Integrated sensing and communication has emerged as a key paradigm in next-generation wireless systems. In orthogonal frequency-division multiplexing (OFDM)-based radar systems with time-division duplexing (TDD), the separation of downlink (DL) and uplink (UL) slots leads to spectral replicas along the Doppler axis, degrading sensing performance. To address this issue, we propose a convolutional long short-term memory autoencoder that predicts the channel state information (CSI) associated with unobserved UL slots between DL transmissions in a TDD frame. By estimating the CSI in unobserved UL slots, the proposed method effectively suppresses spectral replicas in the range-Doppler (RD) map. Simulation results under 3rd Generation Partnership Project New Radio numerologies (μ=1,2,3) demonstrate that the proposed model accurately predicts CSI values across diverse channel conditions. The model achieves a magnitude mean squared error (MSE) of 0.014 and a phase MSE of 0.038, consistently outperforming other methods. In addition, under equal parameter budgets, it achieves superior accuracy–efficiency trade-offs over alternative deep learning models, requiring only 7.2 ms per frame on a modern Internet-of-Things edge device. RD map analysis further confirms replica suppression, improving velocity coverage from 13.05 m/s to 613.5 m/s and signal-to-noise ratio from 45.16 dB to 60.91 dB compared with conventional methods, while maintaining target detection rates above 0.99. These results confirm that accurate CSI reconstruction by the proposed model significantly improves sensing fidelity in TDD-based OFDM radar systems.

This paper presents a method to associate detected objects between automotive vision sensors and radar sensors. Recently, studies on classifying objects by combining radar sensors and deep learning have been conducted. However, it is not easy to acquire labeled radar data to train deep learning models. Labeling these training data requires human effort, which is time-consuming. Furthermore, points obtained using radar sensors lack the ability to describe visual features such as the shape of objects, making the labeling process even more challenging. In this paper, we propose a method to annotate radar sensor data by associating the detection result of radar sensor with vision sensor using the iterative closest point algorithm. The proposed method can be used to annotate additional information such as object’s class information to radar data without human effort.

Urban air mobility (UAM) technologies are evolving rapidly, accompanied by a growing demand for advanced sensing and communication systems to enhance environmental awareness. The orthogonal frequency-division multiplexing (OFDM) radar-based joint radar and communication (JRC) systems have emerged as a key solution. The utilization of these systems in UAM applications allows for efficient use of limited frequency bands and eliminates the need for extra hardware installations. However, employing communication protocols such as long-term evolution presents a challenge due to the diminished range and velocity resolution in the OFDM radar system. Thus, a method for accurate estimation of range and velocity is necessary for OFDM radars. In this study, a geometric sequence decomposition-based accurate range and velocity estimation method is introduced. This method decomposes the received data into a superposition of geometric sequences, where each geometric sequence contains the propagation delay and Doppler shift (i.e., range and velocity of the target). Multiple simulations were conducted to validate the efficacy of our proposed method. When the signal-to-noise ratio is over 8 dB, the difference between the true and estimated range and velocity was less than the ± 1 m or ± 1 m/s. Furthermore, air-to-everything communications efficacy is validated based on the bit error rate analysis. According to the results, utilizing the proposed method can enhance the performance of the JRC system for the application in UAM.

This article explores the application of a stationary multiple-input and multiple-output frequency-modulated continuous-wave radar system for indoor environment mapping. In static situations, it is challenging to create indoor maps using fixed radar systems because only a few strong reflection points are prominently visible in the detection results. Our proposed method uses a stationary radar system, leveraging multipath echoes of radar signals to produce detailed indoor maps. Ghost targets caused by multipath propagation of signals contain information about the geometry of propagation paths. Unlike existing studies that primarily aimed to mitigate ghost targets, our method uses the information provided by both real and ghost targets to improve the mapping of indoor environments. For this purpose, we first extract detected peaks in the range-azimuth domain and compare the estimated Direction of Arrival (DoA) and Direction of Departure (DoD) to identify ghost targets and real targets. Then, the positions of paired real and ghost targets are used to estimate the geometry of reflection points, aiding in the creation of maps for surrounding indoor environments. Finally, we use density-based spatial clustering of applications with noise algorithm and running average filter to eliminate outliers, improving the accuracy and clarity of the generated indoor map. Simulation and experimentation results validate the effectiveness of the proposed method. The average root mean-square error (RMSE) of the estimated indoor map compared to the ground truth was 0.18 m.