A tropospheric duct (TD) is an anomalous atmospheric refraction structure in marine environments that seriously interferes with the propagation path and range of electromagnetic (EM) waves, resulting in serious influence on the normal operation of radar. Since the propagation loss (PL) can reflect the propagation characteristics of EM waves inside the duct layer, it is important to obtain an accurate cognition of the PL of EM waves in marine TDs. However, the PL is strongly non−linear with propagation range due to the trapped propagation effect inside duct layer, which makes accurate prediction of PL more difficult. To resolve this problem, a novel multiscale decomposition prediction method (VMD−PSO−LSTM) based on the long short−term memory (LSTM) network, variational mode decomposition (VMD) method and particle swarm optimization (PSO) algorithm is proposed in this study. Firstly, VMD is used to decompose PL into several smooth subsequences with different frequency scales. Then, a LSTM−based model for each subsequence is built to predict the corresponding subsequence. In addition, PSO is used to optimize the hyperparameters of each LSTM prediction model. Finally, the predicted subsequences are reconstructed to obtain the final PL prediction results. The performance of the VMD−PSO−LSTM method is verified by combining the measured PL. The minimum RMSE and MAE indicators for the VMD−PSO−PSTM method are 0.368 and 0.276, respectively. The percentage improvement of prediction performance compared to other prediction methods can reach at most 72.46 and 77.61% in RMSE and MAE, respectively, showing that the VMD−PSO−LSTM method has the advantages of high accuracy and outperforms other comparison methods.

At present, it is difficult to obtain the indoor propagation loss quickly and accurately by directly using measurements in the millimeter wave band. To solve this problem, in this paper, a ray tracing method suitable for indoor scenes based on geometric optics theory, the uniform theory of diffraction and image theory is presented; the space-alternation generalized expectation-maximization (SAGE) algorithm is used to analyze the measured data and the multipath information of the wireless channel is analyzed; three deep learning models are used to predict the path loss at different receiving distances based on 1600 sets of path loss data. The results show that the comparison between the ray tracing and experimental results shows a good agreement. Moreover, the root-mean-square error (RMSE) and mean absolute error (MAE) of the long short-term memory (LSTM) network are the smallest, and the LSTM has a better fitting effect on the propagation loss sequences predicted at more distant locations when compared with the recurrent neural network (RNN) and gate recurrent unit (GRU) methods, which can better reflect the propagation trend. This provides theoretical support for the layout of base stations and network optimization in typical open indoor environments.

Millimeter-wave channel modeling is the basis of fifth-generation (5G) communication network design and applications. In urban microcellular environments, the roughness of wall surfaces can be comparable to the wavelengths of millimeter waves, resulting in walls that cannot be considered as smooth surfaces. Therefore, channel modeling methods based on only traditional three-dimensional ray tracing (RT) or the three-dimensional parabolic equation (PE) result in the limited computational accuracy of millimeter-wave channel models for urban environments. Based on the scattering theory of a rough surface and the typical scattering characteristics of a millimeter wave, the end field of the three-dimensional vector PE is regarded as the initial field of three-dimensional RT. Moreover, the number of scattered rays and scattering angles are introduced. Finally, a three-dimensional fully polarized millimeter-wave hybrid propagation channel model (3DFPHPCM) is proposed. The proposed model improves the computational accuracy of a single deterministic model. Millimeter-wave channel measurements in non-line-of-sight (NLOS) environments were carried out to verify and optimize the proposed 3DFPHPCM. The results show that the root mean square error (RMSE) and mean absolute error (MAE) of the proposed 3DFPHPCM are both minimized when compared to three-dimensional RT or the three-dimensional PE, which indicates that the proposed 3DFPHPCM has higher computational accuracy. Moreover, its runtime is the shortest among the methods. The results presented herein provide technical support for the layout of base stations.

Accurate acquisition of wave parameters is important for marine monitoring and target identification. Traditional methods are limited in sea state adaptability, real-time observation, and inversion accuracy. Tower-based non-scanning pulsed radar, as a low-cost near-shore method, provides all-weather, high temporal resolution observations without beam scanning. However, it yields two-dimensional range-pulse echo data blocks without azimuth information, making traditional wave spectrum inversion difficult to apply directly. To address this, we propose a wave parameter inversion method based on a ResNet–Transformer hybrid network. It takes sea clutter characteristics, radar parameters, and wind parameters as inputs, and uses 1D ResNet for local feature extraction and a Transformer encoder for cross-feature global interaction modeling, achieving simultaneous inversion of relative wave direction, mean wave period, and significant wave height. Experimental results on measured sea clutter samples demonstrate that the proposed method achieves high accuracy across all tasks. The RMSEs are 24.4 degrees, 0.5s, and 0.3 m for relative wave direction, mean wave period, and significant wave height. Experimental results on the measured dataset demonstrate that the model effectively captures the nonlinear mappings. These results support the feasibility of wave parameter estimation from tower-based non-scanning radar measurements under the evaluated conditions.