A method for modeling a three-dimensional MIMO wireless channel of a UAV air-to-air

Abstract

This invention discloses a method for modeling air-to-air 3D MIMO wireless channels for unmanned aerial vehicles (UAVs), relating to UAV communication, wireless channel modeling, and 6G air-to-ground integrated communication technologies. This method establishes a dual-UAV 3D MIMO scenario, calculates the probability of Loss of Surface (LoS) and probability-weighted path loss, adaptively generates general scattering clusters and rooftop scattering clusters based on UAV altitude, and establishes ground / roof scattering paths and their existence probabilities. The LoS, SR, and NLoS components are used to generate the channel impulse response, calculating path loss, TACF, RMS delay spread, and RMS angle spread. This method differs from the fixed scattering cluster model and can more accurately reflect urban A2A channels from low to mid-high altitudes.

Description

Technical Field

[0001] This invention relates to the fields of unmanned aerial vehicle (UAV) communication, wireless channel modeling, and 6G air-to-ground integrated communication technology, and in particular to a method for UAV air-to-air three-dimensional MIMO wireless channel modeling. Background Technology

[0002] Unmanned aerial vehicle (UAV) communication, including UAV air-to-ground communication and UAV air-to-air communication, is an important component of the 6G integrated air-space-ground communication network. It can be applied to scenarios such as emergency communication, intelligent transportation, low-altitude economy, inspection and monitoring, disaster relief, and military collaboration. With the development of multi-UAV collaborative missions and aerial self-organizing networks, air-to-air links between UAVs have become a key factor affecting system reliability, throughput, and networking capabilities.

[0003] Compared to air-to-ground links, air-to-air links are affected by the motion, altitude changes, three-dimensional attitude rotation of both the launching and receiving drones, as well as the urban built environment. In low-altitude urban canyon scenarios, building facades, the ground, and other obstacles generate rich scattering; in mid-to-high-altitude scenarios, the scattering and specular reflection from the rooftops of tall buildings on the propagation path cannot be ignored. The propagation mechanism changes at different altitudes, making it difficult to accurately describe the air-to-air channel across the entire altitude range using a single scattering cluster or a fixed NLoS assumption.

[0004] Existing UAV air-to-air channel models still have shortcomings: on the one hand, some models are mainly oriented towards open areas or low-altitude scenarios, lacking characterization of roof scattering from urban high-rise buildings; on the other hand, existing models usually treat specular reflection paths as deterministic components, without establishing existence probability models related to building occlusion, UAV altitude, and dynamic link geometry; in addition, although some models give path loss or Loss of Serving (LoS) probabilities, they do not simultaneously couple them to the large-scale fading and small-scale channel impulse response generation processes.

[0005] Therefore, there is an urgent need for a three-dimensional MIMO wireless channel modeling method that can cover low to medium-high altitudes, consider LoS, SR and dual-cluster NLoS propagation mechanisms, and output statistical characteristics such as path loss, time correlation, time delay dispersion and angular dispersion. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a method for modeling air-to-air three-dimensional MIMO wireless channels of UAVs, which can more accurately reflect the A2A channel of cities from low altitude to medium and high altitude.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0008] A method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles (UAVs) according to the present invention includes:

[0009] Step S1: Establish a three-dimensional air-to-air communication scenario between the launching UAV Tx and the receiving UAV Rx;

[0010] Step S2: Establish a stochastic urban building environment model in the three-dimensional air-to-air communication scenario established in step S1, and determine the building coverage ratio in the stochastic urban building environment model. Number of buildings per unit area Building height and its distribution scale parameters And the geometric relationship of the Tx-Rx link, calculate the probability of line-of-sight Loss (LoS) existence, and obtain the probability-weighted path loss from the LoS existence probability;

[0011] Step S3: Based on the three-dimensional air-to-air communication scenario established in Step S1 and the urban building stochastic environment model established in Step S2, establish a dynamic adaptive dual-cluster non-line-of-sight NLoS framework; wherein, the process of establishing the dynamic adaptive dual-cluster non-line-of-sight NLoS framework includes: determining the generation ratio of roof scattering clusters and general scattering clusters according to the relationship between Tx and Rx heights and building height distribution, and the roof scattering clusters and general scattering clusters forming NLoS paths.

[0012] Step S4: Based on the three-dimensional air-to-air communication scenario established in Step S1 and the urban building stochastic environment model established in Step S2, establish a ground or rooftop single-mirror reflection SR path model; calculate the SR path existence probability, mirror reflection point height, SR path propagation distance, propagation delay, departure angle, and arrival angle based on the ground or rooftop single-mirror reflection SR path model.

[0013] Step S5: Using the LoS existence probability obtained in step S2, the SR path existence probability obtained in step S4, the cluster birth and death process, and the UAV attitude rotation parameters in the three-dimensional air-to-air communication scenario established in step S1, a three-dimensional MIMO channel is generated; wherein, the three-dimensional MIMO channel includes the LoS path, the SR path, and the NLoS path formed in step S3.

[0014] A three-dimensional MIMO channel impulse response is generated based on the LoS component formed by the LoS path, the SR component formed by the SR path, and the NLoS component formed by the NLoS path.

[0015] Step S6: Calculate the statistical characteristics of the UAV air-to-air wireless channel based on the three-dimensional MIMO channel impulse response generated in step S5.

[0016] As a further optimization scheme for the UAV air-to-air 3D MIMO wireless channel modeling method described in this invention, the statistical characteristics of the UAV air-to-air wireless channel include path loss, time autocorrelation function, root mean square delay spread, and root mean square angle spread; the urban building stochastic environment model is the International Telecommunication Union (ITU) urban building stochastic environment model.

[0017] As a further optimization scheme for the UAV air-to-air 3D MIMO wireless channel modeling method described in this invention, in step S1, the 3D air-to-air communication scenario is represented by a Tx local coordinate system, an Rx local coordinate system, and a global coordinate system, and the multipath propagation geometry is determined based on the position, altitude, velocity, direction of motion, array structure, antenna polarization, and attitude rotation parameters of Tx and Rx; Tx configuration One transmit antenna element, Rx configuration There are n receiving antenna elements, and the p-th transmitting antenna element is denoted as . The q-th receiving antenna element is denoted as ,in , ;

[0018] The three-dimensional air-to-air communication scenario includes the LoS path from Tx to Rx, the SR path generated by the ground or roof, and the NLoS path generated by general scattering clusters and roof scattering clusters.

