6G PERVASIVE CHANNEL MODELING METHOD SUITABLE FOR ALL FREQUENCY BANDS AND ALL SCENARIOS

Abstract

A 6G pervasive channel modeling method includes the following steps: S1, setting a propagation scenario and a propagation condition, and determining a carrier frequency, an antenna type, a layout of a transmitting end and a receiving end, and the like; S2, generating large-scale fadings such as path loss, shadowing and blocking effect loss; S3, generating large-scale parameters having spatial consistency; S4, generating scatterer positions in ellipsoid Gaussian scattering distribution, and calculating a delay, an angle and a power of a cluster according to the positions of the transmitting end, the receiving end and the scatterers to generate a channel coefficient, and S5, on the basis of movements of the transmitting end and the receiving end and a birth-death process of each cluster, updating the large scale parameters and small-scale parameters, and generating a new channel coefficient.

Description

TECHNICAL FIELD

The present disclosure relates to a method for pervasively modeling 6G channels for all frequency bands and all scenarios, which belongs to the technical field of wireless communication.

BACKGROUND

As illustrated in FIG. 2 below, 6G wireless channels can be summarized as all spectra (sub-6 GHz/millimeter wave (mmWave)/terahertz (THz)/optical wireless frequency bands), global-coverage scenarios (space-air-ground-sea integration, including the communication channels of satellite, unmanned aerial vehicle (UAV), terrestrial and maritime) and full-application scenarios (such as vehicle-to-vehicle (V2V), high-speed train (HST) channels, (ultra-)massive multiple-input multiple-output (MIMO), reconfigurable intelligent surfaces (RIS), industrial Internet of things (IIoT)) channels. At the same time, the 6G channels suitable for all frequency bands and all scenarios also exhibit multitudinous new channel characteristics, which brings new challenges to the 6G channel modeling.

In terms of the all spectra, due to the application of high frequency bands such as mmWave and THz, wireless channels exhibit the characteristics such as wide bandwidth, frequency non-stationarity, diffuse scattering, large path loss, blockage effects and atmosphere absorption. In the visible light communication (VLC) band, the channels will no longer have small-scale fading, and exhibit negligible Doppler effect and frequency non-stationarity. In terms of global-coverage scenarios, in addition to the terrestrial mobile communication scenario, satellite communication, UAV communication and maritime communication scenarios are also included. In the satellite communication channels, the Doppler shift caused by the rapid movement of satellites, the rain attenuation and the ionospheric effect should be taken into account. In the UAV communication system, the arbitrary three-dimensional (3D) trajectories of UAV and altitudes-dependent large-scale parameters should be mainly considered. In terms of the full-application scenarios, the V2V channels exhibit Doppler shift and time-domain non-stationary characteristics due to the multiple mobility of the transceiver and the clusters. At higher moving speeds of more than 500 km/h, the channel experiences stronger Doppler shifts and more pronounced time-domain non-stationarity. In the scenario of ultra-high speed train (UHST) running in vacuum tube, the influence of vacuum tube waveguide effect should also be considered. The ultra-massive MIMO channel exhibits spherical wavefront and spatial non-stationary characteristics. Ultra-dense scatterer distribution and multiple mobility need to be considered in IIoT channels. In addition, the precise modeling for wireless channels utilizing RIS technology also needs to be studied.

Considering that the mixed application of various new technologies will bring about the combination of different channel characteristics, an important challenge of modeling 6G channels is how the various channel characteristics be considered comprehensively and to propose a pervasive channel model suitable for all frequency bands and all scenarios. For example, when mm Wave/THz band and massive MIMO technology are applied at the same time in high-speed moving scenarios, wireless channels will exhibit spatial-time-frequency non-stationary characteristics, spatial consistency (that is, in multi-user scenarios, channel coefficients of neighborhood users are correlated or different trajectory points of a single user are spatially correlated), and multi-band correlation.

To sum up, it is urgent to establish an accurate, pervasive, and flexible 6G channel model. The problems are tried to be solved in standard 5G channel models such as B5GCM, 3GPP TR 38.901, IMT-2020 and QuaDRiGa, but all of them cannot accurately and comprehensively describe all of the above-mentioned characteristics. In terms of the all spectra, these models do not apply to the VLC band, and ignore some characteristics of the mmWave/THz band. For example, QuaDRiGa neglects to model the atmospheric absorption and blockage effects, and 3GPP TR 38.901 and IMT-2020 neglects the frequency non-stationary characteristics of high frequency bands. In terms of all coverage, these channel models are aimed at land mobile communication channels, and cannot be applied to the scenarios of satellite, UAV and maritime communication. In terms of full application, they do not support modeling for HST, UHST, RIS, and IIoT channels, and the spherical wavefront and spatial non-stationary properties of (ultra-)massive MIMO are not taken into consideration in 3GPP TR 38.901 and IMT-2020. In summary, these models still lack pervasiveness and do not take all the channel characteristics mentioned above into consideration. In order to fill the research gap, the pervasive channel modeling theory is proposed and applied to the geometric random channel model, and a modeling method for 6G pervasive channels suitable for all frequency bands and all scenarios is proposed and disclosed.

SUMMARY

Technical problems: the objectives of the present disclosure are to provide a method for pervasively modeling 6G channels for all spectra (sub-6 GHz/mmWave/THz/optical wireless frequency bands), global-coverage scenarios (space-air-ground-sea integration, including the communication channels of satellites, UAV, terrestrial and maritime) and full-application scenarios (such as V2V, HST, UHST, (ultra-)massive MIMO, RIS, IIoT scenarios).

Technical solutions: the present disclosure provides a method for pervasively modeling 6G channels for all frequency bands and all scenarios. Massive uniform linear arrays are adopted at both a transmitter side and a receiver side in a 6G pervasive geometry-based stochastic channel model named 6G pervasive channel model (6GPCM), and the model is a multi-bounce propagation model, where A_p^T denotes a p-th array element of a transmitter antenna array, A_q^R denotes a q-th array element of a receiver antenna array, and a distance between the transmitter antenna array and the receiver antenna array is δ_T(δ_R); β_A^T(R) denotes an azimuth angle of the transmitter antenna array and the receiver antenna array in an xy plane, and β_E^T(R) denotes an elevation angle of the transmitter antenna array and the receiver antenna array; for a n-th propagation path from A_p^T to A_q^R, n=1, 2, 3, . . . , N_qp(t), where C_n^A denotes a first-bounce cluster of the n-th path proximity to the transmitter side, C_n^Z denotes a last-bounce cluster proximity to the receiver side, and a propagation path between the two clusters is modeled as a virtual link; when a delay of the virtual link between the first-bounce cluster and the last-bounce cluster is zero, the model is reduced to a single-bounce model; besides, N_qp(t) is the number of paths from A_p^T to A_q^R at a time instant/corresponding to N_qp(t) cluster pairs in a double-cluster model with the first-bounce cluster and the last-bounce cluster in one-to-one correspondence with each other, and corresponding to N_qp(t) clusters in a single-cluster model; on a microscopic level, analyzing clusters C_n^A and C_n^Z on the n-th path, and M_n(t) scatterers are existed in the clusters, C_mn^A denotes an m-th scatterer in C_n^A, C_mn^Z denotes an m-th scatterer in C_n^Z; from a view point of the path, C_mn^A, is understood as a scatterer connected by an m-th sub-path from A_p^T to C_n^A, and C_mn^Z is understood as a scatterer connected by an m-th sub-path from A_q^R to C_n^Z; besides, ϕ_A,mn^T(t) and ϕ_E,mn^T(t) denote an azimuth departure angle and an elevation departure angle corresponding to an m-th sub-path from A₁^T to C_n^A at the time instant t, ϕ_A,mn^T(t) and ϕ_E,mn^R(t) are an azimuth arrival angle and an elevation arrival angle corresponding to an m sub-path from A₁^R to C_n^Z at the time instant t; besides, motion conditions at the transmitter side, the receiver side and a motion conditions of the clusters are modeled by the model respectively, and three-dimensional motions of an arbitrary speed and an arbitrary trajectory of a transceiver and the clusters are supported, where ν^T(t), ν^R(t), ν^An(t), ν^Zn(t) denote motion speeds at the transmitter side, the receiver side, the first-bounce and cluster the last-bounce cluster respectively, α_A^T(t), α_A^R(t), α_A^An(t), α_A^Zn(t) denote azimuth angles of the motor directions at the transmitter side, the receiver side, the first-bounce cluster and the last-bounce cluster respectively, α_E^T(t), α_E^R(t), α_E^An(t), α_E^Zn(t) denote elevation angles of the motor direction at the transmitter side, the receiver side, the first-bounce cluster and the last-bounce cluster, respectively.

A channel matrix of the 6GPCM is represented as:

$H={\left[\mathrm{PL}·\mathrm{SH}·\mathrm{BL}·\mathrm{WE}·\mathrm{AL}\right]}^{1/2}·H_s,$

where PL, SH, BL, WE, AL denote large-scale fadings, PL denotes a path loss, SH denotes a shadowing, BL denotes a blockage loss, AL denotes an atmospheric gas absorption loss, WE denotes a weather effect loss, H_s denotes a small-scale fading channel matrix.

