To overcome the limitations of traditional satellite navigation technologies in complex and signal-obstructed industrial environments, this paper presents MSM-Pnet, a novel semi-supervised FM-based positioning framework leveraging FM signals of opportunity. By integrating wavelet packet decomposition with a multi-scale Vision Transformer and a hybrid masking strategy that combines random and time–frequency-aware masking, MSM-Pnet introduces a masked autoencoder architecture capable of robust positioning with limited labeled data. Experimental results demonstrate that MSM-Pnet consistently outperforms conventional supervised learning methods in both indoor and outdoor environments, while also significantly reducing model complexity. These results highlight the method’s potential as a cost-effective and scalable solution for seamless indoor–outdoor positioning for Internet of Things systems.

In covert communications, the position of the warden has a variety of situations, which leads to different scenes that require different covert communication schemes to ensure the security. In response to this situation, in this paper, a beamforming optimization method of covert communications for multi-position warden is proposed. Firstly, we formulate a general optimization problem and optimize it to maximize the covert communication rate of the user based on Dinkelbach’s transform. Subsequently, according to the optimized general optimization problem, we propose three schemes for three scenes corresponding to different fixed warden positions, using appropriate technologies for assistance in each scheme. Specifically, the intelligent reflecting surface (IRS) is used in Scene 1 and the integrated communication and jamming (ICAJ) is used in Scenes 2 and 3, and these technologies can assist the covert communication. Moreover, we propose an alternate optimization (AO) algorithm to solve the optimization problem of Scene 1 for its optimal covert communication performance. Additionally, we also propose an AO algorithm to solve the optimization problems of Scenes 2 and 3 to optimize the active beamforming. Simulation results demonstrate the effectiveness of all three proposed schemes, that outperform their respective benchmark schemes.

Automatic modulation recognition is essential for large-scale wireless communications, but traditional methods often ignore unknown signals in open-set conditions, leading to their incorrect classification as known types and thereby compromising system reliability and communication security. To handle this challenge, a novel automatic modulation open set recognition (AMOSR) model based on deep metric learning with embedding enhancement is proposed in this article. First, for each known example, deep neural model is employed to separately extract the original in-phase and quadrature (IQ) signal and its instantaneous features, which are then fused to obtain embedding. Second, random erasing is employed to the original IQ signal and instantaneous features separately to obtain an augmented example, which has similar structure but different semantics with known example, and the embedding of this example is obtained by using first step. Then, the embedding space, in which tuplet loss and margin loss are combined with the embeddings of known and augmented examples, is trained to improve the overall performance of the model. Finally, after training, AMOSR is implemented using the class centers of known classes. Experiments on three automatic modulation datasets show that our model has better average performance than several mainstream methods in the field of computer vision.

Abstract Integrated sensing and communication (ISAC) is a promising technique to increase spectral efficiency and support various emerging applications by sharing the spectrum and hardware between these functionalities. However, the traditional ISAC schemes are highly dependent on the accurate mathematical model and suffer from the challenges of high complexity and poor performance in practical scenarios. Recently, artificial intelligence (AI) has emerged as a viable technique to address these issues due to its powerful learning capabilities, satisfactory generalization capability, fast inference speed, and high adaptability for dynamic environments, facilitating a system design shift from model-driven to data-driven. Intelligent ISAC, which integrates AI into ISAC, has been a hot topic that has attracted many researchers to investigate. In this paper, we provide a comprehensive overview of intelligent ISAC, including its motivation, typical applications, recent trends, and challenges. In particular, we first introduce the basic principle of ISAC, followed by its key techniques. Then, an overview of AI and a comparison between model-based and AI-based methods for ISAC are provided. Furthermore, the typical applications of AI in ISAC and the recent trends for AI-enabled ISAC are reviewed. Finally, the future research issues and challenges of intelligent ISAC are discussed.

Understanding the enigmatic black-box representations within Deep Neural Networks (DNNs) is an essential problem in the community of deep learning . An initial step towards tackling this conundrum lies in quantifying the degree of similarity between these representations. Various approaches have been proposed in prior research, however, as the field of representation similarity continues to develop, existing metrics are not compatible with each other and struggling to meet the evolving demands. To address this, we propose a comprehensive similarity measurement framework inspired by the natural graph structure formed by samples and their corresponding features within the neural network . Our novel Graph-Based Similarity (GBS) framework gauges the similarity of DNN representations by constructing a weighted, undirected graph based on the output of hidden layers. In this graph, each node represents an input sample, and the edges are weighted in accordance with the similarity between pairs of nodes. Consequently, the measure of representational similarity can be derived through graph similarity metrics, such as layer similarity. We observe that input samples belonging to the same category exhibit dense interconnections within the deep layers of the DNN. To quantify this phenomenon, we employ a motif-based approach to gauge the extent of these interconnections. This serves as a metric to evaluate whether the representation derived from one model can be accurately classified by another. Experimental results show that GBS gets state-of-the-art performance in the sanity check. We also extensively evaluate GBS on downstream tasks to demonstrate its effectiveness, including measuring the transferability of pretrained models and model pruning.

