A channel modeling and phase design method and system for an aerial intelligent reflector

Abstract

The present invention proposes a channel modeling and phase design method and system for an aerial intelligent reflective surface, which belongs to the field of wireless communication technology. The method includes: establishing a random rotation angle model of an unmanned aerial vehicle, and obtaining the local coordinates of the reflection unit on the intelligent reflective surface that change with time based on the random rotation angle of the unmanned aerial vehicle; using a deep learning time series prediction algorithm to establish a deflection angle prediction model for the unmanned aerial vehicle, and predict the rotation angle of the unmanned aerial vehicle; according to the prediction result of the rotation angle of the unmanned aerial vehicle, the phase offset of the intelligent reflective surface is calculated; based on the local coordinates of the reflection unit on the intelligent reflective surface, the position, movement speed, movement direction of the transmitting end, the receiving end and the unmanned aerial vehicle, and the system carrier frequency, an aerial intelligent reflective surface-assisted channel model is established, and the channel impulse response is obtained, and the statistical characteristics of the channel are obtained based on the channel impulse response. It can effectively reduce the impact of the instability of the unmanned aerial vehicle on the channel characteristics, and significantly improve the system performance of the aerial intelligent reflective surface.

Description

Technical Field

[0001] The invention belongs to the technical field of wireless communications, and in particular relates to a channel modeling and phase design method and system for an aerial intelligent reflection surface. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] With the development of 5G and future 6G communication technologies, improving wireless communication quality and system capacity has become an important research direction. As an emerging technology, reconfigurable smart reflective surface has shown great potential in improving communication quality due to its advantages of low cost and high energy efficiency. Smart reflective surface is composed of many passive reflective units, and the reflection coefficients of these units can be flexibly adjusted by programming, thereby reconfiguring its electromagnetic response and realizing dynamic regulation of the wireless propagation environment.

[0004] Traditional smart reflective surfaces are usually deployed on the surface of buildings or walls. This deployment method requires the transceiver to be located on one side of the reflective surface, that is, it only supports half-space signal reflection. In addition, the channel between the smart reflective surface and the transceiver is easily blocked by obstacles such as trees and buildings, thereby reducing system performance. Due to its excellent maneuverability, drones can flexibly move in three-dimensional space and avoid obstacles, thereby establishing a stable communication link with ground nodes. Based on this advantage, combining smart reflective surfaces with drone communication technology, that is, mounting smart reflective surfaces on drones to form aerial smart reflective surfaces, has become a research hotspot in the field of wireless communications. Aerial smart reflective surfaces can be dynamically deployed according to actual needs, and can provide 360° full-angle signal reflection, thereby significantly improving the performance of the communication system. However, compared with traditional ground smart reflective surfaces, aerial smart reflective surfaces face more complex environmental factors. For example, aerial smart reflective surfaces may be affected by factors such as wind, atmospheric turbulence, and propeller rotation, causing them to deflect randomly. Studies have shown that aerial smart reflective surface systems are sensitive to changes in the attitude of drones. For high-frequency band communications such as millimeter waves, even a slight rotation of the reflecting surface may significantly affect the phase alignment of the reflected signal, thereby degrading system performance.

[0005] In the current research on intelligent reflector channel modeling, the instability modeling of UAVs is too simple, and the phase design of the reflector still follows the traditional design ideas, without fully considering the impact of UAV instability on channel characteristics and system performance. Summary of the invention

[0006] In order to overcome the above-mentioned shortcomings of the prior art, the present invention provides a channel modeling and phase design method and system for an aerial intelligent reflecting surface, aiming to solve the challenges faced by the aerial intelligent reflecting surface communication system in the prior art, especially the problem of how to maintain and improve the performance of the communication system under the influence of UAV instability and environmental factors.

[0007] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0008] The first aspect of the present invention provides a channel modeling and phase design method for an aerial intelligent reflection surface;

[0009] A channel modeling and phase design method for an aerial intelligent reflection surface, comprising:

[0010] Step S1, establishing a random rotation angle model of the drone, and obtaining the local coordinates of the reflection unit on the smart reflection surface that change with time based on the random rotation angle of the drone;

[0011] Step S2: Based on the acquired random rotation angle of the drone, a drone deflection angle prediction model is established using a deep learning time series prediction algorithm to predict the rotation angle of the drone;

[0012] Step S3: Calculate the phase shift of the intelligent reflection surface according to the prediction result of the rotation angle of the UAV to optimize the coherent superposition of the signal at the receiving end;

[0013] Step S4: Based on the local coordinates of the reflection unit on the intelligent reflection surface, the position, movement speed, movement direction, and system carrier frequency of the transmitter, the receiver, and the drone, a channel model assisted by an aerial intelligent reflection surface is established; based on the phase shift of the intelligent reflection surface, a channel impulse response is obtained; and based on the channel impulse response, statistical characteristics of the channel are obtained.

