Indoor Internet of Things (IoT) applications, such as smart buildings and office automation, often require dense node deployments with power constraints. Reliable connections between gateways (GWs) and end nodes (ENs) rely on accurate knowledge of the path loss (PL). In this paper, we present a coordinate-attention (CA)-based scheme for PL prediction in indoor IoT scenarios. Within this scheme, a network termed CATransPropa integrates a CA module into the convolutional neural network (CNN) to capture the spatial distribution of indoor obstacles, aiming to enhance prediction accuracy. The network utilizes a multimodal input that combines environmental images and propagation features, including indoor-specific characteristics such as wall penetration thickness. Measurements are carried out in an indoor scenario at 433 MHz to verify the performance of the proposed scheme. It is shown that our scheme achieves a root mean square error (RMSE) value of 3.62 dB between the predicted and actual PL, outperforming the compared methods.

The performance of the sixth-generation (6G) vehicle-to-vehicle (V2V) communication systems will be significantly improved, but they are also confronted with many technical challenges like massive terminal access and low transmission delay. A fundamental and difficult problem is how to establish an intelligent 6G V2V channel model with high accuracy, low complexity, and generality. In this paper, we propose a dynamic V2V channel model in complicated urban scenarios driven by effective environment information, channel data, and physical statistics. To begin with, the bimodal features representing the environment information are extracted from vector maps by a set of fully automatic algorithms. Heuristic graph datasets are constructed using features coupled with locations and ground-truth large-scale parameters (LSPs), i.e., the channel data reflecting realistic statistical properties. Then, we design a novel network based on attention-assisted graph convolution and pooling layers, which enables us to perform prediction for path loss, delay spread, and angular spreads. Compared with convolutional neural networks-based methods, the proposed LSPs prediction model can reduce both the number of trainable parameters and the FLOPs by two orders of magnitude with higher accuracy. Moreover, the predicted LSPs are next fed into multi-link V2V simulations based on physical statistics. Dynamic channel impulse response generation is implemented based on a spatially consistent geometrical modeling methodology. Eventually, we validate our model by comparing key channel characteristics with those of the ground-truth values, and better agreements are shown compared with existing methods.

Radio frequency fingerprint identification (RFFI) is a critical technique for ensuring the stability and security of Internet of Things (IoT) networks by exploiting unique hardware-induced characteristics of received signals. However, the extraction of hardware-induced characteristics from the transmitter is frequently obfuscated by the receivers’ hardware impairments. This challenge is particularly pronounced in low signal-to-noise ratio (SNR) conditions, where it significantly diminishes the accuracy of RFFI systems. To address this issue, we propose a novel receiver-agnostic RFFI approach to mitigate the impact of noise and enhance the extraction of transmitter features. The proposed approach uses demodulation and reconstruction techniques and employs adversarial training to achieve noise-robust feature extraction. An adaptive noise mitigation mechanism is designed to further reduce the interference resulting from noise. The results demonstrate that the accuracy of the proposed approach is 38.4% higher than the comparative methods at 0 dB SNR. Furthermore, a comparative experiment confirms the individual advantages of each component, validating the effectiveness of our proposed approach.

Long-range wide-area network (LoRaWAN) is a widely used technology in the Internet of Things (IoT), which provides long-range (LoRa) communication with low power consumption. In LoRaWAN, an accurate path loss (PL) model is essential to realize link budget and network coverage planning. In this article, we present an artificial intelligence (AI)-enhanced generalizable scheme for PL prediction in LoRaWAN. We propose a network that performs corrective adjustments to improve the PL estimates of empirical models. The network termed STransRadio benefits from the self-attention computation in Swin Transformer to model the LoRa correlation about propagation for enhancing the adjustment prediction accuracy. To generalize our scheme to new scenarios, an multiscenario deep transfer learning (MDTL) algorithm is proposed, which finetunes the pretrained STransRadio network with limited data. We conduct simulations and measurements in the 868-MHz bands to assess the performance of the scheme in terms of prediction accuracy and generalization ability. The effectiveness of the proposed scheme has been verified with both simulations and measurements. Moreover, the STransRadio network in the scheme outperforms the convolutional neural network (CNN) and deep vision transformer (DeepViT). With the MDTL algorithm, our scheme can achieve excellent prediction performances when it is applied in a new scenario with limited training data. Furthermore, we verify that the scheme utilized in the simulated scenario can be transferred to both the new simulated scenario and the realistic scenario. With only 100 samples, the scheme achieves root mean square error (RMSE) values of 7.27 and 5.96 dB between the predicted and actual PL, respectively.