[0019] In the NLoS path, the nth scattering cluster is either a general scattering cluster or a roof scattering cluster, and includes the first hop scattering point. and last jump scattering point The m-th ray within the n-th scattering cluster has azimuth departure angle, elevation departure angle, azimuth arrival angle, elevation arrival angle, propagation delay, path power, and polarization phase; the total number of general scattering clusters and rooftop scattering clusters. and the number of rays within the nth scattering cluster Changes over time.

[0020] As a further optimization of the UAV air-to-air 3D MIMO wireless channel modeling method described in this invention, in step S2, the urban building stochastic environment model, the building height... It follows a Rayleigh distribution, and its probability density function is: ,in, , Indicates building height The probability density function, The scale parameter represents the building height distribution; the building location follows a Poisson process; the probability of building height below the Tx-Rx link line is calculated and combined with the expected number of buildings on the link. Obtain the probability of Loss ;

[0021] Probability-weighted path loss ,in, This represents the path loss in the Loss of Path (LoS) state. This represents the path loss under NLoS conditions.

[0022] As a further optimization scheme for the UAV air-to-air 3D MIMO wireless channel modeling method described in this invention, in step S3, the dynamic adaptive dual-cluster non-line-of-sight (NLoS) framework, let... Given the minimum values ​​of Tx height and Rx height at time t, the theoretical proportion of roof scattering clusters in the generation ratio of roof scattering clusters and general scattering clusters is: ;when When the height is lower than the average height of urban buildings, The corresponding number of general scattering clusters is greater than the number of rooftop scattering clusters; when the minimum values ​​of Tx height and Rx height at time t are not less than the average height of urban buildings, The number of corresponding roof scattering clusters is not less than the number of general scattering clusters, and the roof scattering clusters participate in NLoS channel generation;

[0023] The height of the roof scattering cluster is generated based on the building height distribution. The azimuth, pitch, path delay, and path power of the roof scattering cluster are generated according to the geometric random channel model. General scattering clusters are used to characterize scattering from building facades, ground, and obstacles, while roof scattering clusters are used to characterize scattering from the roof of high-rise buildings.

[0024] As a further optimization scheme for the UAV air-to-air three-dimensional MIMO wireless channel modeling method described in this invention, in step S4, the probability of SR path existence in the ground or rooftop single specular reflection SR path model is... The probability of roof illumination or the probability that the reflection point is below the minimum height of the transmitting UAV Tx and the receiving UAV Rx is determined by the expected number of buildings on the link; when an SR path exists, the height of the specular reflection point, the SR path propagation distance, the propagation delay, the departure angle, and the arrival angle are calculated, wherein the height of the specular reflection point... Generated according to the mixture probability density function, where Corresponding to ground mirror reflection, Corresponding to the reflection of the roof mirror.

[0025] As a further optimization of the UAV air-to-air 3D MIMO wireless channel modeling method described in this invention, in step S5, the 3D MIMO channel impulse response is determined by the 3D MIMO channel matrix. Represented as ,in For the shadow to fade, This is a small-scale fading matrix;

[0026] The channel impulse response from the p-th transmit element to the q-th receive element in the three-dimensional MIMO channel impulse response. From the Los component SR components and NLoS components The weighted summation is obtained;

[0027] ,in Rice factor, For SR path power coefficient, Let be the total power coefficient of the NLoS path, and ;

[0028] The result is obtained by multiplying the transmit and receive antenna patterns and the polarization matrix, where Let m be the time-varying power of the m-th ray within the n-th cluster. To correspond to the time-varying propagation delay, For carrier frequency, The polarization phase is given by j, which is the imaginary unit. It is the Dirac function;

[0029] Where c is the speed of light; and These represent the mapping distance of the p-th transmitting element caused by the attitude rotation of the launching UAV Tx and the mapping distance of the q-th receiving element caused by the attitude rotation of the receiving UAV Rx, respectively. for to distance, for to The distance; for and Virtual link latency between; , , and These represent the p-th transmitting antenna element, the first hop scattering point, the last hop scattering point, and the q-th receiving antenna element, respectively.

[0030] As a further optimization scheme for the UAV air-to-air 3D MIMO wireless channel modeling method described in this invention, the cluster generation and extinction process in step S5 includes: according to the sampling time interval Calculate the survival probability of existing scattering clusters And based on the cluster generation rate Cluster recombination rate Calculate the expected number of new scattering clusters .

[0031] As a further optimization of the UAV air-to-air 3D MIMO wireless channel modeling method described in this invention, in step S6, the time autocorrelation function is the channel transfer function from the p-th transmit antenna element to the q-th receive antenna element. With the The transmitting antenna array element to the first Each receiving antenna element in time, Channel transfer function at frequency The correlation between them is obtained; the correlation characterizes the channel response of the two transmit / receive antenna array elements in the time interval. Frequency interval and spatial interval The degree of similarity, among which, and Let these represent the indexes of the other transmitting antenna element and the other receiving antenna element, respectively; let the frequency interval be... and spatial interval When the value is zero, the time autocorrelation function TACF is obtained.

[0032] As a further optimization scheme for the UAV air-to-air 3D MIMO wireless channel modeling method described in this invention, in step S6, the root mean square angle expansion... Root mean square delay spread ,in and These are the power-weighted average angle and the power-weighted average time delay, respectively. For the i-th path from the p-th transmit antenna element to the q-th receive antenna element, at angle... The relative power contribution at that location Let be the angle of the i-th path; For the path of this array element pair in time delay The relative power contribution at that location This refers to path delay.

[0033] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects:

[0034] (1) In the low-altitude urban canyon scene, the general scattering cluster is retained, and the roof scattering cluster is introduced in the mid-to-high altitude scene, so as to realize the adaptive modeling of the air-to-air channel of the UAV at all altitudes.