The small-scale fading channel matrix H_s is represented as follows:

$H_s={\left[h_{qp,f_c}\left(t,\tau \right)\right]}_{M_R\times M_T},$

where M_T denotes the number of antenna elements in the transmitter antenna array, M_R denotes the number of antenna elements in the receiver antenna array, h_qp,fc(t, τ) denotes a channel impulse response between the array element A_p^T in the transmitter antenna array and the array element A_q^R in the receiver antenna array at the time instant t, which is represented as a superposition of an LoS component h_qp,fc^LoS(t, τ) and a NLoS component h_qp,fc^NLoS(t, τ):

$h_{qp,f_c}\left(t,\tau \right)=\sqrt{\frac{K_R\left(t\right)}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{LoS}\left(t,\tau \right)+\sqrt{\frac{1}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{NLoS}\left(t,\tau \right),$

where K_R(t) denotes a Rice factor, h_qp,fc^LoS(t, τ) and h_qp,fc^NLoS(t, τ) are respectively represented as follows:

$h_{\mathrm{qp},f_c}^{LoS}\left(t,\tau \right)={\left[\begin{array}{c}{F}_{q,f_c,V}\left({\varphi }_{E,L}^R\left(t\right),{\varphi }_{A,L}^R\left(t\right)\right)\\ F_{q,f_c,H}\left({\varphi }_{E,L}^R\left(t\right),{\varphi }_{A,L}^R\left(t\right)\right)\end{array}\right]}^T\left[\begin{array}{cc}{e}^{j{\theta }_L^{\mathrm{VV}}}& 0\\ 0& -e^{j{\theta }_L^{HH}}\end{array}\right]$ $F_r\left[\begin{array}{c}{F}_{q,f_c,V}\left({\varphi }_{E,L}^T\left(t\right),{\varphi }_{A,L}^T\left(t\right)\right)\\ F_{p,f_c,H}\left({\varphi }_{E,L}^T\left(t\right),{\varphi }_{A,L}^T\left(t\right)\right)\end{array}\right]e^{j2\pi f_c{\tau }_{\mathrm{qp}}^L\left(t\right)}\delta \left(\tau -{\tau }_{qp}^L\left(t\right)\right)$ $h_{\mathrm{qp},f_c}^{\mathrm{NLoS}}\left(t,\tau \right)={\sum }_{n=1}^{N_{qp}\left(t\right)}{{\sum }_{m=1}^{M_n\left(t\right)}\left[\begin{array}{c}{F}_{q,f_c,V}\left({\varphi }_{E,m_n}^R\left(t\right),{\varphi }_{A,m_n}^R\left(t\right)\right)\\ F_{q,f_c,H}\left({\varphi }_{E,m_n}^R\left(t\right),{\varphi }_{A,m_n}^R\left(t\right)\right)\end{array}\right]}^T\left[\begin{array}{cc}{e}^{j{\theta }_{m_n}^{VV}}& \sqrt{\mu {\kappa }_{m_n}^{-1}\left(t\right)}{e}^{j{\theta }_{m_n}^{VH}}\\ \sqrt{{\kappa }_{m_n}^{-1}\left(t\right)}{e}^{j{\theta }_{m_n}^{\mathrm{HV}}}& \sqrt{\mu }{e}^{j{\theta }_{m_n}^{\mathrm{HH}}}\end{array}\right]$ $F_r\left[\begin{array}{c}{F}_{p,f_c,V}\left({\varphi }_{E,m_n}^T\left(t\right),{\varphi }_{A,m_n}^T\left(t\right)\right)\\ F_{p,f_c,H}\left({\varphi }_{E,m_n}^T\left(t\right),{\varphi }_{A,m_n}^T\left(t\right)\right)\end{array}\right]\sqrt{P_{\mathrm{qp},m_n,f_c}\left(t\right)}{e}^{j2\pi f_c{\tau }_{\mathrm{qp},m_n}\left(t\right)}\delta \left(\tau -{\tau }_{qp,m_n}\left(t\right)\right),$

where {*}^T denotes a transposition operation, f_c denotes a carrier frequency, F_p(q),fc,v and F_p(q),fc,H denote antenna patterns of the array element A_p^T (A_q^R) for vertical and horizontal polarizations at different frequency bands, κ_mn(t) denotes a cross polarization power ratio, μ denotes a co-polar imbalance, ϕ_A,L^T(t) and ϕ_E,L^T(t) denote an azimuth departure angle and an elevation departure angle corresponding to the LoS path from A₁^T to A₁^R at the time instant t, ϕ_A,L^R(t) and ϕ_E,L^R(t) denote an azimuth arrival angle and an elevation arrival angle corresponding to the LoS path from A₁^T to A₁^R at the time instant t, θ_L^VV, θ_L^HH, θ_mn^VV, θ_mn^VH, θ_mn^HV and θ_mn^HH are random phases uniformly distributed over (0, 2π],

$F_r=\left(\begin{array}{cc}\cos {\psi }_{l,m}& -s\mathrm{in}{\psi }_{l,m}\\ \sin {\psi }_{l,m}& \cos {\psi }_{l,m}\end{array}\right),{\psi }_{l,m}=108/f_c^2$

denotes a Faraday rotation angle, a unit of f_c in which the Faraday rotation angle is calculated here in GHZ, P_qp,mn,fc(t) denotes a power of m-th sub-path in n-th path from A₁^T to A₁^R at the NLoS condition, τ_qp^L(t) denotes a delay of the LoS path at the time instant t,

${\tau }_{qp}^L\left(t\right)=\frac{{\overrightarrow{d}}_{qp}\left(t\right)}{c},$

{right arrow over (d)}_qp(t) denotes a vector distance between the transmitter antenna array A_p^T and the receiver antenna array A_q^R at the time instant t, c denotes the speed of light, τ_qp,mn(t) denotes a delay of the m-th sub-path in the n-th path from A₁^T to A₁^R at the time instant t, P_qp,mn,fc(t) denotes a power of the m-th sub-path in the n-th path from A₁^T to A₁^R at the time instant t, all of the above parameters are time-varying parameters.

When the method for pervasively modeling 6G channels is utilized in maritime communication scenarios, a LoS path component and multipath components of both a rough ocean surface and an evaporation duct over a sea surface are modeled as h_qp,fc^LoS(t, τ), h_qp,fc^NLoS1(t, τ) and h_qp,fc^NLoS2((t, τ) by the model, and power control factors S₁ and S₁ are used to manipulate a disappearance and an appearance of corresponding parts with variations of distances between two ships, that is, a NLoS part of a formula for calculating h_qp,fc^NLoS(t, τ) is divided into two parts: h_qp,fc^NLoS1(t, τ) and h_qp,fc^NLoS2(t, τ), and S₁+S₂=1; in IIoT scenarios, specular multipath components and dense multipath components are modeled as h_qp,fc^NLoSSC(t, τ) and h_qp,fc^NLoSDMC(t, τ) respectively, and modeling methods for h_qp,fc^NLoS1(t, τ), h_qp,fc^NLoS2(t, τ), h_qp,fc^NLoSSC(t, τ) and h_qp,fc^NLoSDMC(t, τ) are the same as that for h_qp,fc^NLoS(t, τ) merely with different parameter values and different distributions of clusters.

When the method for pervasively modeling 6G channels is utilized in RIS scenarios, channels are divided into a sub-channel H_TI from the transmitter side to the RIS, a sub-channel H_IR from the RIS to the receiver side and a sub-channel H_TR from the transmitter side to the receiver side, the three sub-channels are modeled respectively and a phase shift diagonal matrix Φ is introduced to implement an intelligent control for channel environments, calculation methods of H_IR, H_TI and H_TR are the same as that of H_s merely with different parameter values and different distributions of clusters.

When the method for pervasively modeling 6G channels is utilized for VLC channels, on one hand, wavelengths of optical signals are extremely short, a size of the receiver is commonly multi-million wavelengths, with no rapid signal fading on multi-wavelengths, on another hand, due to an incoherent light emitted by an LED light in a VLC system, the optical signals has no phase information, and no rapid signal fading is caused after a superposition of real-valued multipath signals at the receiver side with an exhibition on a slow-varying shadowing, therefore although a current VLC model representation is a channel impulse response form of a multipath superposition, the representation is essentially a large-scale scale model of modeling PL and SH, that is, H_s=1, PL·SH=h_pVpH^LoS(t, τ)+h_pVpH^NLoS(t, τ)=P_pVpH^LoS(t)·δ(τ−τ_pVpH^LoS(t))+P_pVpH,mn^NLoS(t)·δ(τ−τ_pVpH,mn(t)), p_H, p_Vdenote the number of rows and the number of columns in an LED array.

When the method for pervasively modeling 6G channels is utilized in multi-link scenarios:

Assuming that the number of base stations is N_BS and the number of users is N_MS, a channel transmission matrix of a multi-link channel model is represented as a following formula:

$H_M={\left[\begin{array}{ccc}{H}_{{\mathrm{BS}}_1{\mathrm{MS}}_1}& \dots & H_{{\mathrm{BS}}_1{\mathrm{MS}}_{N_{\mathrm{MS}}}}\\ ⋮& \ddots & ⋮\\ H_{{\mathrm{BS}}_{N_{\mathrm{BS}}}{\mathrm{MS}}_1}& \dots & H_{{\mathrm{BS}}_{N_{\mathrm{BS}}}{M}_{N_{\mathrm{MS}}}}\end{array}\right]}_{N_{BS}\times N_{MS}},$

H_BSiMSj, i=1, 2 . . . N_BS, j=1, 2 . . . N_MS corresponding to each link is a single-link channel model H described above.

In the method for pervasively modeling 6G channels, detailed steps for generating the channel matrix H are specifically as follows.

In S1, propagation scenarios and conditions are set; a carrier frequency, an antenna type, a layout of the channel and a motion trajectory of the transceiver are determined.

In S2, path loss, shadowing, oxygen absorption and blockage effect loss are generated; the method mainly focuses on a modeling for a small-scale fading, and standard channel models are referable to a calculation the large scale fadings.

