Unmanned aerial vehicle (UAVs) are a key platform in the sixth generation (6G) communication networks, where integrated sensing and communication (ISAC) is also a promising technology that requires real-time channel estimation. This paper proposes a robust and efficient angle-of-arrival (AOA) estimation method based on a field-trained neural network (NN) for low-latency UAV ISAC applications. In this method, the NN is pre-trained quickly in the field with each receiving antenna element's channel state information (CSI) and the transceivers' locations. We extract the channel multi-paths from the CSI and calculate the path phases as the training data set in real time. Then the pre-trained NN is used for high-efficient AoA estimation in real time. A real-time UAV channel sounder is utilized to verify the proposed method. The measurement results show that the proposed field-trained estimation method is faster and more robust compared with the traditional method and fixed-trained NN. The proposed angle estimation method is valuable for UAV channel estimation and low-latency UAV ISAC applications.

Unmanned aerial vehicle (UAV) is considered as an important component for future communication networks. In this paper, an air-to-ground (A2G) channel sounder is designed and implemented for UAV communication channel measurement and characterization. The channel impulse response (CIR) extraction is implemented on a field programmable gate array (FPGA) to improve extraction efficiency. Based on the channel characteristics under measured (or baseline) scenarios, a transfer learning neural network (TNN) framework is also proposed to predict the channel characteristics of other unmeasured (or transferred) scenarios. In the proposed framework, the baseline matrices of neural network parameters are obtained from the measurement data of baseline scenarios. The ray tracing (RT) simulation data is only used to obtain the extrapolation matrices where we utilize imperfect digital map and do not require a highly accurate RT simulation. Then the neural network driven by the baseline and extrapolation matrices is used to predict the channel characteristics of transferred scenarios. To verify the proposed prediction method, the channel characteristics including path loss, K-factor, and root mean square delay spread of a near-urban scenario are firstly measured. Then, the corresponding channel characteristics of a transferred dense-urban scenario are predicted by the proposed TNN method and validated by the measurement data. It is shown that the predicted channel characteristics are well consistent with the measured ones.

Unmanned aerial vehicle (UAV) millimeter wave (mmWave) technologies can provide flexible link and high data rate for future communication networks. By considering the new features of three-dimensional (3-D) scattering space, 3-D velocity, 3-D antenna array, and especially 3-D rotations, a machine learning (ML)-integrated UAV-to-Vehicle (U2V) mmWave channel model is proposed. Meanwhile, an ML-based network for channel parameter calculation and generation is developed. The deterministic parameters are calculated based on the simplified geometry information, while the random ones are generated by the backpropagation-based neural network (BPNN) and generative adversarial network (GAN), where the training data set is obtained from massive ray-tracing (RT) simulations. Moreover, theoretical expressions of channel statistical properties, i.e., power delay profile (PDP), autocorrelation function (ACF), Doppler power spectrum density (DPSD), and cross-correlation function (CCF), are derived and analyzed. Finally, the U2V mmWave channel is generated under a typical urban scenario at 28 GHz. The generated PDP and DPSD show good agreement with RT-based results, which validates the effectiveness of proposed method. Moreover, the impact of 3-D rotations, which has rarely been reported in previous works, can be observed in the generated CCF and ACF, which are also consistent with the theoretical and measurement results.

A GPS sparse multipath signal estimation method based on compressive sensing is proposed. A new 0 norm approximation function is designed, and the parameter of the approximate function is gradually reduced to realize the approximation of 0 norm. The sparse signal is reconstructed by a modified Newton method. The reconstruction performance of the proposed algorithm is better than several commonly reconstruction algorithms at different sparse numbers and noise intensities. The GPS sparse multipath signal model is established, and the sparse multipath signal is estimated by the proposed reconstruction algorithm in this paper. Compared with several commonly used estimation methods, the estimation error of the proposed method is lower.