Accurate and efficient positioning in complex environments remains a critical challenge where satellite-based systems (e.g., GNSS) suffer from signal attenuation and multipath interference. This paper proposes WK-Pnet, a lightweight positioning framework that utilizes frequency modulation (FM) signals and integrates Wavelet Packet Decomposition (WPD) with knowledge distillation. WK-Pnet first decomposes raw FM IQ signals using WPD to extract fine-grained multi-scale time-frequency features, preserving both spectral and phase information. These features are then fed into a deep neural network for location estimation. To reduce computational complexity, we employ a knowledge distillation strategy that transfers knowledge from a large-capacity ResNeXt-based teacher model—enhanced with a spatial attention mechanism—to a compact student network with significantly fewer parameters and FLOPs. The proposed method is validated on publicly available indoor and outdoor datasets, showing that WK-Pnet achieves comparable positioning accuracy to the teacher model while reducing FLOPs by 95.9%, model parameters by 99.3%, and inference latency by 90.5% on edge devices. Experimental comparisons also reveal that WPD outperforms STFT and EMD in positioning stability and accuracy, especially in outdoor scenarios. WK-Pnet demonstrates strong robustness, low-latency inference, and high accuracy, making it highly suitable for real-time, resource-constrained mobile and IoT applications.

Graph neural networks (GNNs) have achieved remarkable advances in graph-oriented tasks. However, real-world graphs invariably contain a certain proportion of heterophilous nodes, challenging the homophily assumption of traditional GNNs and hindering their performance. Most existing studies continue to design generic models with shared weights between heterophilous and homophilous nodes. Despite the incorporation of high-order messages or multi-channel architectures, these efforts often fall short. A minority of studies attempt to train different node groups separately but suffer from inappropriate separation metrics and low efficiency. In this paper, we first propose a new metric, termed Neighborhood Confusion (NC), to facilitate a more reliable separation of nodes. We observe that node groups with different levels of NC values exhibit certain differences in intra-group accuracy and visualized embeddings. These pave the way for Neighborhood Confusion-guided Graph Convolutional Network (NCGCN), in which nodes are grouped by their NC values and accept intra-group weight sharing and message passing. Extensive experiments on both homophilous and heterophilous benchmarks demonstrate that our framework can effectively separate nodes and yield significant performance improvement compared to the latest methods.

Automatic Modulation Open Set Recognition (AMOSR) has practical significance in detecting unknown classes. However, a challenge arises when unknown samples closely resemble known samples, posing a formidable task for accurate detection. A novel AMOSR framework based on multi-view discriminators’ joint judgment is proposed to handle this challenge. Firstly, utilizing signal domain knowledge, multi-dimensional features are extracted through varied signal time-frequency transforms and encoders, baesd on which multiple discriminators are created. Secondly, Constrained Clustering Prototype Loss and Geodesic Contrastive Loss are introduced to pretrain these discriminators, providing more space for unknown signals. Then, collaborative learning is employed to further fine-tune the aforementioned discriminators, enhancing information sharing between modalities. Furthermore, a set of indicators is constructed, and multi-criteria fusion is performed using the TOPSIS algorithm to evaluate the discrimination capabilities of different classifiers in both closed-set and open-set scenarios. Furthermore, a decision tree is constructed to segregate test signals into known and unknown classes, in which discriminators with higher confidence levels are given precedence. Finally, TOPSIS hierarchical ensemble pruning algorithm that considers diversity and open-set recognition capabilities is adopted to reduce model complexity while maintaining original performance. Extensive experiments conducted on modulation datasets demonstrate the superiority of this framework over state-of-the-art AMOSR results.

Due to the decentralized and public nature of the blockchain ecosystem, malicious activities on the Ethereum platform impose immeasurable losses on users. At the same time, the transparency of cryptocurrency transactions provides a unique opportunity to analyze illegal activities, such as phishing scams, from a network perspective. Most existing phishing scam detection methods focus primarily on analyzing account interaction networks, which limits their ability to uncover transaction behavior patterns embedded within transaction interactions. To address this, we construct theTransactionSubGraphNetwork (TSGN) by using transaction subgraphs as basic elements and further propose a novel framework for Ethereum phishing account detection. Specifically, we rebuild the graph structures via three well-designed mapping mechanisms, yielding TSGN and its two variants, i.e., Directed-TSGN and Temporal-TSGN, to obtain direction-aware and time-aware transfer flow features. By further incorporating the mapping strategy into transaction multidigraphs, we develop the Multiple-TSGN, which could preserve more transaction flow features while concurrently reducing the time consumption of modeling large-scale networks. TSGN models based on transaction subgraph interactions can capture complex higher-order dependencies, which lay beyond the reach of models that exclusively capture pairwise account interactions. As a general framework, our model can incorporate various feature extraction methods to improve the performance of phishing detection. Extensive experimental results on Ethereum datasets show that our method achieves superior performance in phishing detection, yielding 3.27%$\sim$6.71% relative improvement over previous state-of-the-art.