In S3, according to positions and motion conditions of the transceiver, large-scale parameters with spatial consistency for a delay spread (DS) and 4 angle spreads are generated.

Except SH, other corresponding large-scale parameters include a delay spread DS, an azimuth spread of arrival (ASA), an azimuth spread of departure (ASD), an elevation spread of arrival (ESA), an elevation spread of departure (ESD), a Rice factor (K_R) and a cross-polarization ratio (XPR), a generation of the DS is represented as a following formula:

${\mathrm{DS}}_{f_c}\left(P\right)=D{S}_{\mu ,f_c}+X^{DS}\left(P\right)·{\mathrm{DS}}_{\sigma ,f_c},$ $\mathrm{where}P=\left(P^T,P^R\right)$

is composed of transceiver position vectors, P^T(t)=(x^T(t), y^T(t), z^T(t)) and P^R(t)=(x^Rt), y^R(t), z^R(t)) denote a coordinate vector at the transmitter side and a coordinate vector at the receiver side at the time instant t, respectively, and initial values of which are generated according to simulation environments and requirements; X^DS(P) denotes a normal distribution variable generated by a sine wave superposition method and following a spatial consistency with a mean value of 0 and a variance of 1, DS_μ,fc denotes a mean value for DS in a frequency band f_c, and DS_σ,fc denotes a variance of DS in the frequency f_c, configuration values for DS_σ,fc are divided into three types according to a height h_UT of a user terminal; for terrestrial mobile communication scenarios 1.5 m≤h_UT≤22.5 m, values set from Table 7.5-6 of 3GPP TR 38.901 are referable; for UAV scenarios 22.5 m≤h_UT≤300 m, values set from Table B1.2 of 3GPP TR 36.777 standardization document are referable; for satellite communication scenarios, values set from Table 6.7-2 of 3GPP TR 38.811 standardization document are referable; in NLoS conditions of urban macro Uma scenarios, when a carrier frequency ranges from 2 to 4 GHZ, DS_μ,fc is calculated as follows:

${\log }_{10}\left(D{S}_{\mu ,f_c}/1s\right)=\{\begin{array}{cc}-0.204{\log }_{10}\left(f_c\right)-6.28,& 1.5m<h_{UT}\le 22.5m,\mathrm{NLoS}\\ 0.0965{\log }_{10}\left(h_{UT}\right)-7.503,& 22.5m<h_{UT}\le 300m,\mathrm{NLoS}\\ -7.21& \left(\mathrm{An}\mathrm{elevation}\mathrm{angle}\mathrm{of}\mathrm{the}\mathrm{link}\mathrm{is}10°\right)\end{array}.$

Generation processes of other large-scale parameters are the same as a generation process of the DS, values for all large-scale parameters with spatial consistency in a logarithm domain can be obtained by multiplying a cross-correlation matrix among the large-scale parameters, after 8 large-scale parameters are generated, subsequently values in the logarithm domain are required to be converted into a linear domain; so that the large-scale parameters of the channel are obtained.

In S4, scatterers following an ellipsoid Gaussian scattering distribution are generated, delays, angles and powers of the clusters are calculated according to geographical location information of the transceiver and the scatterers, and channel coefficients are generated.

In S5, the large-scale parameters and the small-scale parameters are updated according to movements of the transceiver and birth-death processes of the clusters; and new channel coefficients are generated. A space-time-frequency non-stationarity of the model is mainly reflected in two aspects, one is parameters for space-time-frequency variations, and another is birth-death processes of the clusters in a space-time-frequency domain, the number of clusters at the time instant t is calculated as follows:

$N_{qp}\left(t\right)=N_{surv}\left(t\right)+N_{new}\left(t\right),$

where N_qp(t) denotes the number of the clusters, N_surv(t) denotes the number of survived clusters determined by a survived probability P_surv(Δt, Δr, Δf) of the clusters, N_new(t) denotes the number of newly generated clusters following a Poisson distribution with a mean value E[N_new(t)], λ_G is defined as a birth rate of the clusters, λ_R is defined as a combination rate of the clusters, that is, a death rate.

S4 is specifically as follows.

In S401, positions of the scatterers are obtained by using an ellipsoid Gaussian scattering distribution, the scatterers in n-th cluster centered on (d_n^X, ϕ_E,n^X, ϕ_A,n^X) follow a Gaussian distribution with standard deviations of σ_x^X, σ_y^X and σ_z^X on three axes, respectively; after obtaining the positions of the scatterers, the positions of the scatterers are converted into spherical coordinates; positions {right arrow over (C)}_mn^A(t₀) and {right arrow over (C)}_mn^Z(t₀) of the scatterers in n-th cluster corresponding to a position of a first transmitter antenna {right arrow over (A)}₁^T(t₀) and a position of a first receiver antenna {right arrow over (A)}₁^R(t₀) are represented as {right arrow over (C)}_mn^A(t₀)=(d_mn^T(t₀), ϕ_A,mn^T(t₀), ϕ_E,mn^T(t₀)) and {right arrow over (C)}_mn^Z(t₀)=(d_mn^R(t₀), ϕ_A,mn^R(t₀), ϕ_E,mn^R(t₀)), where d_mn^X(t₀), ϕ_A,mn^X(t₀) and ϕ_E,mn^X(t₀) denote a distance, an azimuth angle, and an elevation angle of m-th sub-path of n-th cluster at the transmitter side or the receiver side, respectively, X∈{T, R} denotes the transmitter side and the receiver side.

In S402, in a multi-bounce channel model, delays of sub-paths in the cluster at an initial time instant are calculated by

${\tau }_{\mathrm{qp},m_n}\left(t_0\right)=\frac{\left(d_{p,m_n}^T\left(t_0\right)+d_{q,m_n}^R\left(t_0\right)\right)}{c}+{\tilde{\tau }}_{m_n}\left(t_0\right),$

where {tilde over (τ)}_mn denotes a delay of virtual links between {right arrow over (C)}_mn^A and {right arrow over (C)}_mn^Z, d_p,mn^T(t₀) denotes a distance between A_p^T and C_mn^A at the time instant t₀, and d_q,mn^R(t₀) denotes a distance between A_q^R and C_mn^Z at the time instant t₀,

${\tilde{\tau }}_{m_n}\left(t_0\right)=\frac{{\tilde{d}}_{m_n}\left(t_0\right)}{c}+{\tau }_{\mathrm{link}}\left(t_0\right),$

{tilde over (d)}_mn(t₀) denotes a distance between the first-bounce cluster and the last-bounce cluster, τ_link denotes a non-negative variable following an exponential distribution.

In S403, in (ultra-)massive MIMO scenarios, a sub-paths power P_qp,mn,fc(t) in the clusters is varied along a time axis and an array axis, and the sub-paths power is commonly modeled as a lognormal process varying with time and a lognormal process varying with the array, a non-normalized sub-paths power P′_qp,mn,fc(t) in the clusters is:

$P_{\mathrm{qp},m_n,f_c}^{\prime }\left(t\right)=\underset{A\mathrm{time}\mathrm{domain}}{\underbrace{\exp \left(-{\tau }_{\mathrm{qp},m_n}\left(t\right)\frac{r_{\tau }-1}{r_{\tau }\mathrm{DS}}\right){10}^{-\frac{Z_n}{10}}}}·\underset{A\mathrm{space}\mathrm{domain}}{\underbrace{{\xi }_n\left(p,q\right)}},$

where Z_n denotes a per cluster shadowing term in dB, r_τ denotes a delay distribution proportionality factor, ξ_n(p, q) denotes a two-dimensional spatial lognormal process for simulating smooth power variations over antenna arrays.

In wide bandwidth scenarios, a power value is multiplied by

${\left(\frac{f}{f_c}\right)}^{{\gamma }_{m_n}}$

in a frequency domain by taking frequency domain non-stationary characteristics into account, where γ_mn is a frequency-dependent constant factor, eventually, an ultimate power P_qp,mn,fc(t) of the sub-paths in the clusters is obtained by normalizing the powers of all clusters; if the clusters are newly generated, τ_qp,mn(t) is substituted with τ_qp,mn(t₀) to obtain an initial power of the m-th sub-path in the n-th cluster between A_p^T and A_q^R.

In S404, for the survived clusters, small-scale parameters such as the powers and the delays of the sub-paths in the clusters at different time instants are required to be updated, for a trajectory segment at the time instant t₁, that is, at a subsequent time instant after the clusters are generated, a coordinate of the p-th transmitter antenna A_p^T is:

${\overrightarrow{A}}_p^T\left(t_1\right)={\overrightarrow{A}}_p^T\left(t_0\right)+v^T\left(t_1-t_0\right)·{\left[\begin{array}{c}\cos {\alpha }_A^T·\cos {\alpha }_E^T\\ \sin {\alpha }_A^T·\cos {\alpha }_E^T\\ \sin {\alpha }_E^T\end{array}\right]}^T,$

where a coordinate {right arrow over (A)}_p^T(t₀) of the p-th transmitter antenna at the initial time instant is calculated by

${\overrightarrow{A}}_p^T\left(t_0\right)={\overrightarrow{A}}_1^T\left(t_0\right)+\left(p-1\right)·{\delta }_T·{\left[\begin{array}{c}\cos {\beta }_A^T·\cos {\beta }_E^T\\ \sin {\beta }_A^T·\cos {\beta }_E^T\\ \sin {\beta }_E^T\end{array}\right]}^T,$

a coordinate {right arrow over (C)}_mn^A(t₁) of an m-th scatterer in a n-th first-bounce cluster is calculated by

${\overrightarrow{C}}_{m_n}^A\left(t_1\right)={\overrightarrow{C}}_{m_n}^A\left(t_0\right)+v^{A_n}\left(t_1-t_0\right)·{\left[\begin{array}{c}\cos {\alpha }_A^{A_n}·\cos {\alpha }_E^{A_n}\\ \sin {\alpha }_A^{A_n}·\cos {\alpha }_E^{A_n}\\ \sin {\alpha }_E^{A_n}\end{array}\right]}^T$

at the time instant t₁. A distance from A_p^T to C_mn^A, is obtained by calculating d_p,mn^T(t₁)=∥{right arrow over (C)}_mn^A(t₁)−{right arrow over (A)}_p^T(t₁)∥ similarly, a distance d_q,mn^R(t₁) from A_q^R to C_mn^Z is obtained; a delay of the sub-path in the clusters at the time instant t₁ is τ_qp,mn(t₁)=(d_p,mn^T(t₁)+d_q,mn^R(t₁))/c+{tilde over (τ)}_mn; τ_qp,mn(t) and P_qp,mn(t) are obtained by using geographical locations of the transmitter, the receiver, and the scatterer at a previous time instant, (t=t₂, t₃, . . . ).