[0014] As a further technical solution, the process of establishing the UAV random rotation angle model is as follows:

[0015] The random rotation angle of the drone is modeled by a sinusoidal random process, and the formula is:

[0016] ;

[0017] ;

[0018] ;

[0019] in, , and are the vibration amplitudes of yaw, pitch and roll angles, is the vibration frequency;

[0020] Gauss-Markov is used to model the vibration amplitude of yaw, pitch and roll angles; the formula is as follows:

[0021] ;

[0022] ;

[0023] ;

[0024] in, , and Respectively represent the vibration amplitudes of the yaw angle, pitch angle and roll angle at the initial moment The value of ， and Used to adjust randomness, between 0 and 1; , and is the asymptotic mean of the rotation angle, Equal to 0 or 1, used to control the coefficient of the noise term.

[0025] As a further technical solution, the local coordinates of the reflection unit on the smart reflection surface obtained based on the random rotation angle of the drone are as follows:

[0026] The smart reflective surface is deployed on the drone and experiences random deflection as the drone time, The first Reflection unit, The local coordinates of can be calculated as follows:

[0027]

[0028] in, , and They are the rotation matrices corresponding to the yaw angle, pitch angle, and roll angle respectively.

[0029] As a further technical solution, the UAV deflection angle prediction model adopts a temporal convolution-long short-term memory network model, which includes:

[0030] Input layer: used to receive the time series of three channels, which are used to input the historical data of yaw angle, pitch angle and roll angle changing over time;

[0031] TCN layer: The convolution operation is used to process the historical data of the input rotation angle changing over time, and the local temporal features and short-term dependencies in the historical data of the yaw angle, pitch angle and roll angle changing over time are extracted.

[0032] ReLU activation layer: The output of the TCN layer is transformed nonlinearly through the ReLU activation function, which introduces nonlinear characteristics to enhance the expressiveness of the model.

[0033] LSTM layer: The LSTM layer captures long-term dependencies in time series data through a gating mechanism;

[0034] Output layer: The output layer consists of a fully connected layer, and the output dimension of the fully connected layer is 3, corresponding to the prediction results of yaw angle, pitch angle and roll angle respectively. Through the fully connected layer, the high-dimensional time series features extracted by the LSTM layer are compressed and converted into the final prediction results.

[0035] As a further technical solution, the phase shift of the smart reflective surface is:

[0036]

[0037] in, For The distance from the transmitter to the center of the smart reflective surface at any moment; For The distance from the center of the smart reflector to the receiving end at any moment; and They are the horizontal angle and elevation angle from the transmitter to the smart reflector surface; and They are the horizontal angle and elevation angle from the receiving end to the smart reflector surface; , , , are the x-axis and y-axis coordinates of the nth reflection unit in the reflection surface; , , , , , are the parameters obtained in the process of calculating the transmission distance from the transmitter to the smart reflector and from the smart reflector to the receiver, respectively reflecting the influence of the yaw angle, pitch angle and roll angle of the smart reflector on the reflection phase; , , are the yaw angle, pitch angle and roll angle respectively. The predicted value at time.

[0038] As a further technical solution, the channel impulse response includes a channel impulse response reflected by the smart reflection surface and a channel impulse response not affected by the smart reflection surface; wherein the channel impulse response reflected by the smart reflection surface is expressed as:

[0039]

[0040] in, is the wavelength, and Respectively represent the spacing between adjacent columns and rows of reflection units; and are the path losses from the transmitter to the smart reflector and from the smart reflector to the receiver, and are the gains of the transmitting antenna and the receiving antenna respectively, is the angle between the incoming wave direction and the normal vector of the smart reflection surface, is the angle between the reflected wave direction and the normal vector of the smart reflection surface;

[0041] Indicates the initial distance between the transmitter and the center of the smart reflective surface; Indicates the initial distance between the receiving end and the center of the smart reflective surface; and It is the Doppler shift caused by the relative motion between the drone and the transmitter and receiver; and It is the Doppler shift caused by the motion of the transmitter and receiver.