The rampant fraudulent activities on Ethereum hinder the healthy development of the blockchain ecosystem, necessitating the reinforcement of regulations. However, multiple imbalances involving account interaction frequencies and interaction types in the Ethereum transaction environment pose significant challenges to data mining-based fraud detection research. To address this, we first propose the concept of meta-interactions to refine interaction behaviors in Ethereum, and based on this, we present a dual self-supervision enhanced Ethereum fraud detection framework, named Meta-IFD. This framework initially introduces a generative self-supervision mechanism to augment the interaction features of accounts, followed by a contrastive self-supervision mechanism to differentiate various behavior patterns, and ultimately characterizes the behavioral representations of accounts and mines potential fraud risks through multi-view interaction feature learning. Extensive experiments on real Ethereum datasets demonstrate the effectiveness and superiority of our framework in detecting common Ethereum fraud behaviors such as Ponzi schemes and phishing scams. Additionally, the generative module can effectively alleviate the interaction distribution imbalance in Ethereum data, while the contrastive module significantly enhances the framework’s ability to distinguish different behavior patterns. The source code will be available inhttps://github.com/GISec-Team/Meta-IFD.

The data compression based transmission has been envisioned as a promising solution to improve the data transmission efficiency with the limited radio resources in the future sixth-generation (6G) wireless networks. In this paper, we propose an integrated communication and computation resource allocation system for image transmission based on compressive sensing (CS), which consists of several camera devices and a base station (BS). The device side first compresses the images, after which the compressed images are transmitted using non-orthogonal multiple access (NOMA) transmission, and finally the BS restores the received compressed images. Due to the limited energy supply, the total system energy consumption is minimized by jointly optimizing the image sampling rate, the image data transmission power, the number of floating point operations per second (FLOPS), the time of image compression and the time of data transmission under the constraints of latency and the peak signal-to-noise ratio (PSNR). Due to the non-convexity of the proposed problem, after a series of equal substitutions we convexify the problem. Then, the Karush-Kuhn-Tucker (KKT) condition and the gradient descent method are used to obtain the optimal solution of the target problem. After simulation experiments, it is concluded that the proposed CS-based image transmission scheme effectively reduces the total energy consumption by a factor of 2.7 compared with frequency division multiple access (FDMA), and the total latency by 180% compared with the original image transmission.

Satellite communication has emerged as a promising technology for achieving a wide range of communication coverage and providing a variety of services in the marine Internet of Things. This paper investigates the efficient data collection scheme of the low earth orbit (LEO) satellite from different sensing devices (SDs) deployed in the offshore areas. To be specific, these marine SDs in each time slot utilize the non-orthogonal multiple access (NOMA) to upload their respective sensing data to the LEO satellite passing over the relative areas. To ensure efficient data collection, we then aim to minimize the overall energy consumption needed to upload all sensing data from SDs to the LEO satellite subject to the minimum transmission latency. To tackle the proposed non-convex joint optimization problem, we designed an efficient algorithm based on successive convex approximation (SCA) to approach the optimal solutions. Finally, numerous results are presented to illustrate the convergence performance of the proposed SCA-based algorithm as well as the performance gains of the proposed scheme.

With the growing demand for image transmission, there is a need for solutions that offer low energy consumption and low latency. In this paper, we present a novel relay-assisted system based on non-orthogonal multiple access (NOMA) and mobile edge computing (MEC). Our proposed system compresses images at the device end, decompresses them at either a relay or a cloud server (CS). The primary objective is to minimize system energy consumption under given task delay constraints. Considering that this is a non-convex optimization problem, we solve it by decomposing it into a continuous subproblem and a discrete subproblem. To solve the continuous subproblem, we convexify it by introducing new parameters and change variables to get the optimal the computing power of devices, relay and CS, sampling rate of devices, transmission power of devices and relay. To solve the discrete subproblem, we propose a cross-entropy (CE) algorithm to obtain the optimal decompression decision and subcarrier allocation decision. Simulation results demonstrate the accuracy and effectiveness of our algorithm in optimizing total energy consumption compared to the Linear Interactive and General Optimizer (LINGO) and frequency division multiple access (FDMA) methods.

Along with the prevalent use of deep neural networks (DNNs), concerns have been raised on the security threats from DNNs such as backdoors in the network. While neural network repair methods have shown to be effective for fixing the defects in DNNs, they have been also found to produce biased models, with imbalanced accuracy across different classes, or weakened adversarial robustness, allowing malicious attackers to trick the model by adding small perturbations. To address these challenges, we propose INNER, an INterpretability-based NEural Repair approach. INNER formulates the idea of neuron routing for identifying fault neurons, in which the interpretability technique model probe is used to evaluate each neuron's contribution to the undesired behaviour of the neural network. INNER then optimizes the identified neurons for repairing the neural network. We test INNER on three typical application scenarios, including backdoor attacks, adversarial attacks, and wrong predictions. Our experimental results demonstrate that INNER can effectively repair neural networks, by ensuring accuracy, fairness, and robustness. Moreover, the performance of other repair methods can be also improved by re-using the fault neurons found by INNER, justifying the generality of the proposed approach.