[0035] (2) Establish existence probability models for LoS path and SR path respectively, and apply the probability models to both path loss and channel impulse response generation process simultaneously;

[0036] (3) By incorporating ground / roof specular reflection, roof scattering, general scattering, cluster formation and extinction processes and UAV attitude rotation into the three-dimensional MIMO GBSM, a more accurate wireless channel model can be provided for UAV link budget, aerial networking and 6G air-space-ground integrated system simulation. Attached Figure Description

[0037] Figure 1 This is a flowchart illustrating a method for modeling an air-to-air 3D MIMO wireless channel for unmanned aerial vehicles (UAVs) provided in this embodiment.

[0038] Figure 2 This is a schematic diagram of the UAV air-to-air three-dimensional MIMO geometric random channel model provided in this embodiment;

[0039] Figure 3 This is a comparison chart of the path loss of the model proposed in this embodiment and the 3GPP TR 36.777 standard model at different altitudes and horizontal distances.

[0040] Figure 4 This is a comparison chart of the root mean square delay spread (CDF) of the model, measurement data, and 3GPP TR 38.901 standard model proposed in this embodiment;

[0041] Figure 5 This is a graph showing the time autocorrelation function results under different channel components and different motion directions in this embodiment;

[0042] Figure 6 This is a diagram showing the root mean square angle extended CDF results for different launch drone altitudes and Rice factors in this embodiment. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] The technical problem to be solved by this invention is to overcome the shortcomings of existing UAV air-to-air channel models in terms of the random existence of urban building roof scattering, specular reflection and full-height adaptive modeling capabilities, and to provide a UAV air-to-air three-dimensional MIMO wireless channel modeling method.

[0045] Figure 1This is a flowchart illustrating a UAV air-to-air three-dimensional MIMO wireless channel modeling method provided in this embodiment. In this embodiment, Tx represents the transmitting UAV, Rx represents the receiving UAV, MIMO represents Multiple-Input Multiple-Output, GBSM represents Geometric Random Channel Model, A2A represents Air-to-Air Communication, LoS represents Line-of-Sight Path, NLoS represents Non-Line-of-Sight Path, SR represents Specular Path, PL represents Path Loss, SH represents Shadowing Fading, CIR represents Channel Impulse Response, TACF represents Time Autocorrelation Function, RMS represents Root Mean Square Function, and CDF represents Cumulative Distribution Function. Indicates building height, This represents the lower of Tx and Rx at time t. Indicates the building coverage ratio. This indicates the number of buildings per unit area. This represents the scale parameter indicating the distribution of building height. Indicates the carrier frequency. denoted by , where represents the propagation delay, and c represents the speed of light.

[0046] Symbol and parameter descriptions

[0047] To avoid ambiguity in the meaning of symbols in the formulas, this embodiment provides a unified explanation of the main symbols and parameters used in subsequent modeling. Unless otherwise specified, quantities with the subscript p are related to the p-th transmitting antenna element or the p-th link, quantities with the subscript q are related to the q-th receiving antenna element or the q-th link, and quantities with the subscripts n and m are related to the n-th scattering cluster and the m-th ray within that cluster, respectively. The meaning and purpose of each symbol and parameter are shown in Table 1.

[0048] Table 1

[0049]

[0050] Step S1: Establish a three-dimensional air-to-air communication scenario

[0051] Step S101, see Figure 2 Establish local coordinate systems Tx, Rx, and global coordinate systems to describe the three-dimensional motion of Tx / Rx, the antenna array attitude, and the geometric relationship of multipath propagation. Figure 2 In the Tx configuration, "scattering clusters" refers to general scattering clusters, and "rooftop clusters" refers to rooftop scattering clusters. One transmit antenna element, Rx configuration There are n receiving antenna elements, and the p-th transmitting antenna element is denoted as . The q-th receiving antenna element is denoted as ,in , .

[0052] Step S102: Set the initial position, altitude, velocity, and direction of motion for Tx and Rx. Both Tx and Rx can undergo three-dimensional attitude rotation, which is described by yaw angle, pitch angle, and roll angle, and is used to calculate the propagation distance offset of the array element relative to the reference point.

[0053] Step S103: Divide the propagation path between Tx and Rx into LoS ​​path, SR path, and NLoS path. The LoS path is the direct path between Tx and Rx; the SR path is the single specular reflection path generated by the ground or building roof; the NLoS path consists of general scattering clusters and roof scattering clusters.

[0054] Step S104: For the nth NLoS scattering cluster, Indicates the first scattering point. Indicates the final scattering point. This represents the number of rays in the cluster at time t. The m-th ray within the n-th cluster has azimuth departure angle, elevation departure angle, azimuth arrival angle, elevation arrival angle, and time-varying path power. and time-varying propagation delay .

[0055] Step S105: In the global coordinate system, the position vectors of Tx and Rx are respectively represented as... and .

[0056] in, , , These represent the x, y, and z coordinates of Tx in the global coordinate system, respectively. , , These represent the x, y, and z coordinates of Rx in the global coordinate system, respectively.

[0057] When using the uniform motion approximation, the position update can be written as:

[0058]

[0059] in, and Let Tx and Rx represent the velocity vectors, respectively. This indicates the sampling time interval. This update is used to describe the relative position changes of the two UAVs, thereby further updating the LosS distance, SR path distance, NLoS cluster angle, and path delay.

[0060] Step S106: The position of the array element is determined by the position of the platform reference point and the local coordinates of the array. Let... Let p be the position of the p-th transmitting element in the Tx local coordinate system. Let q be the position of the q-th receiving array element in the Rx local coordinate system. and Let the rotation matrices be determined by the yaw, pitch, and roll angles, respectively. Then the global position of the array element can be represented as:

[0061]

[0062] The rotation matrix is ​​used to map changes in the UAV platform's attitude to changes in the antenna array element positions. Through this process, the model can characterize the impact of UAV attitude changes on array phase, path delay, and angular parameters.

[0063] Step S107, the LosS path distance is directly obtained from the positions of the transmitting and receiving array elements:

[0064]

[0065] The azimuth departure angle, pitch departure angle, azimuth arrival angle, and pitch arrival angle corresponding to the LoS path are given by vectors. The angle is calculated by projection into the local coordinate systems Tx and Rx. This angle calculation result is also used to generate the antenna pattern, array response, and polarization response.