[0042] The channel impulse response without the influence of the smart reflector is expressed as follows:

[0043] ;

[0044] in, yes The path loss is , S is the number of clusters in the channel, Indicates The number of multipaths in a cluster, is the power of the sth cluster, , , , , , denote the transmission distance, elevation departure angle, horizontal departure angle, elevation arrival angle and horizontal arrival angle of the ith path in the sth cluster respectively; the movement speed and azimuth of the transmitter are and ; The speed and azimuth of the receiving end are and .

[0045] As a further technical solution, the statistical characteristics of the channel include time correlation function, Doppler power spectrum density, and system achievable rate;

[0046] The time correlation function is:

[0047] ;

[0048] in, represents the mathematical expectation, represents the complex conjugate, Indicates a time interval;

[0049] The Doppler power spectral density is obtained by Fourier transform of the time autocorrelation function, and the formula is as follows:

[0050] ;

[0051] The system's achievable rate formula is as follows:

[0052] ；

[0053] in, is the signal transmission power, is the noise power.

[0054] A second aspect of the present invention provides a channel modeling and phase design system for an aerial intelligent reflection surface.

[0055] A channel modeling and phase design system for an aerial intelligent reflection surface, comprising:

[0056] The UAV random rotation angle generation module is configured to: establish a UAV random rotation angle model, and obtain the local coordinates of the reflection unit on the smart reflection surface that change with time based on the UAV random rotation angle;

[0057] The UAV rotation angle prediction module is configured to: establish a UAV deflection angle prediction model based on the acquired random rotation angle of the UAV using a deep learning time series prediction algorithm to predict the rotation angle of the UAV;

[0058] The smart reflective surface phase shift calculation module is configured to: calculate the phase shift of the smart reflective surface according to the predicted result of the rotation angle of the UAV, so as to optimize the coherent superposition of the signal at the receiving end;

[0059] The channel modeling module is configured to: establish a channel model assisted by an aerial intelligent reflection surface based on the local coordinates of the reflection unit on the intelligent reflection surface, the position, movement speed, movement direction, and system carrier frequency of the transmitter, receiver, and drone; obtain the channel impulse response based on the phase offset of the intelligent reflection surface; and obtain the statistical characteristics of the channel based on the channel impulse response.

[0060] The third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in the channel modeling and phase design method for an aerial intelligent reflection surface as described in the first aspect of the present invention.

[0061] The fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the channel modeling and phase design method of an aerial intelligent reflection surface as described in the first aspect of the present invention are implemented.

[0062] One or more of the above technical solutions have the following beneficial effects:

[0063] (1) The channel model of the present invention takes into account the random deflection of the UAV fuselage caused by adverse weather conditions such as wind and atmospheric turbulence, which in turn leads to phase mismatch of the reflected signal and system degradation. The present invention proposes a UAV deflection angle prediction algorithm based on deep learning. The prediction results are applied to the reflection phase design of the intelligent reflective surface. The results show that the present invention can effectively reduce the impact of UAV instability on channel characteristics and significantly improve the system performance of the aerial intelligent reflective surface.

[0064] (2) The present invention proposes a three-dimensional channel model and system for an aerial intelligent reflective surface. The system can achieve 360° panoramic signal reflection, thereby providing a wider signal coverage capability. Due to the high maneuverability of the drone, it can flexibly avoid obstacles. This allows the intelligent reflective surface to maintain a stable line-of-sight connection with the ground node. By setting reasonable model parameters, the model can be applied to different communication scenarios.

[0065] (3) The use of the temporal convolution-long short-term memory network (TCN-LSTM) can effectively capture the short-term local dependencies in the time series and enhance the feature extraction capability of the model. At the same time, the model can further process long-term dependencies and better capture complex time series features, thereby improving prediction accuracy.

[0066] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0068] Figure 1 This is a flow chart of the method of the first embodiment.

[0069] Figure 2 Schematic diagram of the aerial intelligent reflective surface communication scenario in the first embodiment.

[0070] Figure 3 This is a comparison chart between the TCN-LSTM and LSTM model’s prediction results of the drone’s deflection angle and the true value in the first embodiment.

[0071] Figure 4 This is a comparison chart of the system achievable rates at different UAV flight altitudes in the first embodiment.