Digital twin has been emerging as a promising paradigm that connects physical entities and digital space, and continuously evolves to optimize the physical systems. In this article, we focus on studying efficient communication and computation scheme when constructing the Marine Internet of Things (M-IoT)’s digital twin with secrecy provisioning. Specifically, the digital twin model is trained based on federated learning, in which all the unmanned surface vehicles deliver the trained models with nonorthogonal multiple access (NOMA) to the high-altitude platform (HAP) for global model aggregation. Considering the potential eavesdropping on the radio signals of HAP, we utilize the chaotic sequences to spread the model information before the global model broadcasting. In this framework, we aim to minimize the total energy consumption for constructing the digital twin of M-IoT by jointly optimizing the global accuracy, the local accuracy, the HAP’s transmission power and NOMA transmission duration, subject to the secrecy provisioning and latency constraint. An effective low-complexity algorithm is proposed to tackle this joint optimization problem with the use of a layered feature. Finally, numerical results are given to validate the performance gain of the proposed scheme, in comparison with the fixed accuracy scheme, the nonspread spectrum scheme and the time division multiple access transmission scheme.

Cognitive radio-based Internet of Things (CR-IoTs) provide an efficient spectrum management for IoT networks with massive wireless access and data transmission needs. As one of the key technologies of CR-IoT, spectrum sensing is of great research significance. Motivated by the recent boom on applications of deep learning in wireless communications networks and IoT, several spectrum sensing methods based on deep learning have emerged. However these algorithms train the sensing models with the extracted features of received signals and require a retraining of sensing models when the number of sensing antennas changes. Thus, we develop multiple-antenna spectrum sensing methods based on convolutional neural networks (MASSnet) using the in-phase (I) and quadrature (Q) components of the signals as the input. The three schemes of MASSnet also provide the flexibility to choose between retraining the sensing models or using the obtained models for different sensing antenna configurations. Experiment results demonstrate the superior performance of the proposed methods over existing deep learning-based spectrum sensing methods in terms of probability of detection especially in very low signal-to-noise ratio (SNR) condition. Furthermore, the proposed methods have good generalization ability to new noise distribution, new fading channel, different frequency offsets, and detecting signals with a new modulation even without retraining.

Frequency Modulation (FM) broadcast signals, regarded as opportunistic signals, hold significant potential for indoor and outdoor positioning applications. The existing FM-based positioning methods primarily rely on Received Signal Strength (RSS) for positioning, the accuracy of which needs improvement. In this paper, we introduce FM-Pnet, an end-to-end FM-based positioning method that leverages deep learning. This method utilizes the time-frequency representation of FM signals as network input, enabling automatically learning of deep features for positioning. We also propose two strategies, noise injection and enriching training samples, to enhance the model’s generalization performance over long time spans. We construct datasets for both indoor and outdoor scenarios and conduct extensive experiments to validate the performance of our proposed method. Experimental results demonstrate that FM-Pnet significantly outperforms traditional RSS-based positioning methods in terms of both positioning accuracy and stability.

The Doppler dispersion of the received signal is very severe when the beam width of the antenna is wide, resulting in a decrease in localization accuracy for Multiple Radiation Source Localization. To resolve the problem, we propose a passive signal focusing algorithm (PSFA) for multiple radiation sources localization based on a full aperture model. The full-aperture model overcomes the resolution degradation in conventional sub-aperture processing. In the PSFA, the residual frequency correction (RFC) eliminates the localization bias in the azimuth domain and the instantaneous Doppler compensation (IDC) resolves the Doppler dispersion in the range domain, improving localization accuracy. The matched filtering is used to complete precise azimuthal focusing and the locations are obtained according to the focusing results. Moreover, the Cramer-Rao lower bound (CRLB) for synthetic aperture localization is derived. The CRLB is essential for evaluating algorithms in theoretical studies and designing system parameters in practical applications. Finally, The CRLB, simulation, and acquired data are used to evaluate the localization performance.

In satellite synthetic aperture positioning (SAP), the curvature of the Earth’s surface and the curved orbit lead to nonlinear and asymmetric instantaneous Doppler frequencies, especially when dealing with signals of very long durations. This phenomenon significantly affects the accuracy of center frequency estimation and radiating source positioning. This study presents an ultra-high-resolution positioning algorithm designed to process ultra-long-duration data collected by a single satellite. Initially, a method for estimating the zero-Doppler moment based on sub-aperture chirp rates is proposed to obtain an unbiased estimate of the radiation source’s center frequency. Subsequently, a nonlinear instantaneous Doppler compensation method is proposed, utilizing the estimated center frequency and chirp rates to enhance the coherence of the long-duration data. Furthermore, a long coherent positioning is suggested to generate an ultra-high-resolution positioning image. Ultimately, the efficacy of the proposed algorithm is validated through simulations and acquired data.

Automatic modulation recognition (AMR) of radio signal is an important research topic in the area of non-cooperative communication and cognitive radio. Recently deep learning (DL) techniques enable significant progress in AMR. However, the techniques of adversarial machine learning cause the threats of adversarial attacks in DL-based AMR. In this paper, we aim to make AMR model robust, accurate and lightweight, thus propose a multi-distillation mechanism for robust training of DL-based AMR models, namely Adversarial Multi-Distillation (AMD). In the framework of AMD, by knowledge distillation, two powerful teacher models transfer the learned classification knowledge and defense knowledge, respectively, to the student model to form robust training. Our experiments with public dataset RML2016.10a show that the proposed method can significantly improve the defense of AMR models to against adversarial perturbations and keep relatively high classification accuracy, which enables robust decision making with lightweight models under adversarial attacks.