In S5, in order to model a space-time-frequency evolution process of the clusters more accurately, two types of sampling intervals are introduced, one type is a time domain sampling interval Δt, a frequency domain sampling interval Δf and a space domain (array domain) sampling interval Δr, and channel parameters are updated continuously, another type is described by Δt_BD, Δf_BD and Δr_BD that are integer multiples of corresponding Δt, Δf and Δr, and during the birth-death processes and the evolution processes of the clusters occurred at sampling points, survival probabilities of the transmitter side and receiver side clusters along the array axis and time axis are as follows:

$P_{\mathrm{surv}}^T\left(\Delta t_{\mathrm{BD}},{\delta }_p\right)=e^{-{{\lambda }_R\left({\left({ϵ}_1^T\right)}^2+{\left({ϵ}_2^T\right)}^2+2{ϵ}_1^T{ϵ}_2^T\cos \left({\alpha }_A^T-{\beta }_A^T\right)\right)}^{1/2}}$ $P_{\mathrm{surv}}^R\left(\Delta t_{\mathrm{BD}},{\delta }_q\right)=e^{-{{\lambda }_R\left({\left({ϵ}_1^R\right)}^2+{\left({ϵ}_2^R\right)}^2+2{ϵ}_1^R{ϵ}_2^R\cos \left({\alpha }_A^R-{\beta }_A^R\right)\right)}^{1/2}}$ $\mathrm{where}{ϵ}_1^T=\frac{{\delta }_p\cos {\beta }_E^T}{D_c^A}\left({ϵ}_1^R=\frac{{\delta }_q\cos {\beta }_E^R}{D_c^A}\right)\mathrm{and}{ϵ}_2^T=\frac{v^T\Delta t_{\mathrm{BD}}}{D_c^s}\left({ϵ}_2^R=\frac{v^R\Delta t_{\mathrm{BD}}}{D_c^s}\right)$

denote position differences of a transmitter antenna element and a receiver antenna element on the array axis and the time axis, respectively, D_c^A and D_c^S denote scenario-dependent factors on the array axis and the time axis, respectively, a joint survived probability of the transmitter side and receiver side clusters is represented as follows:

P _surv(Δt_BD, δ_p, δ_q)=P_surv^T(Δt_BD, δ_p)P_surv^R(Δt_BD, δ_q).

The average number of the newly generated clusters is:

$E\left(N_{\mathrm{new}}\right)=\frac{{\lambda }_G}{{\lambda }_R}\left(1-P_{\mathrm{surv}}\left(\Delta t_{\mathrm{BD}},\Delta r_{\mathrm{BD}}\right)\right).$

When wide bandwidth scenarios are studied, the birth-death processes of the clusters also exist on a frequency axis, and a survival probability of the clusters on the frequency axis is:

$P_{\mathrm{surv}}\left(\Delta f_{\mathrm{BD}}\right)=e^{-{\lambda }_R\frac{F\left(\Delta f_{\mathrm{BD}}\right)}{D_c^f}},$

where F(Δf_BD) and D_c^f are determined by channel measurements, D_c^f denotes a scenario-dependent factor on the frequency axis, in summary, when the birth-death processes of the space-time-frequency domain clusters are taken into account, the survival probability of the clusters is:

P _surv(Δt_BD, Δr_BD, Δf_BD)=P_surv^T(Δt_BD, δ_p)P_surv^R(Δt_BD, δ_q)P_surv(Δf_BD).

The average number of the newly generated clusters is:

$E\left(N_{\mathrm{new}}\right)=\frac{{\lambda }_G}{{\lambda }_R}\left(1-P_{\mathrm{surv}}\left(\Delta t_{\mathrm{BD}},\Delta r_{\mathrm{BD}},\Delta f_{\mathrm{BD}}\right)\right).$

In UHST scenarios, by taking account of a waveguide effect and an impact of tube wall roughness on channels in vacuum tube UHST scenarios, the average number of the newly generated clusters is:

$E\left(N_{\mathrm{new}}\right)=\frac{{\lambda }_G}{{\lambda }_R}\left(1-P_{\mathrm{surv}}\left(\Delta t_{\mathrm{BD}},\Delta r_{\mathrm{BD}},\Delta f_{\mathrm{BD}}\right)\right)\left(1-\frac{D_{\mathrm{qp}}\left(t\right)}{D}\right)·\frac{{\rho }_s}{{\rho }_{s_0}}$ ${\rho }_s=e^{\left(-8{\left(\frac{{\mathrm{\pi \sigma }}_h\cos \left(E\left[{\varphi }_{E,m_n}^T\right]\right)}{\lambda }\right)}^2\right)},$

where D_qp(t) denotes a linear distance between the transmitter side and the receiver side at the time instant t, D denotes an initial distance between the transmitter side and the receiver side, ρ_s denotes a scattering coefficient of the tube wall, and ρ_s0 denotes a scattering coefficient with a roughness of σ_h=0.

Technical effects: proposed in the present disclosure is a pervasive channel modeling theory, and the theory is applied to the geometry-based stochastic channel model (GBSM). By using the cluster-based geometry-based stochastic channel modeling method and framework, and using the unified channel impulse response expression, the 6G channels characteristics for all frequency bands and all scenarios can be modeled, and a 6GPCM based on the pervasive channel modeling theory is proposed, which is basically suitable for all spectra such as sub-6 GHz, mm Wave, THz and VLC channels, full-coverage scenarios channels such as satellites, UAV and maritime communications, as well as full-application scenarios channels such as ultra-massive MIMO, IIoT, and RIS. Moreover, the 6GPCM can be simplified into a dedicated channel model of specific frequency bands and specific scenarios by adjusting the parameters. 6GPCM is extremely important for 6G channel model standardization, 6G generic theory technology research and system fusion construction.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 illustrates a flow diagram of the embodiments in the present disclosure.

FIG. 2 illustrates a schematic diagram of 6G wireless channels in the embodiments of the present disclosure.

FIG. 3 illustrates a schematic diagram of a pervasive channel modeling theory in the present disclosure.

FIG. 4 illustrates a schematic diagram of a 6GPCM in the present disclosure.

DETAILED DESCRIPTION OF THE EMBODIMENTS

In order to realize the above objectives, the present disclosure proposes a pervasive channel modeling theory, and proposes a 6GPCM based on the theory. Therefore, the present disclosure mainly includes two parts: the pervasive channel model modeling theory and the 6GPCM construction.

1. The Pervasive Channel Model Modeling Theory

The pervasive channel modeling theory is utilizing a unified channel modeling method and framework, a unified channel impulse response expression, and a comprehensive consideration of the characteristics of 6G channels for all frequency bands and all scenarios, to construct a 6G pervasive channel model that is generally applicable to all frequency bands and scenarios of 6G and that can accurately reflect the channel characteristics of 6G, as illustrated in FIG. 3. At the same time, the 6G pervasive channel model can be simplified into a dedicated channel model of specific frequency bands and specific scenarios by adjusting the parameters for the channel model. Through the analysis on the 6G pervasive channel model, the complex mapping relationship between channel model parameters, channel characteristics and communication system performance can be studied. As a unified channel model framework, 6GPCM is extremely important for 6G channel model standardization, 6G generic theory technology research and system fusion construction.