Non-reciprocal antenna beam patterns are promising to be utilized in asymmetric massive multiple-input multiple-output (MIMO) systems for future sixth-generation communications. The inconsistency of uplink (UL) and downlink (DL) channels makes channel modeling in this scenario challenging. In this paper, a novel geometry-based stochastic model (GBSM) is proposed for non-reciprocal terahertz (THz) channels. A directional effective scatterer generation algorithm is designed to depict the inconsistency of bidirectional propagation conditions. The correlation function between UL and DL is derived and analyzed, which validates the ability to characterize the non-reciprocal channels. To mimic THz propagation features, molecular absorption and diffuse scattering are introduced to the model, which is verified by measured data. In addition, the non-stationarities in space, time, and frequency domains are characterized, respectively. Statistical properties are compared between analytical and simulation results, and good agreements are shown. Finally, the accuracy of the model is verified by comparing with the ray tracing data.

Automatic modulation recognition (AMR) plays an important role in wireless communication system monitoring, non-cooperative communications, and cognitive communications. Recently, the applications of deep learning in AMR improve classification accuracy. However, it is difficult to deploy a deep learning-based model on resource-constrained devices because of its huge model size. In this paper, we propose a neural network called double pooling convolutional neural network (DP-CNN) and a stochastic pruning-asymmetric quantization (SPAQ) algorithm to realize lightweight and accurate modulation recognition. With the SPAQ algorithm, unimportant parameters are pruned by designing probability intervals and evaluation criteria. In addition, the storage type of parameters will be transformed by creating quantization intervals and mapping criteria. The performance of our method is verified using an open-source dataset RadioML2016.10a. Experimental results show that the SPAQ algorithm has better recognition performance than other lightweight methods at high compression ratios. In addition, the DP-CNN compressed by the SPAQ algorithm outperforms the existing lightweight network in recognition accuracy under the same model size.

Automatic modulation recognition (AMR) plays an essential role in intelligent communication networks monitoring, management, and optimization. Recently, it has been shown that deep learning-based methods perform well in AMR. However, most existing methods are based on real-valued networks, e.g., convolution neural network (CNN), which are not specifically designed for AMR. Thus, the recognition performance is limited. In this paper, we propose a selective kernel fusion complex-valued convolution neural network (SKF-CCNN) for the fulfillment of the AMR task. The proposed method uses parallel complex-valued convolution for raw in-phase/quadrature (I/Q) sequence together with real-valued convolution for amplitude. The complex-valued features and amplitude features are then fused by a selective kernel block to combine all sorts of information. Lastly, a ResNeXt block and two convolution layers are employed to extract further information from the fused feature map for the task of the final classification. Experiments on the benchmark dataset show that the proposed method outperforms the existing complex-valued methods, especially at high SNRs.

Asymmetric massive multiple-input multiple-output (MIMO) systems have been proposed to reduce the burden of data processing and hardware cost in sixth-generation mobile networks (6G). However, in the asymmetric massive MIMO system, reciprocity between the uplink (UL) and downlink (DL) wireless channels is not valid. As a result, pilots are required to be sent by both the base station (BS) and user equipment (UE) to predict double-directional channels, which consumes more transmission and computational resources. In this paper we propose an ensemble-transfer-learning-based channel parameter prediction method for asymmetric massive MIMO systems. It can predict multiple DL channel parameters including path loss (PL), multipath number, delay spread (DS), and angular spread. Both the UL channel parameters and environment features are chosen to predict the DL parameters. Also, we propose a two-step feature selection algorithm based on the SHapley Additive exPlanations (SHAP) value and the minimum description length (MDL) criterion to reduce the computation complexity and negative impact on model accuracy caused by weakly correlated or uncorrelated features. In addition, the instance transfer method is introduced to support the prediction model in new propagation conditions, where it is difficult to collect enough training data in a short time. Simulation results show that the proposed method is more accurate than the back propagation neural network (BPNN) and the 3GPP TR 38.901 channel model. Additionally, the proposed instance-transfer-based method outperforms the method without transfer learning in predicting DL parameters when the beamwidth or the communication sector changes.