Deep learning technology has found a promising application in lightweight model design, for which pruning is an effective means of achieving a large reduction in both model parameters and float points operations (FLOPs). The existing neural network pruning methods mostly start from the consideration of the importance of model parameters and design parameter evaluation metrics to perform parameter pruning iteratively. These methods were not studied from the perspective of network model topology, so they might be effective but not efficient, and they require completely different pruning for different datasets. In this article, we study the graph structure of the neural network and propose a regular graph pruning (RGP) method to perform a one-shot neural network pruning. Specifically, we first generate a regular graph and set its node-degree values to meet the preset pruning ratio. Then, we reduce the average shortest path-length (ASPL) of the graph by swapping edges to obtain the optimal edge distribution. Finally, we map the obtained graph to a neural network structure to realize pruning. Our experiments demonstrate that the ASPL of the graph is negatively correlated with the classification accuracy of the neural network and that RGP has a strong precision retention capability with high parameter reduction (more than 90%) and FLOPs reduction (more than 90%) (the code for quick use and reproduction is available at https://github.com/Holidays1999/Neural-Network-Pruning-through-its-RegularGraph-Structure).

A wireless communications system usually consists of a transmitter which transmits the information and a receiver which recovers the original information from the received distorted signal. Deep learning (DL) has been used to improve the performance of the receiver in complicated channel environments and state-of-the-art (SOTA) performance has been achieved. However, its robustness has not been investigated. In order to evaluate the robustness of DL-based information recovery models under adversarial circumstances, we investigate adversarial attacks on the SOTA DL-based information recovery model, i.e., DeepReceiver. We formulate the problem as an optimization problem with power and peak-to-average power ratio (PAPR) constraints. We design different adversarial attack methods according to the adversary’s knowledge of DeepReceiver’s model and/or testing samples. Extensive experiments show that the DeepReceiver is vulnerable to the designed attack methods in all of the considered scenarios. Even in the scenario of both model and test sample restricted, the adversary can attack the DeepReceiver and increase its bit error rate (BER) above 10%. It can also be found that the DeepReceiver is vulnerable to adversarial perturbations even with very low power and limited PAPR. These results suggest that defense measures should be taken to enhance the robustness of DeepReceiver.

Unmanned aerial vehicle (UAV)-enabled mobile edge computing (MEC) can provide flexible computing service for terminal-devices (TDs). However, malicious active aerial eavesdroppers can perform air-to-ground eavesdropping and air-to-air attacking, which makes TDs’ tasks offloading computation more vulnerable, posing significantly secure threats to UAV-enabled MEC. To overcome this challenge, we aim to design collaborative communication and computation schemes for the secure UAV-enabled MEC system, where an active aerial eavesdropper is capable of wiretapping the tasks information offloaded from TDs and transmitting attack signals to the legitimate network. The total weighted energy consumption of the system is minimized via optimizing time allocation, transmit power, local and offloading computation bits, as well as UAV trajectory. First, considering the given number of computational tasks of TDs, a block coordinate descent (BCD)-based scheme is proposed to decompose the original multi-variables-coupling and close-form-lacking problem into several tractable subproblems that can be addressed by iterations. Next, considering that there are dynamic and random tasks arriving to TDs’ original tasks, a deep reinforcement learning (DRL)-based scheme is proposed to maintain the stability of tasks, where the solution of computation, communication and trajectory optimization is intelligently obtained by adopting double-deep Q-learning (DDQN). Simulation results demonstrate that the proposed schemes outperform the respective benchmarks for secure UAV-enabled MEC against active aerial eavesdropping.

Deep learning has been widely used in automatic modulation classification (AMC) recently. Most of deep learning-based AMC uses a single network model to deal with radio signals with a single input format. In this paper, we propose a hybrid heterogeneous modulation classification architecture named DeepSIG, which integrates Recurrent Neural Network (RNN), Convolutional Neural Network (CNN) and Graph Neural Network (GNN) models in a single framework to process radio signals with heterogeneous input formats, i.e., in-phase (I) and quadrature (Q) sequences, images mapped from IQ signals and graphs converted from IQ signals, to extract and integrate the features from different perspectives. A fusion training mechanism is presented to train DeepSIG. We use three different radio signal datasets for simulations. Results show that our proposed DeepSIG performs the best in terms of classification accuracy compared with the three methods with single input, i.e., sequence, image or graph. The performance gain is larger in few-shot scenarios.

Mobile edge computing has been a promising technology that enables diverse applications of computation-intensive yet latency-sensitive in marine Internet of Things networks. In this paper, we propose a framework of hierarchical computing offloading with the assistance of high altitude platforms (HAPs), and a hybrid transmission scheme of non-orthogonal multiple access (NOMA) and frequency division multiple access (FDMA) is designed for achieving efficient computation offloading. Specifically, the offshore sensing devices (SDs) initially perform computation offloading to the HAPs by forming NOMA groups, and the HAPs further offload partial workload to the onshore base station (BS) via FDMA. For efficient calculation, we aim to minimize the overall delay in completing all the workload processing of these SDs by jointly optimizing the durations of NOMA and FDMA transmission as well as the hierarchical computation offloading workload. Though the problem is in the form of non-convexity, we design an efficient SCA-based algorithm to tackle it. Finally, numerical results demonstrate the optimality and convergence of the proposed algorithm, as well as the performance gains of the proposed scheme.