[0072] Figure 5 It is a system structure diagram of the second embodiment. DETAILED DESCRIPTION

[0073] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0074] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0075] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0076] Embodiment 1

[0077] This embodiment discloses a channel modeling and phase design method for an aerial intelligent reflection surface;

[0078] like Figure 1 As shown, a channel modeling and phase design method for an aerial intelligent reflection surface includes:

[0079] Step S1, establishing a random rotation angle model of the drone, and obtaining the local coordinates of the reflection unit on the smart reflection surface that change with time based on the random rotation angle of the drone;

[0080] In step S1, Figure 2 As shown in Figure 2, due to the influence of wind and atmospheric turbulence, the drone can experience random rotation. At this moment, its yaw angle, pitch angle, and roll angle are respectively , , It can be modeled as the following sinusoidal random process:

[0081] ;

[0082] ;

[0083] ;

[0084] in, , and are the vibration amplitudes of yaw, pitch and roll angles, is the vibration frequency. The present invention uses the Gauss-Markov process to model the vibration amplitude, which can not only accurately characterize the stability and randomness of the vibration, but also capture the time-dependent characteristics of the vibration through the Markov property and autoregressive structure. The vibration amplitude is expressed as follows:

[0085] ;

[0086] ;

[0087] ;

[0088] in, , and Respectively represent the vibration amplitudes of the yaw angle, pitch angle and roll angle at the initial moment The value of , and Used to adjust the randomness, between 0 and 1. , and is the asymptotic mean of the rotation angle, Equal to 0 or 1, used to control the coefficient of the noise term.

[0089] Smart reflective surface Reflection units, and Represent the number of columns and rows of the reflective unit. The spacing between adjacent columns and rows is and . Smart reflective surface The reflection unit is denoted as At the initial moment, The local coordinates of the center point of the intelligent reflective surface can be expressed by the vector It indicates that, , , , , represents transpose, Indicates rounding up. Consider deploying a smart reflective surface on a drone, which can experience random deflection with the drone. time, The local coordinates of can be calculated as follows:

[0090]

[0091] in, , and They are the rotation matrices corresponding to the yaw angle, pitch angle, and roll angle respectively.

[0092] Step S2: Based on the acquired random rotation angle of the drone, a drone deflection angle prediction model is established using a deep learning time series prediction algorithm to predict the rotation angle of the drone;

[0093] In the current time series prediction problem, the Long Short-Term Memory (LSTM) network is often used. The recursive structure of this network is good at processing complex nonlinear time series relationships, can retain different information at each time step, and infer future time steps based on the context of the sequence. However, LSTM has gradient vanishing and explosion problems, and there may be limitations in modeling long-term dependencies. The Temporal Convolutional Network (TCN) can better model dependencies in sequence data by using convolution kernels across time steps to capture long-term dependencies. In addition, TCN can be calculated in parallel through convolution layers, and computing resources can be used more efficiently when processing long sequence data. Therefore, in this embodiment, TCN and LSTM are combined for drone deflection angle prediction. The TCN-LSTM model combines two different types of network structures. TCN effectively captures short-term local dependencies in time series through local convolution, enhancing the feature extraction ability of the model, while LSTM further processes long-term dependencies. This combination enables the TCN-LSTM model to better capture complex time series features, thereby improving prediction accuracy.

[0094] Specifically, the TCN-LSTM model includes an input layer, a TCN layer, a ReLU activation layer, a LSTM layer, and an output layer. It processes the input time series data step by step and outputs the prediction result of the next time step.

[0095] The model structure is as follows:

[0096] Input layer: The function of the input layer is to receive the original time series data. For each time step, the input data contains multiple features. Assume that the input data is a three-dimensional tensor with a shape of (None, T, N), where None represents the batch size, that is, the number of samples input each time. T is the number of time steps, indicating the length of the sequence. N is the number of features for each time step, such as the number of channels for multi-channel data. For the implementation of the present invention, the dimension of the input data is (None, num_steps-1, 3), where 3 represents the three channels for each time step, namely the yaw angle, pitch angle and roll angle.

[0097] TCN layer: In order to extract local features in time series data, a time series convolution layer, namely the TCN layer, is added after the input data. This layer uses causal convolution to ensure that the convolution operation only depends on the current time step and the previous time step, avoiding leakage of future information. The configuration of this convolution layer is: filters=50, which means that this layer outputs 50 convolution kernels and extracts 50 features of the input data. kernel_size=3, which means that the size of the convolution kernel is 3, which means that each convolution operation considers 3 consecutive time steps. padding= 'causal' ensures that the convolution operation complies with causality and avoids information leakage. The output dimension of the convolution layer is (None, T-1, 50), that is, 50 features are output.

[0098] ReLU activation layer: After the TCN layer, the ReLU activation function is used to enhance the model's expressiveness by introducing nonlinear transformations. The ReLU activation function can effectively reduce the gradient vanishing problem and improve training efficiency.