[0066] Step S108: The NLoS scattering cluster is described by the cluster center, intra-cluster ray angular spread, time delay spread, and power attenuation. For the nth cluster, if its first hop scattering point and last hop scattering point are respectively... and The propagation distance of the m-th ray is determined by... arrive , arrive Virtual links and arrive It consists of three parts. It preserves virtual link latency. This model can represent both multi-hop cluster propagation and... It degenerates into a single-hop scattering model.

[0067] Step S2: Loss probability and probability-weighted path loss

[0068] Step S201: In the three-dimensional air-to-air communication scenario established in step S1, a stochastic environment model of urban buildings is established. The building positions follow a Poisson point process, and the building heights... It follows a Rayleigh distribution, and its probability density function is:

[0069]

[0070] in, This represents the building height distribution scale parameter. α represents the building coverage ratio, β represents the number of buildings per unit area, and D represents the horizontal distance between Tx and Rx. The desired number of buildings on the link. It is determined by α, β and D.

[0071] Step S202: For any normalized position on the Tx-Rx link The link height is represented as ,in For the height of Tx, Let Rx be the height. The probability that a single building does not obstruct the LoS path can be obtained from the probability that the building height is lower than the straight-line height of the link:

[0072]

[0073] Assuming that the buildings on the link are independent of each other, the probability of Loss of Space (LoS) between Tx and Rx is expressed as:

[0074]

[0075] Step S203: Construct a probability-weighted path loss model based on the probability of LoS existence. Let... For path loss under Loss conditions, Given the path loss under NLoS conditions, the probability-weighted path loss is:

[0076]

[0077] The above model makes the path loss no longer determined by a single LosS or NLoS state, but changes continuously with the drone's altitude, horizontal distance, and the probability of occlusion by urban buildings.

[0078] Step S204, Expected number of buildings This represents the number of buildings that may cause shading near the Tx-Rx horizontal link. It references random building shading modeling methods. It can be calculated from the building coverage ratio, the number of buildings per unit area, and the horizontal distance Tx-Rx, and can be written in the form of:

[0079]

[0080] Where D is the projected distance between Tx and Rx on the horizontal plane. This formula shows that when the horizontal distance between Tx and Rx increases or urban buildings become more densely packed, the number of potential obstructing buildings that the link passes through increases, and the probability of a Loss of Space (LoS) path decreases.

[0081] Step S205, the LoS path loss and NLoS path loss can be expressed using the corresponding expressions in the standard channel model, or they can be obtained by fitting based on actual measurement results. This embodiment preferably uses the air-to-air / air-to-ground standard path loss as... and Basic model, and then through A probability weighting method is applied. This preserves the engineering usability of the standard model while introducing the randomness of urban building occlusion.

[0082] Step S206: The probability-weighted path loss is not only used in the link budget, but also incorporated into the large-scale fading term of the three-dimensional MIMO channel matrix. Unlike the approach of using the LoS probability only in the large-scale path loss, this invention also uses the LoS probability in the small-scale CIR generation stage to control whether the LoS component participates in the channel response, thereby achieving consistent modeling at both large and small scales.

[0083] Step S3: Dynamic Adaptive Dual-Cluster NLoS Framework

[0084] Step S301: Based on the three-dimensional air-to-air communication scenario established in Step S1 and the urban building stochastic environment model established in Step S2, a dynamic adaptive dual-cluster non-line-of-sight (NLoS) framework is established, simultaneously setting up a general scattering cluster and a roof scattering cluster in the NLoS channel. The general scattering cluster is used to characterize the scattering caused by building facades, ground, and obstacles in the low-altitude urban canyon scenario; the roof scattering cluster is used to characterize the scattering caused by the roofs of high-rise buildings in the mid-to-high-altitude scenario.

[0085] Step S302, let The probability that the roof is illuminated by Tx and Rx, or the theoretical proportion of the roof scattering cluster, is expressed as:

[0086]

[0087] Step S303, according to Determine the number of rooftop scattering clusters and the number of general scattering clusters ,For example , When Tx or Rx is lower than the average building height, The NLoS component is relatively small, mainly composed of general scattering clusters; when both Tx and Rx are located in the mid-to-high altitude, As the number of rooftop scattering clusters increases, they participate in channel generation.

[0088] Step S304: For rooftop scattering clusters, their height is generated by the building height distribution, and their azimuth, pitch, path delay, and path power are generated according to a geometrical random channel model; for general scattering clusters, their parameters are generated according to the statistical distribution of the low-altitude scattering environment. Therefore, the dual-cluster NLoS framework can automatically switch propagation mechanisms based on the UAV's altitude.

[0089] Step S305: The rooftop scattering cluster is not fixed in all scenes, but is caused by... Controls whether to generate and how much to generate. When At lower levels, Tx or Rx is located between buildings, and the visibility of the link to the roof is weak; NLoS energy mainly comes from building facades, the ground, and discrete obstacles. At higher values, both Tx and Rx have a greater probability of illuminating the building roof, and the roof scattering cluster begins to become a non-negligible NLoS propagation component.

[0090] Step S306: If the total number of NLoS clusters is It can be classified into roof scattering clusters and general scattering clusters using the following formula:

[0091]

[0092] in, Indicates the number of roof scattering clusters. This represents the number of general scattering clusters. Rounding can be used to obtain an integer number of clusters in the simulation, or Bernoulli sampling can be used to determine whether each cluster is a roof scattering cluster.

[0093] Step S307: For rooftop scattering clusters, the cluster center height can be determined from the conditional distribution. Extraction is performed from the middle; for general scattering clusters, the cluster center can be set near building facades, roads, or other obstacles, depending on the low-altitude urban scattering environment. Both types of clusters retain their respective angular spread and time delay spread parameters, thus enabling the simultaneous description of low-altitude rich scattering and mid-to-high-altitude roof scattering.

[0094] Step S308, Number of rays within the cluster It can be generated according to the Poisson distribution, that is ,in The average number of rays. The relative power of each ray is allocated according to the cluster power and the power normalization coefficient within the cluster, satisfying that the sum of the powers of all rays within the same cluster equals the power of that cluster.

[0095] Step S4: SR path and SR existence probability

[0096] Step S401: Based on the 3D air-to-air communication scenario established in Step S1 and the urban building stochastic environment model established in Step S2, establish a single specular reflection (SR) path on the ground or roof. The SR path originates from Tx, and reaches Rx after reflection from the ground or building roof. If the height of the specular reflection point... Then the SR path is a ground specular reflection; if If so, then the SR path is a roof mirror reflection.