Internet of Things (IoT) is one of the key applications of 5G, and the data transmission is the basis of IoT networks. In this paper, we investigate the data transmission scheme in non-orthogonal multiple access (NOMA) for IoT networks to minimize the transmission latency. In order to improve the communication efficiency between devices and the base station (BS) without more energy consumption, we deploy an energy harvesting (EH) relay node between devices and the BS for data transmitting and forwarding. Based on this networking model, we first aim at minimizing the transmission latency by jointly optimizing the transmit power of devices and the relay, forwarding ratios among devices, and forwarding time fraction when transmitting a fixed data bits from devices to the BS via the relay under the constraints of energy buffer and data buffer. Noted that the formulated problem is discrete-continuous mixed and non-convex, we apply the deep deterministic policy gradient (DDPG) algorithm in the framework of bisection searching to obtain the optimal solution. Specifically, the bisection searching is used to seek the possible transmission latency, and the DDPG is to check the feasibility of the chosen transmission latency. Finally, the effectiveness of the proposed model-data-driven algorithm is verified by comparing it with other benchmark algorithms, such as LINGO.

Digital twin has been emerging as a promising paradigm that connects the physical entities and digital space, and continuously evolves and optimizes the physical systems. In this paper, we focus on studying the efficient data communication and computation when constructing the marine digital twin network with secrecy provisioning. Specifically, we leverage the federated learning (FL) to train the digital twin model. In the process of FL, all unmanned surface vehicles (USVs) deliver the trained models with non-orthogonal multiple access (NOMA) to the high altitude platform (HAP) for the global model aggregation. Considering the possible eavesdropping on the HAP, we utilize the chaotic sequences to spread the model information during the global model broadcasting. In this framework, we further want to minimize the total energy consumption of completing the digital twin training by jointly optimizing the global accuracy, local accuracy, HAP's transmission power, and model uploading duration subject to the secrecy provisioning and latency constraint. Despite the non-convexity, we propose a low-complexity search algorithm (LCS-Algorithm) to solve this joint optimization problem. Finally, the numerical results validate the performance of the proposed algorithm in terms of optimality and time efficiency.

Automatic modulation classification (AMC) is an important technology for the monitoring, management, and control of communication systems. In recent years, machine learning approaches are becoming popular to improve the effectiveness of AMC for radio signals. However, the automatic modulation open-set recognition (AMOSR) scheme that aims to identify the known modulation types and recognize the unknown modulation signals is not well studied. Therefore, in this paper, we propose a novel multi-modal marginal prototype framework for radio frequency (RF) signals (MMPRF) to improve AMOSR performance. First, MMPRF addresses the problem of simultaneous recognition of closed and open sets by partitioning the feature space in the way of one versus other and marginal restrictions. Second, we exploit the wireless signal domain knowledge to extract a series of signal-related features to enhance the AMOSR capability. In addition, we propose a GAN-based unknown sample generation strategy to allow the model to understand the unknown world. Finally, we conduct extensive experiments on several publicly available radio modulation data, and experimental results show that our proposed MMPRF outperforms the state-of-the-art AMOSR methods.

Although deep learning methods have made significant advancements across various domains, recent research has shown that carefully crafted adversarial samples can lead to a significant degradation in the performance of deep learning models. Such adversarial examples raise concerns about the reliability and safety of deep learning-based models. Currently, there is a lack of research on the robustness of deep learning based DOA methods against adversarial samples. This letter aims to fill this research gap by leveraging the differentiability of the transformation process from the original signal to the covariance matrix. By utilizing this differentiability, the robustness of the DOA estimation model, which takes the covariance matrix as input, is investigated. Four different white-box attack methods are considered to generate adversarial samples to evaluate the resilience of the model. The experimental results demonstrate that all four methods employed significantly increase the estimation error of the DOA estimation model, posing a serious threat to the model's security.

Edge computing has been envisioned as a promising approach to enable the computation-intensive yet latencysensitive marine mobile services in the fifth generation and beyond wireless networks. In this paper, we investigate the edge computing in Marine Internet of Things (M-IoT) via the assistance of unmanned surface vehicles (USVs) subject to the eavesdropping attack. In particular, we consider a scenario in which USVs are exploited to provide cooperative jamming for the communication security at the physical layer when the high altitude platform (HAP) is performing task offloading transmission. We jointly optimize the workload offloaded by HAP, the HAP's transmission power as well as each USV's interfering signal power with the objective of minimizing the total energy consumption for completing the total workloads under the latency constraint. The bisection search method is first adopted to obtain the optimal solutions to the offloaded workload and each USV's interfering signal power. Further, by exploiting the monotonicity, the polyblock outer approximation based algorithm (POA-Algorithm) is designed to obtain the HAP's optimal transmission power. Finally, numerical results validate the optimality and effectiveness of our proposed algorithm by comparing it with the results of LINGO and different jamming schemes.