[0099] LSTM layer: used to model long-term dependencies in time series. The LSTM layer effectively captures time-dependent patterns by propagating memories in sequence data. The configuration of the LSTM layer is: num_hidden_units=50, indicating that the LSTM layer has 50 hidden units. return_sequences=True, indicating that the LSTM outputs the hidden state of each time step for subsequent fully connected layers to process. The output dimension of the LSTM layer is (None, T-1, 50), that is, each time step has a 50-dimensional hidden state.

[0100] Output layer: Through a fully connected layer, the output of the LSTM layer is mapped to the prediction result. The configuration of this layer is: num_responses=3, which means that the prediction values ​​of 3 channels (i.e. yaw angle, pitch angle and roll angle) are output. The dimension of the output layer is (None, T-1, 3), that is, the prediction values ​​of 3 channels are output at each time step.

[0101] During the training phase of the model, the model parameters are optimized by minimizing the mean square error loss function. The model loss function is calculated by mean_squared_error, and the optimizer uses the Adam optimization algorithm. During the training process, the model is trained by the model.fit(X, Y, epochs=100) method, where X is the input data, Y is the target output data, and the training cycle is 100 times. After the training is completed, the model can be predicted by the model.predict(X) method in TensorFlow.

[0102] Step S3: Calculate the phase shift of the intelligent reflection surface according to the prediction result of the rotation angle of the UAV to optimize the coherent superposition of the signal at the receiving end;

[0103] In step S3, the intelligent reflective surface controller applies different phase shifts to each reflective unit so that the reflected signal is coherently superimposed in the direction of the receiving end, thereby increasing the received signal power, i.e., passive beamforming. Under harsh transmission conditions such as strong winds, drones often vibrate violently, which will cause significant phase errors. Therefore, the phase shift of the nth reflective unit in the intelligent reflective surface is calculated as follows:

[0104]

[0105] in, For The distance from the transmitter to the center of the smart reflective surface at any moment; For The distance from the center of the smart reflector to the receiving end; and They are the horizontal angle and elevation angle from the transmitter to the smart reflector surface; and They are the horizontal angle and elevation angle from the receiving end to the smart reflector surface; , are the x-axis and y-axis coordinates of the nth reflection unit in the reflection surface; , , are the parameters obtained in the process of calculating the transmission distance from the transmitter to the smart reflector and from the smart reflector to the receiver, respectively reflecting the influence of the yaw angle, pitch angle and roll angle of the smart reflector on the reflection phase; , , are the yaw angle, pitch angle and roll angle respectively. The predicted value at time.

[0106] Step S4: Based on the local coordinates of the reflection unit on the intelligent reflection surface, the position, movement speed, movement direction, and system carrier frequency of the transmitter, the receiver, and the drone, a channel model assisted by an aerial intelligent reflection surface is established; based on the phase offset of the intelligent reflection surface, a channel impulse response is obtained; and based on the channel impulse response, statistical characteristics of the channel are obtained.

[0107] In step S4, a typical aerial intelligent reflector-assisted communication scenario is as follows: Figure 2 As shown. Assume that the initial position of the sender is the origin of the coordinate system, and the initial position of the receiver is ( , the initial position of the smart reflective surface is In addition, the flight speed and azimuth of the UAV are and The movement speed and azimuth of the transmitter are and The speed and azimuth of the receiving end are and The initial distance between the transmitter and the center of the smart reflector is , the initial distance from the center of the smart reflector to the receiving end is .also, and are the horizontal angle and elevation angle from the transmitter to the smart reflector, which can be calculated as and ,in is the four-quadrant inverse tangent function. Similarly, the horizontal angle and elevation angle from the receiving end to the smart reflector can be calculated as , .

[0108] The signal received by the receiving antenna consists of two parts: the signal reflected by the smart reflective surface and the signal not reflected by the smart reflective surface. Therefore, the channel impulse response can be expressed as:

[0109] ;

[0110] in, It can be further expressed as:

[0111]

[0112] in, is the wavelength, which is the ratio of the speed of light to the system carrier frequency. and are the path losses from the transmitter to the smart reflector and from the smart reflector to the receiver, and are the gains of the transmitting antenna and the receiving antenna respectively, is the angle between the incoming wave direction and the normal vector of the smart reflection surface, It is the angle between the reflected wave direction and the normal vector of the smart reflection surface. Indicates RU n The phase shift, and From the sender to the RU n and RU n The phase shift of the signal to the receiving end can be calculated as follows:

[0113] ;

[0114] ;

[0115] in, and From the sender to the RU n Distance and RU n The distance to the receiving end is calculated as follows:

[0116] ;

[0117] ;

[0118] in,

[0119] ；

[0120] ；

[0121] ；

[0122] ；

[0123] because , , , Established, among which, , using the approximate expression , and It can be expressed as follows:

[0124]

[0125] in, , , , , , .Will and Expression introduction In, we can get:

[0126]

[0127] in, , , . and is the Doppler shift caused by the motion of the transmitter and receiver, calculated as follows:

[0128] ;

[0129] ;

[0130] and is the Doppler shift caused by the relative motion between the drone and the transmitter and receiver, and is calculated as follows:

[0131] ;

[0132] ;

[0133] is the channel impulse response not affected by the smart reflector, expressed as follows:

[0134]

[0135] in, yes The path loss is , S is the number of clusters in the channel, Indicates The number of multipaths in a cluster, is the power of the sth cluster, , , , , , They represent the transmission distance, elevation departure angle, horizontal departure angle, elevation arrival angle and horizontal arrival angle of the i-th path in the s-th cluster respectively.

[0136] Furthermore, the statistical characteristics of the channel obtained based on the channel impulse response include: time correlation function, Doppler power spectrum density and system achievable rate. At time t, the time correlation function of the channel is calculated as follows:

[0137] ；

[0138] in, represents the mathematical expectation, represents the complex conjugate, represents the time interval. The Doppler power spectral density of the channel is obtained by Fourier transform of the time autocorrelation function and is calculated as follows:

[0139] ;

[0140] The achievable rate of the system can be calculated as follows:

[0141] ；

[0142] in, is the signal transmission power, is the noise power.

[0143] In order to further verify the effect of the present invention, the following comparative experiments were also conducted. Figure 3 , which shows the prediction results of the yaw angle of the drone using the LSTM model and the TCN-LSTM model proposed in the present invention and the comparison with the real data. The results show that the prediction results of TCN-LSTM are better than those of the LSTM model. Among them, the root mean square errors of the TCN-LSTM and LSTM models are 0.0103 and 0.0296 respectively. The TCN-LSTM model combines two different types of network structures. TCN effectively captures short-term local dependencies in time series through local convolution, enhancing the feature extraction ability of the model, while LSTM further processes long-term dependencies. This combination enables the TCN-LSTM model to better capture complex time series features, thereby improving prediction accuracy.

[0144] See also Figure 4 , which shows the impact of the UAV's flight altitude on the system's achievable rate. The ideal situation refers to the stable flight of the UAV, that is, ignoring the influence of random vibration. As the UAV's flight altitude increases, the path loss of the smart reflector's reflection channel increases significantly, and the smart reflector's ability to regulate the channel is weakened, resulting in a gradual decrease in the system's achievable rate. In the absence of a smart reflector, the system's achievable rate drops to the minimum. In addition, compared with the traditional phase design method, the phase design method proposed in this paper can more accurately predict the signal phase change caused by UAV vibration, making the system's achievable rate closer to the ideal state. This result further verifies the robustness and effectiveness of the proposed method under different flight altitude conditions, indicating that this method can significantly improve the system performance of the aerial smart reflector.

[0145] Embodiment 2

[0146] This embodiment discloses a channel modeling and phase design system for an aerial intelligent reflection surface;

[0147] like Figure 5 As shown, a channel modeling and phase design system for an aerial intelligent reflection surface includes:

[0148] The UAV random rotation angle generation module is configured to: establish a UAV random rotation angle model, and obtain the local coordinates of the reflection unit on the smart reflection surface that change with time based on the UAV random rotation angle;

[0149] The UAV rotation angle prediction module is configured to: establish a UAV deflection angle prediction model based on the acquired random rotation angle of the UAV using a deep learning time series prediction algorithm to predict the rotation angle of the UAV;

[0150] The smart reflective surface phase shift calculation module is configured to: calculate the phase shift of the smart reflective surface according to the predicted result of the rotation angle of the UAV, so as to optimize the coherent superposition of the signal at the receiving end;

[0151] The channel modeling module is configured to: establish a channel model assisted by an aerial intelligent reflection surface based on the local coordinates of the reflection unit on the intelligent reflection surface, the position, movement speed, movement direction, and system carrier frequency of the transmitter, receiver, and drone; obtain the channel impulse response based on the phase offset of the intelligent reflection surface; and obtain the statistical characteristics of the channel based on the channel impulse response.

[0152] Embodiment 3

[0153] The purpose of this embodiment is to provide a computer-readable storage medium.

[0154] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps in the channel modeling and phase design method of an aerial intelligent reflection surface as described in Example 1.

[0155] Embodiment 4

[0156] The purpose of this embodiment is to provide an electronic device.

[0157] An electronic device comprises a memory, a processor and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the channel modeling and phase design method of an aerial intelligent reflection surface as described in Example 1 are implemented.