2. The 6GPCM

The 6GPCM is as illustrated in FIG. 4. The antenna types in the model can be antenna array types such as uniform linear array and uniform planner array, and an arbitrary antenna polarization type is supported. Uniform linear arrays are adopted at both transmitter side and receiver side as illustrated in the schematic diagram and the model in the schematic diagram is a multi-bounce propagation model. A_p^T denotes the p-th array element of a transmitter antenna array, A_q^R denotes the q-th array element of a receiver antenna array, and the distance between the transmitter antenna array and the receiver antenna array is δ_T, (δ_R); β_A^T(R) denotes an azimuth angle of the transmitter (the receiver) antenna array in an xy plane, and β_E^T(R) denotes an elevation angle of the transmitter (the receiver) antenna array; for better understanding, the n-th (n=1, 2, 3, I, N_qp(t)) propagation path from A_p^T to A_q^R is merely described herein, where C_n^A denotes a first-bounce cluster of the n-th path proximity to the transmitter side, C_n^Z denotes a last-bounce cluster proximity to the receiver side, and a propagation path between the two clusters is modeled as a virtual link. When a delay of the virtual link between the first-bounce cluster and the last-bounce cluster is zero, the model is reduced to a single-bounce model; besides, N_qp(t) is the number of paths from A_p^T to A_q^R at a time instant t corresponding to N_qp(t) cluster pairs in a double-cluster model with the first-bounce cluster and the last-bounce cluster in one-to-one correspondence with each other, and corresponding to N_qp(t) clusters in a single-cluster model; on the microscopic level, analyzing clusters C_n^A and C_n^Z on the n-th path, and M_n(t) scatterers are existed in the clusters, C_mn^A denotes the m-th scatterer in C_n^A, C_mn^Z denotes the m-th scatterer in C_n^Z; from the view point of the path, C_mn^A is understood as a scatterer connected by the m-th sub-path from A_p^T to C_n^A, and C_mn^Z is understood as a scatterer connected by the m-th sub-path from A_q^R to C_n^Z; besides, ϕ_A,m,n^T(t) and ϕ_E,mn^T(t) denote an azimuth departure angle and an elevation departure angle corresponding to the m-th sub-path from A₁^T to C_n^A at the time instant t, ϕ_A,mn^T(t) and ϕ_E,mn^R(t) are an azimuth arrival angle and an elevation arrival angle corresponding to the m-th sub-path from A₁^R to C_n^Z at the time instant t; besides, motion conditions at the transmitter side, the receiver side and a motion conditions of the clusters are modeled by the model respectively, and the three-dimensional motions of an arbitrary speed and an arbitrary trajectory of the transceiver and the clusters are supported, where ν^T(t), ν^R(t), ν^An(t), ν^Zn(t) denote motion speeds at the transmitter side, the receiver side, the first-bounce cluster and the last-bounce cluster, respectively, α_A^T(t), α_A^R(t), α_A^An(t), α_A^Zn(t) denote azimuth angles of the motor directions at the transmitter side, the receiver side, the first-bounce cluster and the last-bounce cluster, respectively, α_E^T(t), α_E^R(t), α_E^An(t), α_E^Zn(t) denote elevation angles of the motor direction at the transmitter side, the receiver side, the first-bounce cluster and the last-bounce cluster, respectively.

A channel matrix of a 6GPCM is represented as:

$H={\left[\mathrm{PL}·\mathrm{SH}·\mathrm{BL}·\mathrm{WE}·\mathrm{AL}\right]}^{1/2}·H_s,$

where PL, SH, BL, WE, AL denote large-scale fading, PL denotes path loss, SH denotes shadowing, BL denotes blockage loss, AL denotes atmospheric gas absorption loss, such as the oxygen absorption loss at the mm Wave band and the molecular absorption loss at the THz band, WE denotes weather effect loss, such as rain attenuation loss in satellite communication scenarios. The present disclosure mainly focuses on the calculation of small-scale fading H_s, and the method is as follows:

$H_s={\left[h_{\mathrm{qp},f_c}\left(t,\tau \right)\right]}_{M_R\times M_T},$

where M_T denotes the number of antenna elements in the transmitter antenna array, M_R denotes the number of antenna elements in the receiver antenna array, h_qp,fc(t, τ) denotes a channel impulse response between the array element A_p^T in the transmitter antenna array and the array element A_q^R in the receiver antenna array at the time instant t, which is represented as the superposition of the LoS component h_qp,fc^LoS(t, τ) and the NLoS component h_qp,fc^NLoS(t, τ):

$h_{\mathrm{qp},f_c}\left(t,\tau \right)=\sqrt{\frac{K_R\left(t\right)}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{\mathrm{LoS}}\left(t,\tau \right)+\sqrt{\frac{1}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{\mathrm{NLoS}}\left(t,\tau \right),$

where K_R(t) denotes a Rice factor, h_qp,fc^LoS(t, τ) and h_qp,fc^NLoS(t, τ) are respectively represented as follows:

$h_{\mathrm{qp},f_c}^{\mathrm{LoS}}\left(t,\tau \right)={\left[\begin{array}{c}{F}_{q,f_c,V}\left({\varphi }_{E,L}^R\left(t\right),{\varphi }_{A,L}^R\left(t\right)\right)\\ F_{q,f_c,H}\left({\varphi }_{E,L}^R\left(t\right),{\varphi }_{A,L}^R\left(t\right)\right)\end{array}\right]}^T\left[\begin{array}{cc}{e}^{j{\theta }_L^{\mathrm{VV}}}& 0\\ 0& -e^{j{\theta }_L^{\mathrm{HH}}}\end{array}\right]F_r\left[\begin{array}{c}{F}_{p,f_c,V}\left({\varphi }_{E,L}^T\left(t\right),{\varphi }_{A,L}^T\left(t\right)\right)\\ F_{p,f_c,H}\left({\varphi }_{E,L}^T\left(t\right),{\varphi }_{A,L}^T\left(t\right)\right)\end{array}\right]e^{j2\pi f_c{\tau }_{\mathrm{qp}}^L\left(t\right)}\delta *\tau -{\tau }_{\mathrm{qp}}^L\left(t\right))$ $h_{\mathrm{qp},f_c}^{\mathrm{NLoS}}\left(t,\tau \right)=\sum _{n=1}^{N_{\mathrm{qp}}\left(t\right)}\sum _{m=1}^{M_n\left(t\right)}{\left[\begin{array}{c}{F}_{q,f_c,V}\left({\varphi }_{E,m_n}^R\left(t\right),{\varphi }_{A,m_n}^R\left(t\right)\right)\\ F_{q,f_c,H}\left({\varphi }_{E,m_n}^R\left(t\right),{\varphi }_{A,m_n}^R\left(t\right)\right)\end{array}\right]}^T\left[\begin{array}{cc}{e}^{j{\theta }_{m_n}^{\mathrm{VV}}}& \sqrt{{\mathrm{\mu \kappa }}_{m_n}^{-1}\left(t\right)}{e}^{j{\theta }_{m_n}^{\mathrm{VH}}}\\ \sqrt{{\kappa }_{m_n}^{-1}\left(t\right)}{e}^{j{\theta }_{m_n}^{\mathrm{HV}}}& \sqrt{\mu }{e}^{j{\theta }_{m_n}^{\mathrm{HH}}}\end{array}\right]F_r\left[\begin{array}{c}{F}_{p,f_c,V}\left({\varphi }_{E,m_n}^T\left(t\right),{\varphi }_{A,m_n}^T\left(t\right)\right)\\ F_{p,f_c,H}\left({\varphi }_{E,m_n}^T\left(t\right),{\varphi }_{A,m_n}^T\left(t\right)\right)\end{array}\right]\sqrt{P_{\mathrm{qp},m_n,f_c}\left(t\right)}{e}^{j2\pi f_c{\tau }_{\mathrm{qp},m_n\left(t\right)}\delta (\tau -{\tau }_{\mathrm{qp},m_n\left(t\right)),}}$

where {*}^T denotes a transposition operation, f_c denotes a carrier frequency, F_p(q),fc,V and F_p(q),fc,H denote antenna patterns of the array element A_p^T (A_q^R) for vertical and horizontal polarizations at different frequency bands, κ_mn(t) denotes a cross polarization power ratio, μ denotes a co-polar imbalance, ϕ_A,L^T(t) and ϕ_E,L^T(t) denote an azimuth departure angle and an elevation departure angle corresponding to an LoS path from A₁^T to A₁^R at the time instant t, ϕ_A,L^R(t) and ϕ_E,L^R(t) denote an azimuth arrival angle and an elevation arrival angle corresponding to an LoS path from A₁^T to A₁^R at the time instant t, θ_L^VV, θ_L^HH, θ_mn^VV, θ_mn^VH, θ_mn^HV and θ_mn^HH are random phases uniformly distributed over [0,2π],

$F_4=\left(\begin{array}{cc}\cos {\psi }_{l,m}& -\sin {\psi }_{l,m}\\ \sin {\psi }_{l,m}& \cos {\psi }_{l,m}\end{array}\right),{\psi }_{l,m}=108/f_c^2$

denotes a Faraday rotation angle, the unit of f_c in which the Faraday rotation angle is calculated here in GHz, P_qp,mn,fc(t) denotes a power of the m-th sub-path in the n-th path from A₁^T to A₁^R at the NLoS condition, τ_qp^L(t) denotes a delay of the LoS path at the time instant t,

${\tau }_{\mathrm{qp}}^L\left(t\right)=\frac{{\overrightarrow{d}}_{\mathrm{qp}}\left(t\right)}{c},$

{right arrow over (d)}_qp(t) denotes a vector distance between the transmitter antenna array A_p^T and the receiver antenna array A_q^R at the time instant t, c denotes a speed of light, τ_qp,mn(t) denotes a delay of the m-th sub-path in the n-th path from A₁^T to A₁^R at the time instant t, P_qp,mn,fc(t) denotes a power of the m-th sub-path in the n-th path from A₁^T to A₁^R at the time instant t, all of the above parameters are time-varying parameters.

