This paper validates the concept of spectrum superposing through blind adaptive array (BAA) inter-system interference suppression aided by blind interference estimation. Our proposed framework adaptively switches the BAA algorithm according to the output of the estimator, which classifies the polarity of signal-to-interference power ratio (SIR) trained by convolutional neural network using 2-dimensional time-frequency representation, i.e., spectrogram, of the received signal. Suppose LTE and wireless LAN coexistence scenario; simulation results show that the proposed scheme successfully selects the appropriate BAA algorithm to effectively suppress inter-system interference.

In this paper, we propose denoising autoencoder (DAE)-assisted secret key generation (SKG), where channel noise reciprocity imperfections induced due to wireless channel measurements are suppressed, hence significantly enhancing the reliability and efficiency. More specifically, the DAE is capable of capturing the intrinsic structure of input distributions, reconstructing the original data structure, and removing additive noise while preserving the essential structure of signals. In our analysis, it is demonstrated that the proposed SKG scheme exhibits higher performance than the conventional schemes in terms of key disagreement rate (KDR), secret key capacity (SKC), and randomness of the generated keys.

This paper proposes a novel randomized discrete Fourier transform spread spectrally efficient frequency division multiplexing (RDFT-s-SEFDM) system, which employs an expectation propagation (EP)-based detection algorithm to achieve superior detection performance. Randomized discrete Fourier transform (DFT) spreading is introduced to improve the performance of the EP-based algorithm even under significant bandwidth compression. The simulation results demonstrate that the proposed RDFT-s-SEFDM system achieves superior bit error rate (BER) performance compared to conventional SEFDM in additive white Gaussian noise (AWGN). In addition, the proposed system exhibits robustness against multipath frequency-selective fading channels by exploiting spreading and diversity gain, making it suitable for wireless channels with large delay spreads.

In semantic communication systems, deep learning-based joint source-channel coding (DeepJSCC) has demonstrated superior performance, particularly in low signal-to-noise ratio (SNR) scenarios and under limited bandwidth conditions, compared to traditional communication technologies. However, most existing studies rely on communication models based on autoencoders (AE), where the distribution of transmission symbols is treated as a complete black box, causing inefficient use of limited transmission symbols. Moreover, many existing methods for image transmission focus on optimizing pixel-wise metrics, without considering the semantic information of the images. This pixel-level optimization often compromises the semantic fidelity and perceptual quality of the reconstructed images. To address these problems, we propose a novel semantic communication system that combines a variational autoencoder (VAE) and a generative adversarial network (GAN). Specifically, a VAE is used to arbitrarily control the distribution of transmission symbols according to channel conditions, while a GAN is used to maximize the similarity of semantic information. Simulation results demonstrate that the proposed method improves the perceptual quality of the reconstructed images compared to conventional approaches.

Internet of Vehicles (IoV) is an emerging technology that has been rapidly developing in recent years, which combines the internet, wireless communication technology, and vehicle electronics to realize the intelligence of interaction between vehicles and greatly improve the efficiency and safety of the transportation system. However, high-frequency and cyclical communications between vehicles and the limited system capacity in the region have led to serious conflicts in the allocation of wireless resources for vehicle networking. To solve this problem, this paper models resource allocation as a multi-agent deep reinforcement learning problem and proposes a Multi-Agent Actor-Critic (MAAC) algorithm based on distributed execution. The multi-agent algorithm can achieve distributed computing by modeling the training mechanism and reward function, thus improving resource utilization. Experimental results show that the proposed algorithm improves the total Vehicle-to-Infrastructure (V2I) link capacity and Vehicle-to-Vehicle (V2V) link transmission success rate.

This paper introduces a novel robust signal-to-noise ratio (SNR) estimation against Doppler shift using a convolutional neural network (CNN)-based multimodal network. SNR estimation is valuable for Beyond 5G (B5G) and 6G since it helps to evaluate the channel state information (CSI) reliability that is fed back from users and schedule radio resources. The proposed scheme increases feature diversity by converting the same received signal into multiple modalities, enabling accurate and robust SNR estimation against Doppler shifts. Simulation results show that the proposed scheme provides the most accurate estimation compared to conventional schemes in the presence of Doppler shifts.

In this paper, we propose a novel physical-layer secret key generation (SKG), which randomly activates a pilot sequence during the channel probing phase. Each legitimate user interpolates the received signals to recover the full channel state information (CSI). Due to the random reduction of the pilot sequence, the proposed scheme can degrade the eavesdropping performance regardless of the correlation between legitimate users and eavesdroppers, thereby improving the secret key capacity (SKC). Furthermore, low power consumption during the channel probing phase can be achieved. Our simulation results demonstrate that the proposed scheme exhibits superior performance in the presence of eavesdroppers.

Asynchronous Quick Link (A-QL) is the tri-color-band screen code specified as one of the transmission symbol formats in IEEE802.15.7. An edge detection algorithm is required to extract information from A-QL symbol. However, edge detection wastefully recognizes unwanted objects and it results in communication failure. To improve its detection performance, this paper proposes to apply YOLO as deep learning-based object detection and demonstrates its effectiveness.