Unmanned aerial vehicle-aided (UAV-aided) mobile edge computing (MEC) network can greatly reduce the data growth pressure of Internet of Things (IoT) and expand the wireless communication coverage. However, there is a risk of eavesdropping on the offloading information of terminal users (TUs) because of UAV light-of-sight (LoS) transmission. In this paper, we propose a Dinkelbach-guided deep reinforcement learning (DRL) scheme for secure communication in the UAV-aided MEC network. Specifically, the security calculating efficiency of the network is maximized by optimizing offloading decision and resource allocation under the condition of the data queue stability and minimum calculating requirement. The problem is intractable due to the fractional structure and binary constraint. Firstly, we deal with the fractional structure by taking advantage of Dinkelbach optimization. Then, offloading decision is generated based on DRL and the resource is allocated by successive convex approximation (SCA). Simulation results show that the proposed Dinkelbach-guided DRL scheme efficiently improves the security calculating efficiency of the network.

With the rapid development of wireless communications, it is challenging to guarantee secure wireless transmission and massive connectivity in the process of data collection. In this paper, we consider an unmanned aerial vehicle (UAV)-aided Non-orthogonal Multiple Access (NOMA) communication network. Specifically, the UAV is deployed to collect the data of transmission devices (TDs) in the NOMA manner subject to the eavesdropping attack, while a group of auxiliary devices (ADs) are deployed to provide the cooperative jamming to the eaves-dropper. Driven by this networking model, we aim to maximize the total secrecy capacity by jointly optimizing the TDs’ and ADs’ power allocations and the ADs’ scheduling decisions. Considering the problem’s non-convexity, we propose a deep reinforcement learning based online optimization algorithm to maximize the total secrecy capacity. Numerical results demonstrate that the proposed algorithm can achieve considerable performance gain over some existing algorithms.

In cognitive radio networks, providing accurate recognition of the primary user’s signal is great important for designing the spectrum access strategies. Source number estimation has served as a key fundamental technique to facilitate the signal recognition in the mixed received signals scenario. In this paper, we propose a deep learning based source number estimation method under the single-channel conditions. The architecture of the network is first designed. Then, the received complex signals are reconstructed as in-phase and quadrature (IQ) data in order to adapt to the convolutional neural network for extracting the deep features. Moreover, the cost function that is used to train the proposed network is properly designed by exploring the maximum likelihood function. Supervised training is performed to generate a classifier that can identify the number of sources in the mixtures. The proposed deep leaning-based source number estimation method is tested by experiments on simulation signals and actual signals. The results revealed the effectiveness of the proposed method in solving the problem of single-channel source number estimation compared to the conventional methods.

Frequency hopping (FH) signal is one of the research hotspots of passive positioning. Aiming at the problem of FH signal localization, this paper proposes a synthetic aperture passive positioning method. The method estimates and compensates for the baseband modulation of the received signal. Then the received signal vectors are arranged into a two-dimensional matrix. The Doppler frequency of each pulse is compensated by the Doppler frequency compensate matrix. The cost function is constructed by a two-dimensional focus of the received signal, and the emitter position is directly obtained through a gird search. Simulation and experimental data verify the effectiveness of the proposed method.

The nonlinear variation of viewing angles over a long duration causes a nonlinear initial phase of the received pulse in a passive positioning system with a single moving receiver. Typical positioning systems ignore the phase and perform incoherent accumulation of the long-time data, resulting in a decrease in positioning accuracy, especially at a low signal-to-noise ratio (SNR). A novel passive positioning system with a synthetic aperture technique, named synthetic aperture positioning (SAP) system, is proposed to resolve the issue. First, a new 2-dimensional (2-D) continuous sampling working model is proposed. Then, the SAP system and a cost function are given to analyze the positioning performance. Third, a positioning algorithm based on the maximum likelihood estimation (MLE) is studied to handle the cost function and position the emitter. Simulation and experimental results verify the validity and effectiveness of the proposed SAP system.

Deep Neural Networks (DNN) are known to be vulnerable to adversarial samples, the detection of which is crucial for the wide application of these DNN models. Recently, a number of deep testing methods in software engineering were proposed to find the vulnerability of DNN systems, and one of them, i.e., Model Mutation Testing (MMT), was used to successfully detect various adversarial samples generated by different kinds of adversarial attacks. However, the mutated models in MMT are always huge in number (e.g., over 100 models) and lack diversity (e.g., can be easily circumvented by high-confidence adversarial samples), which makes it less efficient in real applications and less effective in detecting high-confidence adversarial samples. In this study, we propose Graph-Guided Testing (GGT) for adversarial sample detection to overcome these aforementioned challenges. GGT generates pruned models with the guide of graph characteristics, each of them has only about 5% parameters of the mutated model in MMT, and graph guided models have higher diversity. The initial experiments on CIFAR10 validate that GGT performs much better than MMT with respect to both effectiveness and efficiency.