[0097] Step S402, SR path existence probability Determined by the probability that the reflection point is within the visible height range and the expected number of buildings on the link, it can be expressed as:

[0098]

[0099] Step S403: When the SR path exists, the mixture probability density function of the mirror reflection point height hSR is:

[0100]

[0101]

[0102] Step S404: After determining the height and position of the mirror reflection point, calculate the SR path propagation distance. Propagation delay , as well as the azimuth departure angle, pitch departure angle, azimuth arrival angle, and pitch arrival angle corresponding to the SR path.

[0103] Step S405: The SR path propagation distance can be obtained by adding the distances from the transmitting element to the mirror reflection point and from the mirror reflection point to the receiving element. Let the position of the mirror reflection point be... ,but:

[0104]

[0105]

[0106] in, It can be determined by the law of specular reflection and the height of the reflecting plane. When reflecting from the ground, the height of the reflecting plane is 0; when reflecting from the roof, the height of the reflecting plane is... .

[0107] Step S406: Whether the SR path participates in CIR generation is determined by... Control. In actual simulation, uniform random variables can be generated. ,like If the SR component is generated, then the SR component power is set to zero and the corresponding power is reassigned to the NLoS component or processed according to the normalization coefficient.

[0108] Step S407, the SR path, and the LoS path all possess relatively strong determinism, but they are affected differently by the urban environment. The LoS path is mainly affected by building obstruction, while the SR path also depends on the height of the reflecting plane, the visibility of the reflecting point, and the Tx / Rx height. Therefore, this invention establishes... and Instead of simply viewing the SR path as a fixed, deterministic component.

[0109] Step S5: Generation of three-dimensional MIMO channel impulse response

[0110] Step S501: Using the LoS existence probability obtained in step S2, the SR path existence probability obtained in step S4, the cluster birth and death process described in step S507, and the UAV attitude rotation parameters in the three-dimensional air-to-air communication scenario established in step S1, a three-dimensional MIMO channel is generated; wherein, the three-dimensional MIMO channel includes a LoS path, an SR path, and an NLoS path formed in step S3; the three-dimensional MIMO channel impulse response is generated according to the LoS component, SR component, and NLoS component corresponding to the LoS path, SR path, and NLoS path. Let H be a complex channel matrix containing large-scale fading and small-scale fading, then:

[0111]

[0112] in, The probability-weighted path loss obtained in step S2, For the shadow to fade, This is the small-scale fading matrix. Each launch array element to receiving array element The CIR composition between them.

[0113] Step S502, the CIR between the p-th transmitting element and the q-th receiving element is represented as follows:

[0114]

[0115] in, Rice factor, For SR path power coefficient, Let be the total power coefficient of the NLoS path, and . , and These are the LoS, SR, and NLoS components, respectively.

[0116] Step S503: The Loss component is generated from the direct path from Tx to Rx, and its delay... , for and The LoS distance between them. The LoS component includes the transmit and receive antenna patterns, vertical / horizontal polarization phase, and carrier frequency. and Dirac function .

[0117] The LoS component can be further written in the following LaTeX form:

[0118]

[0119] in, The polarization pattern vector of the receiving array element in the direction of the LoS angle of arrival. Let be the polarization pattern vector of the transmitting array element in the direction of the LoS departure angle. This is the LoS path polarization phase matrix. This equation is used to simultaneously describe the array directivity, polarization response, propagation phase, and path delay.

[0120] Step S504: The SR component is generated by reflection from the ground or roof mirrors, and its time delay... The SR component includes the departure angle, arrival angle, transmit / receive antenna pattern, polarization phase, and carrier frequency of the SR path. and Dirac function .

[0121]

[0122] in, This is the polarization phase matrix of the SR path. If the SR path does not exist, then... If an SR path exists, determine whether it is ground reflection or roof reflection based on the height of the reflection point, and calculate the corresponding propagation angle.

[0123] Step S505, the NLoS component is generated jointly by the general scattering cluster and the roof scattering cluster, and its expression can be written as:

[0124]

[0125] in, Let m be the time-varying power of the m-th ray within the n-th cluster. This refers to the time-varying propagation delay of the ray. This represents the initial phase related to polarization. In actual calculations, this formula is also multiplied by the transmit and receive antenna patterns and the polarization matrix to obtain the complete three-dimensional MIMO polarization channel response.

[0126] More completely, each ray in the NLoS component can be multiplied by the transmitter pattern, receiver pattern, and polarization coupling matrix, in the form of:

[0127]

[0128] in, It includes four polarization coupling phases: vertical-vertical, vertical-horizontal, horizontal-vertical, and horizontal-horizontal, which can describe the cross-polarization power ratio and random phase in the polarization channel.

[0129] Step S5051, Path Power It can be obtained from cluster power, intra-cluster ray power, and power normalization coefficient. For the nth cluster, the cluster power is first generated based on the cluster time delay and shadow term. Then allocate the cluster power to A ray. The normalized power of all NLoS rays satisfies:

[0130]

[0131] If SR paths also exist, then the total NLoS power is multiplied by [the factor]. SR path power multiplied by This ensures that the total power of the SR and NLoS components is controllable.

[0132] Step S5052: The angle parameters are calculated from the geometric position and superimposed with the cluster interior angle extension. Taking the azimuth departure angle as an example, the center azimuth departure angle of the nth cluster is from Tx to... The direction vector is determined, and the azimuth departure angle of the m-th ray is generated by random perturbation near the cluster center angle. The pitch departure angle, azimuth arrival angle, and pitch arrival angle are obtained using the same method.

[0133] Step S5053: Unlike channel models that only apply to fixed heights or fixed scattering mechanisms, in this invention, LoS, SR, rooftop scattering clusters, and general scattering clusters can all be updated over time. When Tx and Rx move, causing... Horizontal distance When the visibility of the reflection point or the visibility of the scattering cluster changes, All of these change accordingly, thus giving CIR dynamic adaptive capabilities.

[0134] Step S506: Considering the UAV's attitude rotation, the propagation delay of the m-th ray within the n-th cluster is:

[0135]

[0136] in, and These represent the element mapping distances caused by the attitude rotations of Tx and Rx, respectively. for to distance, for to distance, for and The latency of the virtual link between them. When At that time, the multi-hop cluster model degenerates into a single-hop cluster model.