In time-variant high-speed train (HST) radio channels, the scattering environment changes rapidly with the movement of terminals, leading to a serious deterioration in communication quality. In the system- and link-level simulation of HST channels, this non-stationarity should be characterized and modeled properly. In this paper, the sizes of the quasi-stationary regions are quantified to measure the significant changes in channel statistics, namely, the average power delay profile (APDP) and correlation matrix distance (CMD), based on a measurement campaign conducted at 2.4 GHz. Furthermore, parameters of the multi-path components (MPCs) are estimated and a novel clustering-tracking-identifying algorithm is designed to separate MPCs into line-of-sight (LOS), periodic reflecting clusters (PRCs) from power supply pillars along the railway, and random scattering clusters (RSCs). Then, a non-stationary geometry-cluster-based stochastic model is proposed for viaduct and hilly terrain scenarios. Furthermore, the proposed model is verified by measured channel statistics such as the Rician K factor and the root mean square delay spread. The temporal autocorrelation function and the spatial cross-correlation function are presented. Quasi-stationary regions of the model are analyzed and compared with the measured data, the standardized IMT-Advanced (IMT-A) channel model, and a published non-stationary IMT-A channel model. The good agreement between the proposed model and the measured data demonstrates the ability of the model to characterize the non-stationary features of propagation environments in HST scenarios.

This paper investigates how to build an active attack detection framework that is driven by fundamental channel modeling and practical wireless datasets. Firstly, we propose the concept of interpretable channel fingerprints (ICFs), which correspond to the spatial-temporal parameters in real physical wireless signal propagation channels. Based on this, we design an adversarial autoencoder (AAE) with a semi-supervised learning network, which takes as inputs the power spectrum of quantized ICFs and enables small sample learning multiclassification tasks for different types of wireless channel active attacks. We have experimentally verified the performance of our AAE network using the Wireless InSite ray tracing software. Our results show that the proposed semi-supervised network outperforms the fully-supervised network especially in small sample conditions. We highlight the need for careful selection of the hyperparameters for learning rate and mini-batch size, and the system parameters for the ICF power spectrum resolution. We show that the detection accuracy of the proposed AAE model can reach more than 98% with only a small number of input samples.

Physical layer key generation (PLKG) is one of the promising security solutions for communications in the Internet of Things $(\mathrm{I}\mathrm{o}\mathrm{T})$. Among implementations for the PLKG, a low-cost, integrated, and portable design is still lacking. To this end, we build a cheap and moderately complex testbed with ESP32, which supports on-chip channel estimation. In the proposed testbed, four tasks are created to achieve channel probing and message exchanging for the PLKG. Considering independent clocks and packet mismatches in the real world, we implement a PLKG prototype that generates secret keys at a rate of 19.45 bit/s. Experimental results show that the generated keys can pass all commonly-used National Institute of Standards and Technology (NIST) randomness tests, and the transmitted information cannot be decrypted by the illegitimate user.

This article proposes a wireless key generation solution for secure low-latency communications with active jamming attack prevention in wireless networked control systems (WNCSs) of Industrial Internet of Things (IIoT) applications. We first identify a new vulnerability in physical-layer key generation schemes using wireless channel and random pilots (RPs) in static environments. We derive a closed-form expression for the probability that the RP-based key is successfully attacked by a long-term eavesdropper at a fixed location. To prevent such attacks, we propose a one-time pad (OTP) encrypted transmission solution assisted by one-way self-interference (SI), which has low-latency, high-security benefits, and active attack detection capability. The performance of the proposed scheme is analytically compared with two benchmark RP-based schemes, and its advantages are verified in a ray-tracing-based simulation environment. We further investigate the impact of critical design parameters, which reveal fundamental insights for the deployment and implementation of our proposed secure communications scheme.

Wireless secret key generation has emerged as a promising technique for Internet-of-Things (IoT) systems to establish shared encryption keys between the server and legitimate mobile user. This article focuses on the use of multidomain joint information to achieve a high key generation rate (KGR) and the implementation of a reliable, low-complexity secret key generation mechanism for distributed antenna systems (DAS) with orthogonal-frequency division multiplexing (OFDM). We present a space-time-frequency channel state information (CSI)-based key generation scheme based on a two-step approach of adaptive link selection and stepwise decorrelation algorithms. The performance is evaluated in terms of KGR, key disagreement rate (KDR), randomness, and computational complexity by using both a standardized channel model and real-world measurements. Numerical results show that our proposed low-complexity algorithms effectively utilize the space-time-frequency CSI to multiply the KGR in both indoor and outdoor environments. Through adaptive link selection in DAS, the KDR is maintained within a correctable range, thereby ensuring the validity of generated keys in dynamic environments. Further applying stepwise decorrelation reduces the computational complexity by more than half while satisfying all eight key generation randomness tests in the NIST test suite.