Community detection plays an important role in social networks, since it can help to naturally divide the network into smaller parts so as to simplify network analysis. However, on the other hand, it arises the concern that individual information may be overmined, and the concept community deception has been proposed to protect individual privacy on social networks. Here, we introduce and formalize the problem of community detection attack and develop efficient strategies to attack community detection algorithms by rewiring a small number of connections, leading to privacy protection. In particular, we first give two heuristic attack strategies, i.e., Community Detection Attack (CDA) and Degree Based Attack (DBA), as baselines, utilizing the information of detected community structure and node degree, respectively. Then, we propose an attack strategy called “genetic algorithm (GA)-based Q-Attack,” where the modularity Q is used to design the fitness function. We launch community detection attack based on the above three strategies against six community detection algorithms on several social networks. By comparison, our Q-Attack method achieves much better attack effects than CDA and DBA, in terms of the larger reduction of both modularity Q and normalized mutual information (NMI). In addition, we further take transferability tests and find that adversarial networks obtained by Q-Attack on a specific community detection algorithm also show considerable attack effects while generalized to other algorithms.

A cross-network cross-layer design method is proposed to exploit the trunking, diversity, and best service assignment gains available in a heterogeneous wireless network (HWN), consisting of orthogonal radio access networks (RANs) and interference-limited RANs. Accounting for traffic-level dynamics and channel fading, we jointly design the distribution strategy for elastic and inelastic traffic, and the radio resource management strategy for RANs, in a network-separable control architecture. Optimal and quantified near-optimal radio allocation schemes are proposed for each type of RAN, which are combined into an on-line design framework that over time provides asymptotically optimal performance, maximizing the sum throughput utility for elastic traffic while guaranteeing the throughput requirements of inelastic traffic. Extensive simulation results demonstrate substantial performance improvement against suboptimal alternatives.

Heterogeneous networks are an attractive means of improving the network capacity. Heterogeneous network are typically composed of multiple radio access points (macro, pico and femto) where the base stations are transmitting with variable power. In this study, we consider the uplink of heterogeneous networks, where each terminal can simultaneously connected to multiple radio access points and requests delay-sensitive traffic (e.g., real-time video). We adopt the framework of Hou, Borkar,and Kumar, and study the delay in heterogeneous networks and analyze the multiaccess transmission delay by considering three extreme scheduling schemes. Moreover, a parallel-aggregation multiple radio access points (PA-MRA) algorithm for packet scheduling is presented to improve the transmission delay in heterogeneous networks. Analysis and numerical results show that the proposed algorithm obtains the optimal and robust transmission delay gain compared with the three extreme scheduling policies.

In this paper, we discuss how to exploit the multiple access points diversity gain in the downlink transmission of a multi-access wireless network. To follow the current user equipment hardware constraint as well as to reduce the network control overhead, we confine that each user equipment can only be serviced by one best suited access point. To achieve this, we formulate such problem as a non-linear integer program, which typically is computational intractable. To derive its solution, we first convert it to a linear integer program, which is then used to obtain a fractional solution, and further we refine the fractional solution to be integer to satisfy integer constraints for the access point selection and the resource unit allocation. By doing so, our algorithm has a low computational complexity. We also show the performance improvement of our algorithm over the distance based AP selection and the random AP selection methods through simulation.

A dynamic optimization algorithm is proposed for the joint allocation of subframes, resource blocks, and power in the Type 1 inband relaying scheme mandatory in the LTE-Advanced standard. Following the general framework of Lyapunov optimization, we decompose the original problem into three sub-problems in the forms of convex programming, linear programming, and mixed-integer programming. We solve the last sub-problem in the Lagrange dual domain, showing that it has zero duality gap, and that a primal optimum can be obtained with probability one. The proposed algorithm dynamically adapts to traffic and channel fluctuations, it accommodates both instantaneous and average power constraints, and it obtains arbitrarily near-optimal sum utility of each user's average throughput. Simulation results demonstrate that the joint optimum can significantly outperform suboptimal alternatives.

This paper deals with the blind source separation (BSS) problem in the overdetermined case, i.e. the number of receive antennas is larger than that of sources. First, a method, which converts the over-determined BSS into complete BSS, is presented. Then a maximum ratio combining (MRC) based BSS algorithm (MRC-FastICA) is proposed. The proposed algorithm includes source separation, estimated signals classification and MRC of multiple estimated signals corresponding to a source signal. Theoretical analysis and simulation results show that the proposed algorithm can obtain diversity gain, and has better performance than traditional methods.

Network-assisted diversity multiple access (NDMA) was recently developed to resolve collision and achieve high throughput. The blind NDMA method using independent component analysis (ICA), named ICA-NDMA, has been shown to overcome the difficulties of synchronization and orthogonal identification codes required by the training-based NDMA protocols. In this paper, we propose a novel blind collision resolution strategy by employing cooperative transmission in ICA-NDMA. When K-node collision happened, a cooperative transmission mechanism is adopted to collect different superposition of the K nodes' packets in the following slots. Moreover, a cooperative blind detection algorithm is intro- duced to determine the number of collision nodes. Finally, with ICA, the destination can retrieve the original packets through processing the collided packet and the signals forwarded by relays. Simulation results show that, the proposed method can work effectively under both fast-varying and slow-varying channels, the performances of average packet delay and maximum stable throughput are better than ICA-NDMA.