It is worth noting that in maritime communication channel scenarios, the LoS path component and multipath components of both rough ocean surface and evaporation duct over the sea surface are modeled as h_qp,fc^LoS(t, τ), h_qp,fc^NLoS1(t, τ) and h_qp,fc^NLoS2(t, τ) by the model, and the power control factors S₁ and S₁ are used to manipulate the disappearance and appearance of corresponding parts with variations of distances between two ships, that is, a NLoS part of a formula for calculating h_qp,fc^NLoS(t, τ) is divided into two parts: h_qp,fc^NLoS1(t, τ) and h_qp,fc^NLoS2(t, τ), and S₁+S₂=1; in IIoT scenarios, specular multipath components and dense multipath components are modeled as h_qp,fc^NLoSSC(t, τ) and h_qp,fc^NLoSDMC(t, τ) respectively, and the modeling methods for h_qp,fc^NLoS1(t, τ), h_qp,fc^NLoS2(t, τ), h_qp,fc^NLoSSC(t, τ) and h_qp,fc^NLoSDMC(t, τ) are the same as that for h_qp,fc^NLoS(t, τ) merely with different parameter values and different distributions of clusters. In RIS scenarios, channels are divided into a sub-channel H_TI from the transmitter side to the RIS, a sub-channel H_IR from the RIS to the receiver side and a sub-channel H_TR from the transmitter side to the receiver side, the three sub-channels are modeled respectively and a phase shift diagonal matrix Φ is introduced to implement an intelligent control for channel environments, calculation methods of H_IR, H_TI and H_TR are the same as that of H_s merely with different parameter values and different distributions.

For VLC channels, on one hand, wavelengths of optical signals are extremely short, a size of the receiver is commonly multi-million wavelengths, with no rapid signal fading rapidly on multi-wavelengths, on another hand, due to an incoherent light emitted by an LED light in a VLC system, the optical signals has no phase information, and no rapid signal fading is caused after a superposition of real-valued multipath signals at the receiver side with an exhibition on a slow-varying shadowing, therefore although a current VLC model representation is a channel impulse response form of a multipath superposition, the representation is essentially a large-scale model of modeling PL and SH, that is,

$H_s=1,\mathrm{PL}·\mathrm{SH}=h_{p_V{p}_H}^{\mathrm{LoS}}\left(t,\tau \right)+h_{p_V{p}_H}^{\mathrm{NLoS}}\left(t,\tau \right)=P_{{\mathrm{Pp}}_V{p}_H}^{\mathrm{LoS}}\left(t\right)·\delta \left(\tau -{\tau }_{p_V{p}_H}^{\mathrm{LoS}}\left(t\right)\right)+P_{p_V{p}_H,m_n}^{\mathrm{NLoS}}\left(t\right)·\delta \left(\tau -{\tau }_{p_V{p}_H,m_n}\left(t\right)\right),$

p_H, p_V denote the number of rows and the number of columns in an LED array.

When multi-users scenarios are taken into account, assuming that the number of base station is N_BS and the number of user is N_MS, a channel transmission matrix of a multi-link channel model is represented as a following formula:

$H_M={\left[\begin{array}{ccc}{H}_{{\mathrm{BS}}_1{\mathrm{MS}}_1}& \cdots & H_{{\mathrm{BS}}_1{\mathrm{MS}}_{N_{\mathrm{MS}}}}\\ ⋮& \ddots & ⋮\\ H_{{\mathrm{BS}}_{N_{\mathrm{BS}}}}& \cdots & H_{{\mathrm{BS}}_{N_{\mathrm{BS}}}{\mathrm{MS}}_{N_{\mathrm{MS}}}}\end{array}\right]}_{N_{\mathrm{BS}}\times N_{\mathrm{MS}}},$

H_BSiMSj, i=1, 2 . . . N_BS, j=1, 2 . . . N_MS corresponding to each link is a single-link channel model H described above.

The detailed steps for generating the channel coefficients are specifically as follows.

In S1, propagation scenarios and propagation conditions are set; the carrier frequency, the antenna type, the layout of the transceiver and the motion trajectory of the transceiver are determined.

In S2, the path loss, the shadowing, the oxygen absorption and the large scale fading of blockage effect are generated; the method mainly focuses on a modeling for the small-scale fading, and standard channel models for large scale fading are referable to the calculation of this part.

In S3, according to positions and motion conditions of the transceiver, large-scale parameters with spatial consistency for the DS and 4 angle spreads are generated.

Except SH, other corresponding large-scale parameters include DS, ASA, ASD, ESA, ESD, K_R and XPR. The generation method of large-scale parameters is the same. The generation of the DS is taken as an example herein and represented as the following formula:

${\mathrm{DS}}_{f_c}\left(P\right)={\mathrm{DS}}_{\mu ,f_c}+X^{\mathrm{DS}}\left(P\right)·{\mathrm{DS}}_{\sigma ,f_c},\mathrm{where}P\left(P^T,P^R\right)$

is composed of transceiver position vectors, P^T(t)=(x^T(t), y^T(t), z^T(t)) and P^R(t)=(x^R(t), y^R(t), z^R(t)) denote a coordinate vector at the transmitter side and a coordinate vector at the receiver side, respectively, and initial values of which are generated according to simulation environments and requirements; X^DS(P) denotes a normal distribution variable generated by the sine wave superposition method and following the spatial consistency with the mean value of 0 and the variance of 1, DS_μ,fc denotes a mean value for DS in a frequency f_c, and DS_σ,fc denotes a variance of DS in the frequency f_c, configuration values for DS_σ,fc are divided into three types according to the height h_UT of the user terminal; for terrestrial mobile communication scenarios 1.5 m≤h_UT≤22.5 m, values set from Table 7.5-6 of 3GPP TR 38.901 are referable; for UAV scenarios 22.5 m≤h_UT≤300 m, values set from Table B1.2 of 3GPP TR 36.777 standardization document are referable; for satellite communication scenarios, values set from Table 6.7-2 of 3GPP TR 38.811 standardization document are referable; in NLoS conditions of urban macro Uma scenarios, when a carrier frequency ranges from 2 to 4 GHz, DS_μ,fc is calculated as follows:

${\log }_{10}\left({\mathrm{DS}}_{\mu ,f_c}/1s\right)=\{\begin{array}{cc}-0.204{\log }_{10}\left(f_c\right)-6.28,& 1.5m<h_{\mathrm{UT}}\le 22.5m,\mathrm{NLoS}\\ 0.0965{\log }_{10}\left(h_{\mathrm{UT}}\right)-7.503,& 22.5m<h_{\mathrm{UT}}\le 300m,\mathrm{NLoS}\\ -7.21& \left(\mathrm{An}\mathrm{elevation}\mathrm{angle}\mathrm{of}\mathrm{the}\mathrm{link}\mathrm{is}10°\right)\end{array}$

All 8 large-scale parameters are independently generated by this method, values for all large-scale parameters with spatial consistency in a logarithm domain can be obtained by multiplying a cross-correlation matrix among the large-scale parameters, subsequently values in the logarithm domain are required to be converted into a linear domain; so that the large-scale parameters of the channel are obtained.

In S4, scatterers following an ellipsoid Gaussian scattering distribution are generated, delays, angles and powers of the clusters are calculated according to geographical location information of the transceiver and the scatterers, and channel coefficients are generated.

In S401, positions of the scatterers are obtained by using an ellipsoid Gaussian scattering distribution, the scatterers in the n-th cluster centered on (d_n^X, ϕ_E,n^X, ϕ_A,n^X) follow a Gaussian distribution with standard deviations of σ_x^X, σ_y^X and σ_z^X on three axes respectively; after obtaining the positions of the scatterers, the positions of the scatterers are converted into spherical coordinates; positions {right arrow over (C)}_mn^A(t₀) and {right arrow over (C)}_mn^Z(t₀) of the scatterers in the n-th cluster corresponding to a position of a first transmitter antenna {right arrow over (A)}₁^T(t₀) and a position of a first receiver antenna {right arrow over (A)}₁^R(t₀) are represented as {right arrow over (C)}_mn^A(t₀)=(d_mn^T(t₀), ϕ_A,mn^T(t₀), ϕ_E,mn^T(t₀)) and {right arrow over (C)}_mn^Z(t₀)=(d_mn^R(t₀), ϕ_A,mn^R(t₀), ϕ_E,mn^R(t₀)), where d_mn^X(t₀), ϕ_A,mn^X(t₀) and ϕ_E,mn^X(t₀) denote a distance, an azimuth angle, and an elevation angle of the m-th sub-path of the n-th cluster at the transmitter side or the receiver side, respectively, X∈{T, R} denotes the transmitter side and the receiver side.

In S402, in the multi-bounce channel model, delays of sub-paths in the cluster at the initial time instant are calculated by

${\tau }_{qp,m_n}\left(t_0\right)=\frac{\left(d_{p,m_n}^T\left(t_0\right)+d_{q,m_n}^R\left(t_0\right)\right)}{c}+{\tilde{\tau }}_{m_n}\left(t_0\right),$

where {tilde over (τ)}_mn denotes a delay of virtual links between {right arrow over (C)}_mn^A and {right arrow over (C)}_mn^Z, d_p,mn^T(t₀) denotes a distance between A_p^T and C_mn^A at the time instant t₀, and d_q,mn^R(t₀) denotes a distance between A_q^R and C_mn^Z, at the time instant t₀,

${\tilde{\tau }}_{m_n}\left(t_0\right)=\frac{{\tilde{d}}_{m_n}\left(t_0\right)}{c}+{\tau }_{\mathrm{link}}\left(t_0\right),$

{tilde over (d)}_mn(t₀) denotes a distance between the first-bounce cluster and the last-bounce cluster, τ_link denotes a non-negative variable following an exponential distribution.

In S403, in (ultra-)massive MIMO scenarios, a sub-paths power P_qp,mn,fc(t) in the clusters is varied along a time axis and an array axis, and the sub-paths power is commonly modeled as a lognormal process varying with time and a lognormal process varying with the array, a non-normalized sub-paths power P′_qp,mn,fc(t) in the clusters is:

$P_{\mathrm{qp},m_n,f_c}^{\prime }\left(t\right)=\underset{T\mathrm{he}\mathrm{time}\mathrm{domain}}{\underbrace{\exp \left(-{\tau }_{\mathrm{qp},m_n}\left(t\right)\frac{r_{\tau }-1}{r_{\tau }\mathrm{DS}}\right){10}^{-\frac{Z_n}{10}}}}·\underset{\mathrm{domain}}{\underset{\mathrm{The}\mathrm{space}}{\underbrace{{\xi }_n\left(p,q\right)}}},$

where Z_n denotes a per cluster shadowing term IN dB, r_τ denotes a delay distribution proportionality factor, ξ_n(p, q) denotes a two-dimensional spatial lognormal process for simulating smooth power variations over antenna arrays.