[0158] The steps involved in the apparatuses of the above embodiments 2, 3 and 4 correspond to the method embodiment 1, and the specific implementation methods can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0159] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0160] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A channel modeling and phase design method for an aerial intelligent reflection surface, characterized in that: include: Step S1, establishing a random rotation angle model of the drone, and obtaining the local coordinates of the reflection unit on the smart reflection surface that change with time based on the random rotation angle of the drone; the process of establishing the random rotation angle model of the drone is: The random rotation angle of the drone is modeled by a sinusoidal random process, and the formula is: ； ； ； in, , and are the vibration amplitudes of yaw, pitch and roll angles, is the vibration frequency; Gauss-Markov is used to model the vibration amplitude of yaw, pitch and roll angles; the formula is as follows: ； ； ； in, , and Respectively represent the vibration amplitudes of the yaw angle, pitch angle and roll angle at the initial moment The value of , and Used to adjust randomness, between 0 and 1; , and is the asymptotic mean of the rotation angle, Equal to 0 or 1, used to control the coefficient of the noise term; Step S2: Based on the acquired random rotation angle of the drone, a drone deflection angle prediction model is established using a deep learning time series prediction algorithm to predict the rotation angle of the drone; the drone deflection angle prediction model adopts a time series convolution-long short-term memory network model, and the model includes: Input layer: used to receive the time series of three channels, which are used to input the historical data of yaw angle, pitch angle and roll angle changing over time; TCN layer: uses convolution operation to process the historical data of the input rotation angle changing over time, and extracts the local temporal features and short-term dependencies in the historical data of the yaw angle, pitch angle and roll angle changing over time; ReLU activation layer: The output of the TCN layer is transformed nonlinearly through the ReLU activation function, which introduces nonlinear characteristics to enhance the expressiveness of the model. LSTM layer: The LSTM layer captures long-term dependencies in time series data through a gating mechanism; Output layer: The output layer consists of a fully connected layer, and the output dimension of the fully connected layer is 3, corresponding to the prediction results of the yaw angle, pitch angle and roll angle respectively; through the fully connected layer, the high-dimensional time series features extracted by the LSTM layer are compressed and converted into the final prediction results; Step S3: Calculate the phase shift of the intelligent reflection surface according to the prediction result of the rotation angle of the UAV to optimize the coherent superposition of the signal at the receiving end; Step S4: Based on the local coordinates of the reflection unit on the intelligent reflection surface, the positions of the transmitter, the receiver and the drone, the movement speed of the drone, the movement direction of the drone, and the system carrier frequency, a channel model assisted by an aerial intelligent reflection surface is established, a channel impulse response is obtained based on the phase offset of the intelligent reflection surface, and statistical characteristics of the channel are obtained based on the channel impulse response.

2. The channel modeling and phase design method of an aerial intelligent reflection surface according to claim 1, characterized in that: include: The local coordinates of the reflection unit on the smart reflection surface that changes with time based on the random rotation angle of the drone are: The smart reflective surface is deployed on the drone and experiences random deflection as the drone time, The first Reflection unit, The local coordinates of can be calculated as follows: in, , , , are the x-axis and y-axis coordinates of the nth reflection unit in the reflection surface; and Respectively represent the number of columns and rows of reflection units; , , Indicates rounding up; and are the spacing between adjacent columns and rows, respectively; , and They are the rotation matrices corresponding to the yaw angle, pitch angle, and roll angle respectively.

3. The channel modeling and phase design method of an aerial intelligent reflection surface as claimed in claim 2, characterized in that: The phase shift of the smart reflector is: in, For The distance from the transmitter to the center of the smart reflective surface at any moment; For The distance from the center of the smart reflector to the receiving end at any moment; and They are the horizontal angle and elevation angle from the transmitter to the smart reflector surface; and They are the horizontal angle and elevation angle from the receiving end to the smart reflector surface; , , , , , are the parameters obtained in the process of calculating the transmission distance from the transmitter to the smart reflector and from the smart reflector to the receiver, respectively reflecting the influence of the yaw angle, pitch angle and roll angle of the smart reflector on the reflection phase; , , are the yaw angle, pitch angle and roll angle respectively. The predicted value at time.