[0137] Step S507: Introduce cluster birth and death processes to characterize the channel nonstationarity caused by the high-speed movement of the UAV. Let... This represents the survival probability of existing clusters within the sampling interval Δt. Cluster generation rate, Given the cluster recombination rate, the expected number of newly formed scattering clusters is:

[0138]

[0139] The position, angle, time delay, and power of the scattering clusters are updated based on the surviving and newly formed clusters, thereby obtaining the channel impulse response at the next moment.

[0140] Step S508: In the actual simulation implementation, the CIR generation process at a sampling time includes: First, updating the positions and orientations of Tx and Rx; Second, calculating based on the building environment parameters. , and Third, determine whether Loss and SR exist based on random sampling; Fourth, based on Fifth, assign rooftop scattering clusters and general scattering clusters; sixth, generate the angle, power, time delay, and polarization phase of the rays within each cluster; seventh, obtain the results by superimposing the above LoS, SR, and NLoS formulas. Seventh, repeat this process for all p and q to form a complete three-dimensional MIMO channel matrix.

[0141] Step S509: The CIR output by the method can be directly used for link-level simulation and system-level simulation. In link-level simulation, it can be based on... Calculate received signal, channel capacity, bit error rate, or pilot estimation performance; in system-level simulation, evaluate UAV formation networking, resource allocation, beam design, and multi-antenna transmission schemes based on path loss, delay spread, and angle spread.

[0142] Step S6: Calculation of channel statistical characteristics

[0143] Step S601: Obtain the channel transfer function by performing a Fourier transform on the time delay τ using the CIR. The space-time-frequency correlation function is expressed as:

[0144]

[0145] make and At that time, the space-time-frequency correlation function degenerates into the time autocorrelation function TACF, which is used to describe the correlation of the channel over time.

[0146] Step S602: The root mean square angle extension is used to describe the distribution of signal energy in the angular domain, and its expression is:

[0147]

[0148]

[0149] in, For the i-th path at angle The relative power contribution on The angle is the power-weighted average.

[0150] Step S603: The root mean square delay spread is used to describe the distribution of signal energy in the time delay domain, and its expression is:

[0151]

[0152]

[0153] in, This represents the relative power contribution of the path at time delay τ. This is the power-weighted average time delay.

[0154] Step S604: The path loss statistical characteristics are obtained from the probability-weighted path loss in step S2. By changing the Tx height, Rx height, and horizontal distance, path loss curves at different heights and distances can be obtained and compared with the LoS and NLoS path loss results of the standard channel model.

[0155] Step S605, TACF is used to reflect the correlation of the channel in the time domain. If the TACF decays rapidly, it indicates that the channel changes drastically over time, requiring higher frequency channel estimation and stronger time-varying tracking capabilities; if the TACF is high, it indicates that the Loss or SR components account for a large proportion, and the channel is relatively stable.

[0156] Step S606, RMS delay spread is used to characterize multipath delay dispersion. If A larger frequency selectivity results in stronger channel frequency selectivity, requiring a longer cyclic prefix or finer equalization; if... If the multipath energy is smaller, the multipath energy is more concentrated, and the link can achieve more stable broadband transmission performance.

[0157] Step S607, RMS angle expansion is used to characterize the degree of multipath angular dispersion. If A larger value results in richer spatial multipath propagation, which can affect MIMO spatial multiplexing, beamforming, and angle estimation; if... If the beam is smaller, the energy is concentrated in a few directions, making it suitable for narrow beam tracking and directional transmission.

[0158] Step S608: By comparing the statistical characteristics of components containing only NLoS, components containing LoS+NLoS, and components containing LoS+SR+NLoS, the contributions of the LoS path, SR path, and roof scattering cluster to the air-to-air channel can be quantified respectively. This comparison can be used to verify the necessity of introducing the SR path and roof scattering cluster in this invention.

[0159] Example Parameters and Result Analysis

[0160] This embodiment uses the following simulation parameters to verify the proposed channel model. The simulation parameters are shown in Table 2.

[0161] Table 2

[0162]

[0163] See Figure 3 Path loss (dB) represents the path loss (dB), horizontal distance D (m) represents the horizontal distance D (m), and Proposed Model represents the proposed model. At the same horizontal distance, as the drone's altitude increases, the path loss obtained by the proposed model is closer to the Loss of Path (LoS) result, indicating that the probability of LoS increases at higher altitudes. With increasing horizontal distance, building occlusion intensifies, and the path loss gradually transitions towards an Non-LoS trend. This result demonstrates that the probability-weighted path loss model described in this invention can reflect the changes in propagation loss under the combined effects of drone altitude and urban occlusion.

[0164] See Figure 4 CDF represents the cumulative distribution function, Delay spread represents the root mean square delay spread, Proposedmodel (simulation) represents the proposed model (simulation), and Measurement represents the measured data. The root mean square delay spread (CDF) of the proposed model is close to the measured data, and it can more accurately describe the air-to-air channel delay dispersion in high-rise urban environments compared to the 3GPP TR 38.901 standard model. This is because the present invention concentrates part of the NLoS power to the SR path and characterizes mid-to-high-altitude propagation through roof scattering clusters, thereby reducing the overestimation of multipath dispersion.

[0165] See Figure 5TACF represents the time autocorrelation function, Time difference Δt(s) represents the time difference Δt(s), and horizontal movement and vertical movement represent horizontal and vertical motion, respectively. When LoS, SR, and NLoS components are included, the channel time autocorrelation function is significantly higher than that of the case containing only the NLoS component, indicating that the LoS and SR components have stronger stability. In the case of only NLoS, the time correlation in the vertical motion direction decays faster, indicating that vertical motion leads to more pronounced Doppler spread and faster cluster formation and extinction processes.

[0166] See Figure 6 Here, CDF represents the cumulative distribution function, RMS angular spread represents the root mean square angle spread, hT represents the launch drone's Tx altitude, and K(t) represents the Rice factor. At higher launch drone altitudes, the probability of roof scattering clusters and SR paths increases, and the RMS angle spread changes significantly. When the Rice factor decreases, the power proportion of the NLoS component increases, the multipath effect intensifies, and the angle spread increases accordingly. This result verifies the impact of roof scattering clusters, SR paths, and their existence probability models on the statistical characteristics of channel angles in this invention.