Optical camera communication (OCC) can be realized by commercial LEDs or displays as a transmitter and image sensors as a receiver. One of the challenges to enhance the transmission capacity in OCC is a two-dimensional light source with a display at the transmitting side. So far, the optimization of the imaging process and the method of tracking and detecting the light source have not been studied in detail. This paper proposes a dynamic light source detection system based on the A-QL method specified in IEEE 802.15.7 as a transmission symbol format. It employs YOLO, a deep learning-based object detection algorithm, and optimizes the image capture process as a symbol detection system. The proposed system can detect two-dimensional symbols with adjusting its angle and orientation. Its effectiveness and feasibility are demonstrated through an experimental evaluation.

In order to meet the ever-growing demand for data traffic, highly efficient multiple access schemes, such as OFDMA, are widely used in modern communication standards. In such multiple access schemes, adaptive modulation and coding (AMC) are used to optimize the transmission rate of each user. However, feedback information, such as SNR and Doppler shift, characterizing the communication environment of each user is indispensable of key importance for AMC. In the past, these information and parameters were often estimated using reference signals. However, the reference signal becomes overhead, resulting in throughput degradation and processing delay. Furthermore, the computation burden can be large as it is necessary to perform channel parameter estimation individually for each user. Previously, over the single-user channel, we have proposed a joint SNR and Doppler shift detection method via a spectrogram-based data-driven method, without the reference signal. This paper extends this framework to multiuser OFDM multiple access channels. In the newly proposed method, SNR and Doppler shift for all users can be detected simultaneously via deep learning-guided object detection algorithms from each spectrogram image. Simulation results are provided to validate the effectiveness of the proposed method.

Recently, vehicle to everything (V2X) communications is paid attention in order to realize a smart transportation such as safety driving and automated cruise. It necessitates an ultra-reliable and low-latency communications (URLLC) even at fast fading environment involved by moving vehicles. Generally, forward error correction (FEC) code such as polar code requires long block code length to raise error correction capability while it deteriorates the channel tracking accuracy in pilot-aided estimation. In our previous work, to meet these trade-off demands, we proposed a frame splitting method. It splits information data in half and estimate channel state information (CSI) of the former part. That of latter is simply extrapolated using the former CSI. In this paper, we expand the previous work into multiple-input and multiple-output (MIMO) eigenbeam-space division multiplexing (E-SDM) system, which can improve communication accuracy significantly. To keep the benefit of E-SDM for fast fading, we propose two data assignment methods based on the eigenbeams. The computer simulation results clarified that these two methods could improve the BER performance in the V2X communication.

With the increase in demand for wireless data traffic, high-speed communication systems are required in many environment. Adaptive modulation and coding is an important technique to realize this, but it requires feedback of communication environment information represented by SNR. Conventional SNR estimation methods have a problem of degrading estimation accuracy in fast-moving environments and the presence of carrier frequency offset (CFO). Convolutional neural network (CNN) is capable of estimating the SNR from the trained received signal dataset accurately. This paper investigates an input signal representation to further improve the SNR classification accuracy by CNN. We then propose to combine respective spectrogram data of IQ domains. It can extract the features related to SNR from the received signal to the maximum extent possible and thus achieve highly accurate SNR estimation. Simulation results show that the proposed approach outperforms the other existing candidates.

This paper proposes a novel convolutional neural network (CNN) based joint classification method to characterize the signal-to-noise power ratio (SNR) and Doppler shift using spectrogram images, in order to enable efficient adaptive modulation and coding (AMC) designs. It is necessary to maintain high communication performances even in stringent environments where transceivers move at high speed due to the diversification of wireless applications. To optimize the transmission rate in such dynamic environments, AMC scheme is known to be effective. AMC is generally designed based on feedback information (FBI) such as the SNR and Doppler shift acquired on the receiving side. Here, the challenge is an increase in calculation burden, processing delay and estimation accuracy of the FBI. We focused on the spectrogram which is composed of power values in the time and frequency domains. Its two-dimensional fluctuation represents the Doppler shift as well as noise values. In the proposed method, such key information for AMC can be simultaneously extracted from a single spectrogram via a trained CNN. Therefore it is expected to contribute to reducing the computational burden and speeding up the signal processing. Simulation results are presented to demonstrate that the proposed method achieves better performance than traditional methods.

This paper proposes a highly accurate signal to noise ratio (SNR) estimation method for adaptive modulation and coding (AMC) by learning power spectral density (PSD) images with a convolutional neural network (CNN). Accurate SNR estimation is indispensable for adaptive control schemes such as AMC. Proposed method trains the CNN using PSD images and the corresponding SNR value as a teacher signal, and outputs the SNR in response to an input of unknown PSD image. Once trained, the SNR can be estimated with high speed and high accuracy, so AMC performance can be improved. Furthermore, since PSD is hardly affected by the Doppler shift, proposed method is resistant to high-speed environment. Compared with the previously proposed method which estimates SNR from a PSD data using a neural network, the proposed method can improve the estimation accuracy, bit error rate (BER) and throughput performance when AMC is applied.

This paper proposes the blind interference power estimation via deep learning approach exploiting the visualized wireless signal information. Blind adaptive array (BAA) signal processing is the powerful solution to suppress various kinds of interference such as inter-cell interference (ICI) and intersystem interference (ISysI) for which receivers cannot obtain a priori information represented as channel state information (CSI). However, BAAs cannot always suppress interference due to its blind nature. Depending on signal-to-interference power ration (SIR), adequate BAA algorithms should be switched. In order to estimate SIR in a blind manner, we propose to apply a convolutional neural network (CNN) trained by IQ constellation images where contains the desired and interference signals. This paper presents its methodology and fundamental possibility.