In wide bandwidth scenarios, the power value is multiplied by

${\left(\frac{f}{f_c}\right)}^{{\gamma }_{m_n}}$

in the frequency domain by taking frequency domain non-stationary characteristics into account, where γ_mn is a frequency-dependent constant factor, eventually, the ultimate power P_qp,mn,fc(t) of the sub-paths in the clusters is obtained by normalizing the powers of all clusters; if the clusters are newly generated, τ_qp,mn(t) is substituted with τ_qp,mn(t₀) to obtain the initial power of the m-th sub-path in the n-th cluster between A_p^T and A_q^R.

In S404, for the survived clusters, small-scale parameters such as the powers and the delays of the sub-paths in the clusters at different time instants are required to be updated. For the trajectory segment at the time instant t₁, that is, at the subsequent time instant after the clusters are generated, a coordinate of the p-th transmitter antenna A_p^T is:

${\overrightarrow{A}}_p^T\left(t_1\right)={\overrightarrow{A}}_p^T\left(t_0\right)+v^T\left(t_1-t_0\right)·{\left[\begin{array}{c}\cos {\alpha }_A^T·\cos {\alpha }_E^T\\ \sin {\alpha }_A^T·\cos {\alpha }_E^T\\ \sin {\alpha }_E^T\end{array}\right]}^T,$

where a coordinate {right arrow over (A)}_p^T(t₀) of the p-th transmitter antenna at the initial time instant is calculated by

${\overrightarrow{A}}_p^T\left(t_0\right)={\overrightarrow{A}}_1^T\left(t_0\right)+\left(p-1\right)·{\delta }_T·\left[\begin{array}{c}\cos {\beta }_A^T·\cos {\beta }_E^T\\ \sin {\beta }_A^T·\cos {\beta }_E^T\\ \sin {\beta }_E^T\end{array}\right]$

{right arrow over (C)}_mn^A(t₁) can be calculated by

${\overrightarrow{C}}_{m_n}^A\left(t_1\right)={\overrightarrow{C}}_{m_n}^A\left(t_0\right)+v^{A_n}\left(t_1-t_0\right)·{\left[\begin{array}{c}\cos {\alpha }_A^{A_n}·\cos {\alpha }_E^{A_n}\\ \sin {\alpha }_A^{A_n}·\cos {\alpha }_E^{A_n}\\ \sin {\alpha }_E^{A_n}\end{array}\right]}^T.$

The distance from A_p^T to the first-bounce cluster C_mn^A can be obtained by calculating d_p,mn^T(t₁)=∥{right arrow over (C)}_mn^A(t₁)−{right arrow over (A)}_p^T(t₁)∥ at the time instant t₁, similarly, the distance d_q,mn^R(t₁) from A_q^Z to C_mn^Z is obtained. A delay of the sub-path in the clusters at the time instant t₁ is τ_qp,mn(t₁)=(d_p,mn^T(t₁)+d_q,mn^R(t₁))/c+{tilde over (τ)}_mn; τ_qp,mn(t) and P_qp,mn(t) are obtained by using geographical locations of the transmitter, the receiver, and the scatterer at a previous time instant, (t=t₂, t₃, . . . ).

In S5, the large-scale parameters and the small-scale parameters are updated according to movements of the transceiver and birth-death processes of the clusters; and new channel coefficients are generated.

A space-time-frequency non-stationarity of the model is mainly reflected in two aspects, one is parameters of space-time-frequency variations, and another is the birth-death processes of the clusters in a space-time-frequency domain, the number of clusters at the time instant t is calculated as follows:

$N_{qp}\left(t\right)=N_{surv}\left(t\right)+N_{new}\left(t\right),$

where N_qp(t) denotes the number of the clusters, N_surv(t) denotes the number of survived clusters determined by a survived probability P_surv(Δt, Δr, Δf) of the clusters, N_new(t) denotes the number of newly generated clusters following the Poisson distribution with a mean value E[N_new(t)], λ_G is defined as a birth rate of the clusters, λ_R is defined as a combination rate (death rate) of the clusters. In order to model a space-time-frequency evolution process of the clusters more accurately, two types of sampling intervals are introduced, one type is a time domain sampling interval Δt, a frequency domain sampling interval Δf and a space domain (array domain) sampling interval Δr, and channel parameters are updated continuously, another type is described by Δt_BD, Δf_BD and Δr_BD that are integer multiples of corresponding Δt, Δf and Δr, and during the birth-death processes and the evolution processes of the clusters occurred at sampling points, survived probabilities of the transmitter side and receiver side clusters along the array axis and time axis are as follows:

$p_{\mathrm{surv}}^T\left(\Delta t_{\mathrm{BD}},{\delta }_p\right)=e^{-{{\lambda }_R\left({\left({ϵ}_1^T\right)}^2+{\left({ϵ}_2^T\right)}^2+2{ϵ}_1^T{ϵ}_2^T\mathrm{co}s\left({\alpha }_A^T-{\beta }_A^T\right)\right)}^{1/2}}$ $P_{\mathrm{surv}}^R\left(\Delta t_{\mathrm{BD}},{\delta }_q\right)=e^{-{{\lambda }_R\left({\left({ϵ}_1^R\right)}^2+{\left({ϵ}_2^R\right)}^2+2{ϵ}_1^R{ϵ}_2^R\mathrm{co}s\left({\alpha }_A^R-{\beta }_A^R\right)\right)}^{1/2}}$ $\mathrm{where}{ϵ}_1^T=\frac{{\delta }_p\cos {\beta }_E^T}{D_c^A}\left({ϵ}_1^R=\frac{{\delta }_q\cos {\beta }_E^R}{D_c^A}\right)\mathrm{and}{ϵ}_2^T=\frac{v^T\Delta t_{\mathrm{BD}}}{D_c^s}\left({ϵ}_2^R=\frac{v^R\Delta t_{\mathrm{BD}}}{D_c^s}\right)$

denote position differences of the transmitter antenna element and the receiver antenna element on the array axis and the time axis, respectively, D_c^A and D_c^S denote scenario-dependent factors on the array axis and the time axis, respectively, a joint survived probability of the transmitter side and receiver side clusters is represented as follows:

P _surv(Δt_BD, δ_p, δ_q)=P_surv^T(Δt_BD, δ_p)P_surv^R(Δt_BD, δ_q).

The average number of the newly generated clusters is:

$E\left(N_{new}\right)=\frac{{\lambda }_G}{{\lambda }_R}\left(1-P_{surv}\left(\Delta t_{\mathrm{BD}},\Delta r_{BD}\right)\right).$

When wide bandwidth scenarios are studied, the birth-death processes of the clusters also exist on a frequency axis, and a survival probability of the clusters on the frequency axis is:

$P_{surv}\left(\Delta f_{BD}\right)=e^{-{\lambda }_R\frac{F\left(\Delta f_{\mathrm{BD}}\right)}{D_c^f}},$

where F(Δf_BD) and D_c^f are determined by channel measurements, D_c^f denotes a scenario-dependent factor on the frequency axis, in summary, when the birth-death processes of the space-time-frequency domain clusters are taken into account, the survived probability of the clusters is:

P _surv(Δt_BD, Δr_BD, Δf_BD)=P_surv^T(Δt_BD, δ_p)P_surv^R(Δt_BD, δ_q)P_surv(Δf_BD).

The average number of the newly generated clusters is:

$E\left(N_{new}\right)=\frac{{\lambda }_G}{{\lambda }_R}\left(1-P_{surv}\left(\Delta t_{\mathrm{BD}},\Delta r_{\mathrm{BD}},\Delta f_{BD}\right)\right).$

In UHST scenarios, by taking account of a waveguide effect and an impact of tube wall roughness on channels in vacuum tube UHST scenarios, the average number of the newly generated clusters is:

$E\left(N_{new}\right)=\frac{{\lambda }_G}{{\lambda }_R}\left(1-P_{surv}\left(\Delta t_{\mathrm{BD}},\Delta r_{\mathrm{BD}},\Delta f_{BD}\right)\right)\left(1-\frac{D_{qp}\left(t\right)}{D}\right)·\frac{{\rho }_s}{{\rho }_{s_0}}$ ${\rho }_s=e^{\left(-8{\left(\frac{{\mathrm{\pi \sigma }}_h\mathrm{co}s(E\left[{\varphi }_{E,m_n}^T\right]}{\lambda }\right)}^2\right),}$

where D_qp(t) denotes a linear distance between the transmitter side and the receiver side at the time instant t, D denotes an initial distance between the transmitter side and the receiver side, ρ_s denotes a scattering coefficient of the tube wall, and ρ_s0 denotes a scattering coefficient when the roughness of σ_h=0.

Based on the above method and the geometric relationship between the transmitter, the receiver and the scatterers, the small-scale parameters of different antenna pairs can be obtained, so that all parameter values in the channel matrix can be obtained. The modeling method and corresponding parameters for the model are summarized in the following table.