[0167] In summary, the three-dimensional MIMO wireless channel modeling method provided by this invention can adaptively characterize the LoS, SR, and NLoS propagation mechanisms under different UAV altitudes and urban building environments. It can calculate and analyze key statistical characteristics such as path loss, time correlation, delay spread, and angle spread. It can be used for UAV air-to-air communication system simulation, link budget, low-altitude network planning, and 6G air-space-ground integrated communication technology research.

[0168] Compared with existing fixed scattering cluster UAV channel models, this invention has at least the following differences. First, existing models typically only set up general scattering clusters in low-altitude scenarios, making it difficult to explain the angular spread changes caused by building roofs in mid-to-high-altitude scenarios; this invention, through... This invention links building height distribution with UAV altitude, enabling the generation of rooftop scattering clusters to adaptively change with altitude. Secondly, existing models often treat SR paths as fixed, deterministic paths, failing to reflect the disappearance of SR paths caused by building occlusion and changes in link geometry; this invention addresses this by... This invention performs probabilistic modeling of the SR path and distinguishes between ground-based SR and rooftop SR. Third, existing path loss models often treat LosS / NLoS states separately; this invention addresses this by... A probability-weighted path loss is generated, and the existence of the Loss component is controlled synchronously during CIR generation, thereby maintaining consistency between large-scale and small-scale modeling.

[0169] Compared to large-scale path loss models that only consider the Loss of Sight (LoS) probability, this invention further incorporates the LoS probability, SR probability, and roof cluster ratio into small-scale channel generation. Specifically, Determine whether the Loss of Path (LoS) component exists and its contribution to path loss. Determine whether the ground / roof SR component exists. The number of roof scattering clusters and their power contribution are determined. These three factors work together to enable the channel model to output not only the average path loss, but also small-scale statistical characteristics such as TACF, RMS delay spread, and RMS angular spread.

[0170] The present invention can also be implemented in the following alternative ways. If the application scenario is a suburban or open area, the building coverage ratio can be reduced. and number of buildings per unit area ,at this time rise, The model automatically degenerates into a Loss-dominant air-to-air channel if the signal is reduced. For applications in densely populated, high-rise urban areas, the signal can be increased. , and At this point, increased building obstruction makes rooftop scattering clusters and rooftop SR paths more significant in affecting channel statistical characteristics. If the application scenario is a low-altitude urban canyon, the Tx or Rx height can be set below the average building height, and the model will use general scattering clusters as the dominant NLoS component.

[0171] The model parameters output by this invention can be used in various subsequent communication algorithms. For beam tracking algorithms, the arrival and departure angles of each path (LoS / SR / NLoS) can be used as angular domain priors; for OFDM system design, the parameters can be used... Determine the cyclic prefix length and subcarrier spacing; for MIMO precoding, you can use... Spatial multiplexing capability can be evaluated using the channel matrix H; for link budget, the received power can be calculated using probability-weighted path loss PL and shadow fading SH; for UAV networking simulation, the channel estimation update period can be evaluated using TACF.

[0172] Within a complete simulation cycle, the method of this invention can be executed in the following order: input urban environmental parameters, UAV trajectory parameters, antenna array parameters, and carrier parameters; update Tx / Rx position and attitude at each sampling time; calculate the LoS probability, SR probability, and roof scattering cluster ratio; sample to generate LoS, SR, roof scattering clusters, and general scattering clusters; calculate the distance, delay, angle, power, and polarization phase of each path; superimpose to obtain the CIR; calculate the channel transfer function and statistical characteristics from the CIR; output path loss, TACF, RMS delay spread, RMS angle spread, and corresponding simulation curves.

[0173] The parameters in this invention can be determined by a standard model, measured data, or scene map. Building coverage ratio. and number of buildings per unit area Building height parameters can be obtained from city maps or statistical databases. Rice factor can be obtained by fitting a sample of building heights. The SR power coefficient can be obtained from measurement data or empirical models. Cluster generation rate can be determined by reflection intensity measurement or ray tracing results. Cluster recombination rate The speed of the drone and the dynamics of the environment can be adjusted. Therefore, this invention has strong configurability and engineering applicability.

[0174] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles (UAVs), characterized in that, include: Step S1: Establish a three-dimensional air-to-air communication scenario between the launching UAV Tx and the receiving UAV Rx; Step S2: Establish a stochastic urban building environment model in the three-dimensional air-to-air communication scenario established in step S1, and determine the building coverage ratio in the stochastic urban building environment model. Number of buildings per unit area Building height and its distribution scale parameters And the geometric relationship of the Tx-Rx link, calculate the probability of line-of-sight Loss (LoS) existence, and obtain the probability-weighted path loss from the LoS existence probability; Step S3: Based on the three-dimensional air-to-air communication scenario established in Step S1 and the urban building stochastic environment model established in Step S2, establish a dynamic adaptive dual-cluster non-line-of-sight NLoS framework; wherein, the process of establishing the dynamic adaptive dual-cluster non-line-of-sight NLoS framework includes: determining the generation ratio of roof scattering clusters and general scattering clusters according to the relationship between Tx and Rx heights and building height distribution, and the roof scattering clusters and general scattering clusters forming NLoS paths. Step S4: Based on the three-dimensional air-to-air communication scenario established in Step S1 and the urban building stochastic environment model established in Step S2, establish a ground or rooftop single-mirror reflection SR path model; calculate the SR path existence probability, mirror reflection point height, SR path propagation distance, propagation delay, departure angle, and arrival angle based on the ground or rooftop single-mirror reflection SR path model. Step S5: Using the LoS existence probability obtained in step S2, the SR path existence probability obtained in step S4, the cluster birth and death process, and the UAV attitude rotation parameters in the three-dimensional air-to-air communication scenario established in step S1, a three-dimensional MIMO channel is generated; wherein, the three-dimensional MIMO channel includes the LoS path, the SR path, and the NLoS path formed in step S3. A three-dimensional MIMO channel impulse response is generated based on the LoS component formed by the LoS path, the SR component formed by the SR path, and the NLoS component formed by the NLoS path. Step S6: Calculate the statistical characteristics of the UAV air-to-air wireless channel based on the three-dimensional MIMO channel impulse response generated in step S5.

