sensing is of crucial importance for improving spectrum efficiency and realizing immersive communication. Deep learning (DL) has been introduced for spectrum sensing, with test statistics generated directly from signal samples in an automatic manner. However, most of the existing data-driven spectrum sensing methods are based on supervised learning and they usually require a massive amount of labeled training data to achieve high detection performance. It is difficult to obtain sufficient labeled training data in practice. To address this issue, a pseudo labeling multi-scale network (PLMSNet) for semi-supervised spectrum sensing is proposed to make the best use of a majority of unlabeled samples and achieves well detection performance with only a few of labeled training samples. Moreover, the proposed scheme is implemented in a real-world software defined radio (SDR) communication system. Both simulation and real-world experiments demonstrate that our proposed method achieves superior detection performance compared with the benchmark methods.

Federated Learning (FL) has emerged as a promising paradigm for secure data sharing in Industrial Internet of Things (IIoT), enabling collaborative model training without direct exchange of raw data. However, recent studies have shown that FL still suffers from privacy vulnerabilities, where adversaries can reconstruct sensitive information by analyzing shared model parameters. Although several privacy-preserving FL (PPFL) schemes have been proposed to address these challenges, they primarily focus on protecting local model privacy, with limited attention to protecting global model confidentiality during aggregation. Additionally, their reliance on centralized aggregation servers introduces risks of single points of failure. To address these challenges, we propose a novel privacy-preserving blockchain-based FL framework (PBFL) that integrates blockchain, homomorphic encryption (HE), and a single masking. Specifically, PBFL employs HE to enable secure model training within the ciphertext domain, ensuring global model confidentiality. The single masking technique allows clients to apply unique random masks to their encrypted local model updates, enabling secure aggregation while preserving local privacy. Additionally, PBFL leverages blockchain for decentralized aggregation and encrypted model storage, effectively mitigating the risks associated with centralized servers. Experimental results demonstrate that PBFL achieves comparable model accuracy to state-of-the-art solutions while providing enhanced privacy protection. Furthermore, even with a client dropout rate of up to 30%, PBFL outperforms other blockchain-based PPFL methods in terms of computational and communication efficiency.

Federated Learning (FL) has emerged as a promising paradigm for privacy-preserving collaborative model training in the Industrial Internet of Things (IIoT). By leveraging the decentralization, immutability, and transparency of blockchain technology, Blockchain-enabled FL (BFL) has gained significant attention for enhancing FL’s security and reliability. However, BFL still faces challenges in motivating client participation. While several incentive mechanisms have been proposed, most primarily focus on short-term rewards and overlook the long-term influence of individual contributions on global model performance. To address these challenges, we propose a novel BFL framework that integrates model training with blockchain mining on the client side. Specifically, we design a long-term incentive mechanism based on repeated game theory, where the interactions between participants and the task publisher (TP) are modeled as an infinitely repeated game. We formally prove the existence of a Subgame Perfect Nash Equilibrium, providing theoretical guarantees for stable long-term cooperation. Furthermore, we introduce a hybrid reward scheme that jointly considers contributions to both training and mining tasks, encouraging sustained engagement and attracting new participants. Extensive experiments on MNIST and CIFAR-10 validate that the proposed mechanism enhances the robustness of FL and effectively promotes long-term client participation.

Automatic modulation classification (AMC) plays a vital role in advancing future wireless communication networks. Although deep learning (DL)-based AMC frameworks have demonstrated remarkable classification capabilities, they typically require large-scale training datasets and assume consistent class distributions between training and testing data-prerequisites that prove challenging in few-shot and open-set scenarios. To address these limitations, we propose a novel few-shot open-set automatic modulation classification (FSOS-AMC) framework that integrates a multi-sequence multi-scale attention network (MS-MSANet), meta-prototype training, and a modular open-set classifier. The MS-MSANet extracts features from multi-sequence input signals, while meta-prototype training optimizes both the feature extractor and the modular open-set classifier, which can effectively categorize testing data into known modulation types or identify potential unknown modulations. Extensive simulation results demonstrate that our FSOS-AMC framework achieves superior performance in few-shot open-set scenarios compared to state-of-the-art methods. Specifically, the framework exhibits higher classification accuracy for both known and unknown modulations, as validated by improved accuracy and area under the receiver operating characteristic curve (AUROC) metrics. Moreover, the proposed framework demonstrates remarkable robustness under challenging low signal-to-noise ratio (SNR) conditions, significantly outperforming existing approaches.

Unmanned aerial vehicles (UAVs) are increasingly being integrated into next-generation networks to enhance communication coverage and network capacity. However, the dynamic and mobile nature of UAVs poses significant security challenges, including jamming, eavesdropping, and cyber-attacks. To address these security challenges, this paper proposes a federated learning-based lightweight network with zero trust for enhancing the security of UAV networks. A novel lightweight spectrogram network is proposed for UAV authentication and rejection, which can effectively authenticate and reject UAVs based on spectrograms. Experiments highlight LSNet’s superior performance in identifying both known and unknown UAV classes, demonstrating significant improvements over existing benchmarks in terms of accuracy, model compactness, and storage requirements. Notably, LSNet achieves an accuracy of over 80% for known UAV types and an Area Under the Receiver Operating Characteristic (AUROC) of 0.7 for unknown types when trained with all five clients. Further analyses explore the impact of varying the number of clients and the presence of unknown UAVs, reinforcing the practical applicability and effectiveness of our proposed framework in real-world FL scenarios.

Wireless signal recognition (WSR) is crucial in modern and future wireless communication networks since it aims to identify properties of the received signal. Although many deep learning-based WSR models have been developed, they still rely on a large amount of labeled training data. Thus, they cannot tackle the few-sample problem in the practically and dynamically changing wireless communication environment. To overcome this challenge, a novel SSwsrNet framework is proposed by using the deep residual shrinkage network (DRSN) and semi-supervised learning. The DRSN can learn discriminative features from noisy signals. Moreover, a modular semi-supervised learning method that combines labeled and unlabeled data using MixMatch is exploited to further improve the classification performance under few-sample conditions. Extensive simulation results on automatic modulation classification (AMC) and wireless technology classification (WTC) demonstrate that our proposed WSR scheme can achieve better performance than the benchmark schemes in terms of classification accuracy. This novel method enables more robust and adaptive signal recognition for next-generation wireless networks.

Automatic modulation classification is of crucial importance in wireless communication networks. Deep learning based automatic modulation classification schemes have attracted extensive attention due to the superior accuracy. However, the data-driven method relies on a large amount of training samples and the classification accuracy is poor in the low signal-to-noise radio (SNR). In order to tackle these problems, a novel data-and-knowledge dual-driven automatic modulation classification scheme based on radio frequency machine learning is proposed by exploiting the attribute features of different modulations. The visual model is utilized to extract visual features. The attribute learning model is used to learn the attribute semantic representations. The transformation model is proposed to convert the attribute representation into the visual space. Extensive simulation results demonstrate that our proposed automatic modulation classification scheme can achieve better performance than the benchmark schemes in terms of the classification accuracy, especially in the low SNR. Moreover, the confusion among high-order modulations is reduced by using our proposed scheme compared with other traditional schemes.

Wireless technique classification (WTC) is of crucial importance in Internet of Things for realizing efficient spectrum sharing and interference management. However, the existing deep-learning-based methods have low classification accuracy, especially at low signal-to-noise ratio levels. In this article, a multiscale convolutional neural network framework is proposed for WTC. A multiscale module is exploited to capture the higher abstraction features. Simulation results demonstrate that our proposed scheme can achieve a better classification performance and a higher convergence speed compared to the state-of-the-art schemes.