We consider a cluster-based collaborative spectrum sensing (CSS) scheme in energy harvesting cognitive wireless communication network (EH-CWCN), where cognitive nodes (CNs) are clustered based on their received power levels for enhancing sensing performance. In CSS scheme, time resource is limited and shared by energy harvesting, spectrum sensing, and data transmission. The purpose of this paper is to maximize the average throughput of EH-CWCN by identifying the optimal parameter set, including the durations of energy harvesting and spectrum sensing, local detection threshold, and the number of CNs. Moreover, it is hard to confirm the optimal local detection threshold at CNs with different received power levels. By constructing a fictitious cognitive node (FCN), which is assumed to have the same sensing performance as the sink node, the process for finding the optimal local detection threshold can be converted into the process for searching the optimal received signal-to-noise ratio of the FCN. Then, we formulate the general optimization problem under the collision constraint and energy constraint. Then, the existence and uniqueness of time fractions for energy harvesting and spectrum sensing are proved in this paper. The optimal parameter set is achieved by using the bisection method and a simplified linear search method. Finally, the theoretical analysis and the impact of the optimized parameters on the system performance are verified and shown through numerical simulations.

Specific Emitter Identification (SEI) technology enhances the physical-layer security of the Internet of Things (IoT). However, traditional SEI systems primarily focus on identification tasks within static closed-set environments, limiting their applicability in dynamic open-set scenarios. To address the challenges of Few-Shot Incremental Open-Set Recognition (FIOSR), which involves the expansion of specific emitter categories and Open-Set Recognition (OSR), we propose an Open-Set Few-Shot Class Incremental Learning (OFSCIL) framework. Our framework decouples the feature extractor from the classifier, and optimizes the extractor using an innovative integrated loss function and class augmentation. Moreover, we design a sustainably evolving prototype classifier with a prototype calibration method to alleviate the catastrophic forgetting and overfitting in the Few-Shot Class Incremental Learning (FSCIL) phase. For recognizing unknown categories, we design an open-set classifier based on prototype distance. Additionally, a semi-supervised clustering algorithm using cosine distance is introduced to further distinguish unknown signals. Experimental results on the benchmark dataset show that OFSCIL is well adapted to OSR, FSCIL and FIOSR tasks, highlighting its capability to balance computational efficiency and accuracy.

Spectrum is the indispensable resource for 5G wireless systems and beyond. The dynamic spectrum management (DSM) framework with spectrum sharing among different kinds of users is essential to satisfy the sustainable rapid increase of the bandwidth requirement. Based on the analysis of the DSM as well as the latent applications of the machine learning and artificial intelligence (AI) in the spectrum access, this paper will address the problem of incorporating intelligence into the spectrum management. Firstly, the spectrum opportunity (SOP) related information is layered into the spectrum data, the spectrum information and the spectrum knowledge based on the processing and abstractive level. The spectrum knowledge is formally defined as the extendible and scalable information to reason and predict the SOP usability as well as the outcome of the SOP occupancy. Then the smart spectrum model (SSM) is proposed with the spectrum knowledge as the key enabler and the coupled SOP exploration and exploitation as the core feature. Under the SSM framework, the spectrum knowledge and real-time observing (SKRO) enabled SOP exploration scheme is designed, which can make use of both the historical information and the real-time sensing information for smart SOP exploration. Extensive simulations are provided which demonstrate that the SKRO enabled SSM can achieve much better SOP utilization while satisfying the required legacy assurance. Specifically, in the low signal to noise environment, at least 16.55% gains on the SOP utilization ratio can be obtained compared with the DSM.

Specific Emitter Identification (SEI) is crucial for ensuring the security of physical layer communication. However, signal characteristics can be affected by various factors such as environmental and equipment variations. An effective SEI system must continuously learn and adapt to these changes to maintain accurate signal recognition. This study proposes an advanced domain incremental learning (DIL) framework for SEI, named Adaptive Domain-Incremental Learning with Knowledge Replay and Domain Alignment (ADIRA). ADIRA employs knowledge replay and distillation strategies, along with adaptive coefficients, to balance the model’s performance in recognizing signals across both new and old domains. To address the variations in signal data feature distributions across different domains, we introduce a domain alignment strategy based on adversarial training. This approach integrates embedding distillation loss with supervised contrastive loss, significantly enhancing the model’s adaptability to domain changes. Experimental results on two benchmark datasets demonstrate that ADIRA achieves performance only 0.42% and 1.71% lower than joint training, with replay samples constituting just 1.1% and 1.5% of the training set, effectively mitigating catastrophic forgetting.

The high energy efficiency for cognitive sensor network (CSN) has been widely studied for supporting various applications in military and civil fields. By identifying time fraction, local detection threshold, and the number of cognitive sensors, we discuss some design considerations for a cluster-based collaborative spectrum sensing scheme. We formulate the optimization problem to maximize energy efficiency of CSN under a collision constraint and false alarm probability constraint. In order to overcome the problem of finding the optimal detection threshold when k-out-of-M fusion rule is used by the sink node (SN), we design a fictitious cognitive sensor (FCS), which has the same sensing performance as SN. The process for finding the optimal local detection threshold can be converted into the process for searching an optimal received SNR of FCS. Theoretical analysis shows that there exists a unique time fraction to maximize the energy efficiency. Finally, our theoretical analysis is verified through numeric simulations.

In this paper, we detect illegal spectrum access behaviors in the dynamic spectrum sharing of the cognitive wireless networks, in which multiple authorized users (AUs) work in the same channel. First, we detect whether there are users occupy the channel, and if the channel is occupied, then recognize whether there is the illegal user (IU) and infer the number of AUs roughly. In the light of the thorny challenge that legal spectrum utilization behaviors and illegal spectrum access behaviors coexist probabilistically, the spectrum sensing problem is formulated as a mathematical model of ternary hypothesis test. Moreover, when multiple AUs are working at the same time, it is extremely difficult to detect IU. To tackle the problem, taking into account the various potential combinations of AUs and IU, we exploit a two-step detector and derive related detection schemes based on the generalized likelihood ratio test (GLRT), the Rao test, and the Wald test following the generalized multi-hypothesis Neyman-Pearson (GMNP) criterion respectively. Additionally, we design a cooperative spectrum sensing (CSS) scheme on the basis of the global GMNP criteria, in which the distributed detection framework composed of ${K}$ sensing nodes and a fusion center (FC). Finally, to verify the performance of the proposed detection scheme in a variety of parameter configurations, we provide comprehensive simulations and find that there will be at most $({{2^{N}}-1})$ detection valleys when ${N}$ AUs work at the same time. Compared with single-sensing node detection, the CSS scheme can significantly improve the detection performance of the proposed detection schemes, especially in the upper bound and valley of the detection performance.