Wireless powered sensor networks (WPSNs) have evolved as a promising paradigm for energy-efficient communications. Recently, the proliferation of reconfigurable intelligent surface (RIS) has further been envisioned as a cost-effective solution for improving wireless power transfer (WPT) efficiency. In this paper, from the practical perspective of balancing the network sustainability and reliability, we consider a RIS-enhanced partially WPSN that composed of wireless-powered energy receivers (ERs) and battery-powered information receivers (IRs). Assuming the partially WPSN operates in time division multiple access (TDMA) mode, the joint optimization of covariance matrices, downlink/uplink (DL/UL) time allocation and RIS reflecting coefficients are investigated under the minimum DL rate constraint among all IRs for maximizing the achievable UL sum rate. Specifically, the single-IR single-ER (SISE) case is first studied based on the assumption of separate DL/UL RIS reflecting coefficients, in which an alternating optimization algorithm is proposed with semi-closed-form optimal solutions. In order to reduce the hardware overhead and signal processing complexity, we also investigate the case of identical DL/UL RIS reflecting coefficients, in which an iterative optimization algorithm is developed to tackle the coupled DL/UL transmissions. Then, we extend our work to the multiple-IRs multiple-ERs (MIME) case, where both the optimization problems corresponding to separate and identical DL/UL RIS reflecting schemes become more challenging to solve. To circumvent this intractability, we propose a successive convex relaxation (SCA) based alternating optimization algorithm and a low-complexity two-step algorithm. Finally, numerical results demonstrate the superior UL sum rate performance of our proposed algorithms over the adopted benchmarks.

Physical layer secret key (PLSK) attracts much research interest in recent years due to its lightweight properties and potential to be used in real internet-of-things (IoT) applications. However, most existing works focus on PLSK generation between pairwise users under passive eavesdropping and lack of design for group key generation especially when active attacks may exist. In this paper, the RObust and LIghtweight Group key generation Against Active Attack (ROLIG3A) scheme is proposed to achieve reliable, efficient, and secure group secret key generation (GSKG) against various malicious attack patterns. In ROLIG3A, adaptive channel listening and dual code channel transmission are designed to avoid legitimate users from jamming signals. Comparing with the exiting GSKG schemes, more proactive countermeasures, back interference cancellation, and ECC-based information sharing are involved to realize the identification and cancellation of malicious attacks. Simulation results show that ROLIG3A has better security performance than the state-of-the-art GSKG schemes in the face of both sabotage attacks and manipulative attacks. In addition, ROLIG3A can provide an acceptable key generation rate and it is reliable even in low signal-to-noise-ratio (SNR) environments.

Unmanned aerial vehicle (UAV) will be an essential carrier for future wireless communications due to its flexible deployment and low cost. As such, the information security of UAV communications is of paramount concern. In this paper, a novel physical layer secret key generation scheme is proposed for air-to-ground (A2G) UAV multiple-input-multiple-output (MIMO) communications, which is applicable in frequency division duplex (FDD) systems. In UAV communications, line-of-sight (LoS) propagation is a distinctive feature, which significantly weakens the performance of channel state information (CSI) based keys. Therefore, a novel channel parameter, three-dimension (3D) spatial angle, is employed to combat against a novel active eavesdropping method, which is termed as Environment Reconstruction based Attack for SEcret keys (ERASE). Compared to the existing plane-angle-based method, our scheme can efficiently utilize spatial resources and provide a higher key generation rate (KGR). The advantages of the proposed scheme are shown through both theoretical analysis and simulations.

Recently, the utilization of unmanned aerial vehicles (UAVs) has been increasingly popular in various fields. As a brand new scenario of wireless communication, establishing feasible UAV channel models is necessary for the design and deployment of UAV-aided communication systems. In this paper, a three-dimensional (3D) geometry-based stochastic model (GBSM) is proposed for UAV air-to-air (A2A) channels. Based on this model, we derive the space-time correlation function (STCF), through which the impacts of some key parameters have been analyzed. Finally, the better universality of the proposed model is verified and some useful conclusions giving references for practical design are provided.