4. The channel modeling and phase design method of an aerial intelligent reflection surface as claimed in claim 3, characterized in that: include: The channel impulse response includes a channel impulse response reflected by the smart reflective surface and a channel impulse response not affected by the smart reflective surface; wherein the channel impulse response reflected by the smart reflective surface is expressed as: ； in, is the wavelength, and Respectively represent the spacing between adjacent columns and rows of reflection units; and are the path losses from the transmitter to the smart reflector and from the smart reflector to the receiver, and are the gains of the transmitting antenna and the receiving antenna respectively, is the angle between the incoming wave direction and the normal vector of the smart reflection surface, is the angle between the reflected wave direction and the normal vector of the smart reflection surface; Indicates the initial distance between the transmitter and the center of the smart reflective surface; Indicates the initial distance between the receiving end and the center of the smart reflective surface; and It is the Doppler shift caused by the relative motion between the drone and the transmitter and receiver; and It is the Doppler shift caused by the motion of the transmitter and receiver; The channel impulse response without the influence of the smart reflector is expressed as follows: ； in, yes The path loss is , S is the number of clusters in the channel, Indicates The number of multipaths in a cluster, is the power of the sth cluster, , , , , , denote the transmission distance, elevation departure angle, horizontal departure angle, elevation arrival angle and horizontal arrival angle of the ith path in the sth cluster respectively; the movement speed and azimuth of the transmitter are and ; The speed and azimuth of the receiving end are and .

5. The channel modeling and phase design method of an aerial intelligent reflection surface as claimed in claim 4, characterized in that: include: The statistical characteristics of the channel include time correlation function, Doppler power spectrum density, and system achievable rate; The time correlation function is: ； in, represents the mathematical expectation, represents the complex conjugate, Indicates a time interval; The Doppler power spectral density is obtained by Fourier transform of the time autocorrelation function, and the formula is as follows: ； The system's achievable rate formula is as follows: ； in, is the signal transmission power, is the noise power.

6. A channel modeling and phase design system for an aerial intelligent reflector, characterized in that: include: The UAV random rotation angle generation module is configured to: establish a UAV random rotation angle model, and obtain the local coordinates of the reflection unit on the intelligent reflection surface that change with time based on the UAV random rotation angle; the process of establishing the UAV random rotation angle model is: The random rotation angle of the drone is modeled by a sinusoidal random process, and the formula is: ； ； ； in, , and are the vibration amplitudes of yaw, pitch and roll angles, is the vibration frequency; Gauss-Markov is used to model the vibration amplitude of yaw, pitch and roll angles; the formula is as follows: ； ； ； in, , and Respectively represent the vibration amplitudes of the yaw angle, pitch angle and roll angle at the initial moment The value of , and Used to adjust randomness, between 0 and 1; , and is the asymptotic mean of the rotation angle, Equal to 0 or 1, used to control the coefficient of the noise term; The UAV rotation angle prediction module is configured to: establish a UAV deflection angle prediction model based on the acquired random rotation angle of the UAV using a deep learning time series prediction algorithm to predict the rotation angle of the UAV; the UAV deflection angle prediction model adopts a time series convolution-long short-term memory network model, and the model includes: Input layer: used to receive the time series of three channels, which are used to input the historical data of yaw angle, pitch angle and roll angle changing over time; TCN layer: uses convolution operation to process the historical data of the input rotation angle changing over time, and extracts the local temporal features and short-term dependencies in the historical data of the yaw angle, pitch angle and roll angle changing over time; ReLU activation layer: The output of the TCN layer is transformed nonlinearly through the ReLU activation function, which introduces nonlinear characteristics to enhance the expressiveness of the model. LSTM layer: The LSTM layer captures long-term dependencies in time series data through a gating mechanism; Output layer: The output layer consists of a fully connected layer, and the output dimension of the fully connected layer is 3, corresponding to the prediction results of the yaw angle, pitch angle and roll angle respectively; through the fully connected layer, the high-dimensional time series features extracted by the LSTM layer are compressed and converted into the final prediction results; The smart reflective surface phase shift calculation module is configured to: calculate the phase shift of the smart reflective surface according to the predicted result of the rotation angle of the UAV, so as to optimize the coherent superposition of the signal at the receiving end; The channel modeling module is configured to: establish a channel model assisted by an aerial intelligent reflection surface based on the local coordinates of the reflection unit on the intelligent reflection surface, the positions of the transmitter, the receiver and the drone, the movement speed of the drone, the movement direction of the drone, and the system carrier frequency; obtain the channel impulse response based on the phase offset of the intelligent reflection surface; and obtain the statistical characteristics of the channel based on the channel impulse response.

7. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the steps in the channel modeling and phase design method of an aerial intelligent reflection surface as described in any one of claims 1 to 5 are implemented.

8. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the channel modeling and phase design method of an aerial intelligent reflection surface as described in any one of claims 1-5 are implemented.

Images