2. The method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles according to claim 1, characterized in that, The statistical characteristics of UAV air-to-air wireless channels include path loss, time autocorrelation function, root mean square delay spread, and root mean square angle spread; the urban building stochastic environment model is the International Telecommunication Union (ITU) urban building stochastic environment model.

3. The method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles according to claim 1, characterized in that, In step S1, the three-dimensional air-to-air communication scenario is represented by the Tx local coordinate system, the Rx local coordinate system, and the global coordinate system. The multipath propagation geometry is determined based on the position, height, velocity, direction of motion, array structure, antenna polarization, and attitude rotation parameters of Tx and Rx. Tx configuration... One transmit antenna element, Rx configuration There are p-th receiving antenna elements, and the p-th transmitting antenna element is denoted as . The q-th receiving antenna element is denoted as ,in , ; The three-dimensional air-to-air communication scenario includes the LoS path from Tx to Rx, the SR path generated by the ground or roof, and the NLoS path generated by general scattering clusters and roof scattering clusters. In the NLoS path, the nth scattering cluster is either a general scattering cluster or a roof scattering cluster, and includes the first hop scattering point. and last jump scattering point The m-th ray within the n-th scattering cluster has azimuth departure angle, elevation departure angle, azimuth arrival angle, elevation arrival angle, propagation delay, path power, and polarization phase; the total number of general scattering clusters and rooftop scattering clusters. and the number of rays within the nth scattering cluster Changes over time.

4. The method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles according to claim 1, characterized in that, In step S2, the stochastic environment model of urban buildings, building height It follows a Rayleigh distribution, and its probability density function is: ,in, , Indicates building height The probability density function, The scale parameter represents the building height distribution; the building location follows a Poisson point process; the probability of building height below the Tx-Rx link line is calculated and combined with the expected number of buildings on the link. Obtain the probability of Loss ; Probability-weighted path loss ,in, This represents the path loss in the Loss of Path (LoS) state. This represents the path loss under NLoS conditions.

5. A method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles according to claim 1, characterized in that, In step S3 of the dynamic adaptive dual-cluster non-line-of-sight (NLoS) framework, let Given the minimum values ​​of Tx height and Rx height at time t, the theoretical proportion of roof scattering clusters in the generation ratio of roof scattering clusters and general scattering clusters is: ;when When the height is lower than the average height of urban buildings, The corresponding number of general scattering clusters is greater than the number of rooftop scattering clusters; when the minimum values ​​of Tx height and Rx height at time t are not less than the average height of urban buildings, The number of corresponding roof scattering clusters is not less than the number of general scattering clusters, and the roof scattering clusters participate in NLoS channel generation; The height of the roof scattering cluster is generated based on the building height distribution. The azimuth, pitch, path delay, and path power of the roof scattering cluster are generated according to the geometric random channel model. General scattering clusters are used to characterize scattering from building facades, ground, and obstacles, while roof scattering clusters are used to characterize scattering from the roof of high-rise buildings.

6. A method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles according to claim 1, characterized in that, In step S4, the probability of SR path existence in a single specular reflection model of the ground or roof is... The probability of roof illumination or the probability that the reflection point is below the minimum height of the transmitting UAV Tx and the receiving UAV Rx is determined by the expected number of buildings on the link; when an SR path exists, the height of the specular reflection point, the SR path propagation distance, the propagation delay, the departure angle, and the arrival angle are calculated, wherein the height of the specular reflection point... Generated according to the mixture probability density function, where Corresponding to ground mirror reflection, Corresponding to the reflection of the roof mirror.

7. A method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles according to claim 1, characterized in that, In step S5, the impulse response of the three-dimensional MIMO channel is determined by the three-dimensional MIMO channel matrix. Represented as ,in For the shadow to fade, This is a small-scale fading matrix; The channel impulse response from the p-th transmit element to the q-th receive element in the three-dimensional MIMO channel impulse response. From the Los component SR components and NLoS components The weighted summation is obtained; ,in Rice factor, For SR path power coefficient, Let be the total power coefficient of the NLoS path, and ; The result is obtained by multiplying the transmit and receive antenna patterns and the polarization matrix, where Let m be the time-varying power of the m-th ray within the n-th cluster. To correspond to the time-varying propagation delay, For carrier frequency, The polarization phase is given by j, which is the imaginary unit. It is the Dirac function; Where c is the speed of light; and These represent the mapping distance of the p-th transmitting element caused by the attitude rotation of the launching UAV Tx and the mapping distance of the q-th receiving element caused by the attitude rotation of the receiving UAV Rx, respectively. for to distance, for to The distance; for and Virtual link latency between; , , and These represent the p-th transmitting antenna element, the first hop scattering point, the last hop scattering point, and the q-th receiving antenna element, respectively.

8. A method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles according to claim 1, characterized in that, The cluster generation and extinction process in step S5 includes: based on the sampling time interval Calculate the survival probability of existing scattering clusters And based on the cluster generation rate Cluster recombination rate Calculate the expected number of new scattering clusters .

9. A method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles according to claim 2, characterized in that, In step S6, the time autocorrelation function is the channel transfer function from the p-th transmit antenna element to the q-th receive antenna element. With the The transmitting antenna array element to the first Each receiving antenna element in time, Channel transfer function at frequency The correlation between them was obtained; Correlation characterizes the channel response of two transmit / receive antenna array elements over a time interval. Frequency interval and spatial interval The degree of similarity, among which, and Let these represent the indexes of the other transmitting antenna element and the other receiving antenna element, respectively; let the frequency interval be... and spatial interval When the value is zero, the time autocorrelation function TACF is obtained.

10. A method for modeling an air-to-air three-dimensional MIMO wireless channel for unmanned aerial vehicles according to claim 2, characterized in that, In step S6, the root mean square angle is expanded. Root mean square delay spread ,in and These are the power-weighted average angle and the power-weighted average time delay, respectively. For the i-th path from the p-th transmit antenna element to the q-th receive antenna element, at angle... The relative power contribution at that location Let be the angle of the i-th path; For the path of this array element pair in time delay The relative power contribution at that location This refers to path delay.

Images