Massive multiple-input multiple-output (MIMO) has become a promising solution to provide unprecedented spectral efficiency (SE) to future cellular networks, based on the idea of equipping the base station (BS) with hundreds or thousands of antenna elements operating in a coherent fashion. At the doors of the future fifth generation mobile communication networks (5G), massive MIMO has been widely recognized as one of the cornerstones, given the ambitious key performance indicators of the future standard. The channel state information (CSI) acquisition is one of the core activities in massive MIMO, on which its entire performance depends to a great extent. Pilot contamination has been recognized as the main limiting factor to acquire an accurate CSI, becoming the focus of a large body of research. In this paper, we focus on the pilot contamination problem in massive MIMO systems. An approach is proposed to mitigate this problem based on the use of superimposed (SP) pilots in combination with time-multiplexed (TM) pilot sequences. Specifically, we use the contaminated channel estimates to reduce the amount of interference produced when SP pilots are used. Results show that the amount of interference caused by transmitting pilots alongside the data is substantially reduced when this method is put into place. In turn, the proposed method leads to mitigating the pilot contamination.

Secret key generation based on channel reciprocity is an important branch of wireless physical layer security which provides strong or even theoretically perfect secret. With the pursuit of higher secret key rate, the time-frequency joint key extraction schemes have been recently considered, which can exploit simultaneously the resources contained in the frequency and time domains. However, the strong correlation existing between the broadband time-variant channel samples makes the decorrelation process indispensable before quantization. Due to the large size of these channel sample matrices, decorrelation methods based on full channel covariance matrix corresponding to high calculation complexity are difficult to deploy in realistic mobile communication systems. In this work, we propose a method which separates the decorrelation of time-frequency correlated channel characteristics into two steps. The performance of this proposed method is evaluated through both standardized channel model and practical measurements. It is shown that the decorrelation of samples with the two-step method could be close to that based on the full channel covariance matrix, as well as leading to a dramatical reduction on the calculation complexity. The impact of the signal-to-noise-ratio (SNR) on the secret key rate is analyzed, and no distinct difference is shown in this aspect between these two methods.

Recently, unmanned aerial vehicle (UAV) plays an important role in many applications because of its high flexibility and low cost. To realize reliable UAV communications, a fundamental work is to investigate the propagation characteristics of the channels. In this paper, we propose path loss models for the UAV air‐to‐air (AA) scenario based on machine learning. A ray‐tracing software is employed to generate samples for multiple routes in a typical urban environment, and different altitudes of Tx and Rx UAVs are taken into consideration. Two machine‐learning algorithms, Random Forest and KNN, are exploited to build prediction models on the basis of the training data. The prediction performance of trained models is assessed on the test set according to the metrics including the mean absolute error (MAE) and root mean square error (RMSE). Meanwhile, two empirical models are presented for comparison. It is shown that the machine‐learning‐based models are able to provide high prediction accuracy and acceptable computational efficiency in the AA scenario. Moreover, Random Forest outperforms other models and has the smallest prediction errors. Further investigation is made to evaluate the impacts of five different parameters on the path loss. It is demonstrated that the path visibility is crucial for the path loss.

With the recent developments of high-mobility wireless communication systems, e.g., high-speed train (HST) and vehicle-to-vehicle communication systems, the ability of conventional stationary channel models to mimic the underlying channel characteristics has widely been challenged. Measurements have demonstrated that the current standardized channel models, like IMT-Advanced (IMT-A) and WINNER II channel models, offer stationary intervals that are noticeably longer than those in measured HST channels. In this paper, we propose a non-stationary channel model with time-varying parameters, including the number of clusters, the powers, and the delays of the clusters, the angles of departure, and the angles of arrival. Based on the proposed non-stationary IMT-A channel model, important statistical properties, i.e., the local spatial cross-correlation function and local temporal autocorrelation function are derived and analyzed. Simulation results demonstrate that the statistical properties vary with time due to the non-stationarity of the proposed channel model. An excellent agreement is achieved between the stationary interval of the developed non-stationary IMT-A channel model and that of relevant HST measurement data, demonstrating the utility of the proposed channel model.

Wireless body area network (WBAN) has drawn more and more interests in recent years. As one alternative communication scheme for WBAN, human body communication (HBC) uses human body as the communication medium and it provides better performance for communication security, spectrum efficiency, power consumption, and electromagnetic compatibility. The aims of this paper are to measure and characterize a capacitive HBC channel and to build simple models for it. Measurements have been carried out for different electrode positions and different body shapes. The results show that the path-loss of HBC channel is a function of frequency and it needs to be modeled by separated frequency intervals. A general model is proposed for the path-loss for different scenarios and the model parameters are extracted by fitting methods. The impact of electrode positions on the path-loss is analyzed. To describe the dispersion effects in HBC channel, the coherent bandwidth and delay spread of HBC channels in different scenarios are also extracted from measured data. These results aim at providing references for the design and deployment of HBC systems.