TABLE 1
Model parameters and modeling methods
		Model parameters and
Scenarios	Channel Characteristics	modeling methods
All spectra	MmWave/	High delay resolution	Modeling the delay of sub-paths in
	THz		the clusters (τqp, mn(t))
	Channel	Frequency domain	1) Introducing birth-death process
		non-stationarity	of the cluster in frequency domain
			2) Power varying with
			frequency(   (t))
		Atmosphere	Considering the impact of oxygen
		absorption	absorption at mmWave
			band/molecular absorption at THz
			band on the received power
		Blockage effect	Considering the impact of blockage
			effect (BL) on the received power
	VLC	No multipath rapid fading	Merely modeling powers
	Channel	and negligible Doppler	(   (t)) and propagation
		effect	delays (τqp, mn(t))
		3D rotational receiver side	The angles (βAR(t), βER(t)) of the
			normal vector at the receiver are
			time-variant
		Special LED radiation mode	Supporting all LED radiation modes
			(FpHpV ({tilde over (θ)}pHpV, ET,
			{tilde over (θ)}pHpV, AT))
		Wavelength dependence	Modeling the effective reflectance
			parameters of clusters (ΓpHpV, n)
Global-	Satellite	Ionosphere Faraday effect	Modeling Faraday rotation matrix
coverage	Channel		(Fr)
scenarios		Rain attenuation	Modeling the rain attenuation (RA)
	UAV	3D motion	The motion having elevation
	Channel		directions
			(αET(t), αER(t),
			αEAn(t), αEZn(t))
		Channel difference relate to	Large-scale parameter generation
		the UAV height	following logarithmic Gaussian
			distribution, and the mean value and
			standard deviation of the
			distribution being highly correlated
			with UAVs height
	Maritime	Location	The LoS path component and
	Channel	dependence	multipath components of both
			rough ocean surface and
			evaporation duct over the sea
			surface being modeled as
			hqp, fcLoS(t, τ),
			hqp, fcNLoS1(t, τ) and
			hqp, fcNLoS2(t, τ), and the power
			coefficients KR, S1 and S1 being
			used to manipulate the
			disappearance and appearance of
			corresponding parts with
			variations of distances between
			two ships
		3D fluctuation of	Modeling the ship's 3D trajectory
		sea waves	using the classic Pierson-Moskowitz
			model
Full-	V2V	Arbitrary	The velocity of the transceiver and
application	Channel	trajectory	the cluster being modeled
scenarios		Multi-mobility	respectively, and the velocity size
		property	and direction being variable
			({right arrow over (ν)}T(t), {right arrow over (ν)}R(t),
			{right arrow over (ν)}An(t),
			{right arrow over (ν)}Zn(t))
	(U)HST	Large Doppler	Doppler shift (υD, qp, mn(t)) being
	Channel	shift	time-variant
		Time domain	1) Introducing birth-death process
		non-stationarity	of the cluster in time domain
			2) Channel parameters being time-
			variant
		Waveguide effect	The number of clusters (Nqp(t)) is
			modeled as being affected by the
			vacuum tube waveguide effect
	(U)massive	Spherical	Modeling the arrival angle and
	MIMO	wavefront characteristics	departure angle of each antenna
	Channel		separately
			(ϕA, mnT(t), ϕE, mnT(t),
			ϕA, mnR(t), and ϕE, mnR(t))
		Space domain	1) Introducing birth-death process
		non-stationarity	of the cluster in array domain
			2) Introducing a variation
			(ξn(p, q)) along the array axis in
			power    (t))
	RIS Channel	Cascaded	Introducing HIR, HTI & HTR and
		sub-channel	modeling the three sub-channels
			respectively
		Phase control	Introducing the phase-shift diagonal
			matrix Φ to implement intelligent
			control of channel environment
	IIoT Channel	Dense multipath component	Modeling the dense multipath
			(hqp, fcDMC(t, τ))
Common Characteristics	Spatial consistency	Using the SoS method to generate
		large-scale parameters with spatial
		consistency
	Multi-frequency correlation	1) PL being frequency dependent;
		2) In large-scale parameters, DS
		and angle spreads being related to
		frequency;
		3) Power (   (t)) being
		frequency dependent

3. Model Simplification

By adjusting the parameters, the 6GPCM can be simplified into multiple dedicated channel models, as illustrated in Table 2.

TABLE 2
Simplified summary table of 6GPCM
Scenarios
supported by
6GPCM	Simplified model	Parameter Adjustments
Multi-link	Single-link	1) NT = NR = 1
Multi-	Single-frequency	1) hqp,ƒc(t, τ) = hqp(t, τ)
frequency		2) Multi-band correlation being not considered in
		power Pqp,mn,ƒc(t) calculation
All spectra	Sub-6 GHz	1) OL = 1, BL =1
		2) Mn(t) = 1, each cluster merely having one scatterer,
		and modeling the path is modeled: τmn = τn, Pmn = Pn,
		and the like
		3) Psurv(ΔƒBD) = 1, γmn = 0
	MmWave/THz+ massive MIMO	1) Single-link; single-frequency 2) RA = 1, μ = 1, Mn(t) = Mn $F_r=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right],$
		Mn(t) = Mn
		3) The large-scale parameters following independent
		lognormal distributions.
	Indoor+VLC	1) Single-link; single-frequency; using single-cluster model;
		2) OL = 1, RA =1, μ = 1,
		$F_r=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right],$
		Mn(t) = Mn
		3) The transmitter side being a stationary LED array at the pH-
		th row and the pV-th column, vT = 0, MT = pH ×
		pV and the row and column intervals being δH and δV
		respectively; the receiver side being a photodiode that can
		move and rotate, MR = 1
		4) hpVpH (t, τ) = hpVpHLoS(t, τ) + hpVpHNLoS(t, τ) = PpVpHLoS(t).
		(τ − τpVpHLoS(t)) + PpVpH,mnNLoS(t) · δ (τ − τpVpH,mn(t))
		5) Psurv(ΔƒBD) = Psurv(ΔtBD) = 1, γmn = 0
		Array domains evolving in rows and columns.
Global-	LEO communication	1) Single-link; single-frequency
coverage	channel	2) OL = 1, BL =1, μ = 1, Mn(t) = Mn
scenarios		3) Psurv(ΔƒBD) = 1, γmn = 0
		4) Psurv(ΔrBD) = 1, ξn(p, q) = 1
	UAV communication channel	1) Single-link; single-frequency 2) OL = 1, BL = 1, RA = 1, μ = 1, $F_r=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right],$
		Mn(t) = Mn
		3) Psurv(ΔƒBD) = 1, γmn = 0
		4) ξn(p, q) = 1
	Maritime communication channel	1) Single-link; single-frequency 2) OL = 1, BL = 1, RA = 1, μ = 1, $F_r=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right],$ Mn(t) = Mn
		$3)h_{\mathrm{qp},f_c}\left(t,\tau \right)=\sqrt{\frac{K_R\left(t\right)}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{\mathrm{LoS}}\left(t,\tau \right)+\sqrt{\frac{S_1}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{{\mathrm{NLoS}}_1}\left(t,\tau \right)+\sqrt{\frac{S_2}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{{\mathrm{NLoS}}_2}\left(t,\tau \right)$
		4) Psurv(ΔƒBD) = 1, γmn = 0
		5) ξn(p, q) = 1
Full- application scenarios	V2V communication channel	1) Single-link; single-frequency 2) OL = 1, RA =1, μ = 1, $F_r=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right],$ Mn(t) = Mn
		3) Psurv(ΔƒBD) = 1, γmn = 0
		4) Psurv(ΔrBD) = 1, ξn(p, q) = 1
	MmWave+UHST	1) Single-link; single-frequency; clusters are distributed on
		the wall of the vacuum tube
		2) RA = 1, μ = 1,
		$F_r=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right],$
		Mn(t) = Mn
		3) vAn = 0, vZn = 0, vT = 0
		4) Psurv(ΔƒBD) = 1, γmn = 0
		5) Psurv(ΔrBD) = 1, ξn(p, q) = 1
	ultra-massive	1) Single-frequency
	MIMO	2) RA = 1, μ = 1,
		$F_r=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right],$
		Mn(t) = Mn
	RIS communication	1) Single-link; single-frequency
	channel	2) OL = 1, BL = 1, RA =1, μ = 1,
		$F_r=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right],$
		Mn(t) = Mn
		3) Psurv(ΔƒBD) = Psurv(ΔtBD) = 1, γmn = 0
		Array domains evolving in rows and columns.
	IIOT communication	1) Single-link; single-frequency
	channel	2) RA = 1, μ = 1,
		$F_r=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right],$
		$3)h_{\mathrm{qp},f_c}\left(t,\tau \right)=\sqrt{\frac{K_R\left(t\right)}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{\mathrm{LoS}}\left(t,\tau \right)+\sqrt{\frac{S_1}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{{\mathrm{NLoS}}_1}\left(t,\tau \right)+\sqrt{\frac{S_2}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{{\mathrm{NLoS}}_2}\left(t,\tau \right)$
		4) Psurv(ΔƒBD) = 1, γmn = 0
Pervasive	B5GCM	1) Single-link; single-frequency
		2) RA = 1,
		$F_r=\left[\begin{array}{cc}1& 0\\ 0& 1\end{array}\right],$
		Mn(t) = Mn
		3) Psurv(ΔƒBD) = 1
		4) The large-scale parameters following independent
		lognormal distribution

The present disclosure is described in detail in combination with the drawings and specific embodiments. The embodiments are implemented on the premise of the technical solutions of the present disclosure, and the specific embodiments and specific operation processes are given, but the protection scope of the present disclosure is not limited to the following embodiments.

By taking the scenario of the ultra-massive MIMO at millimeter band as an example, a channel matrix of a 6GPCM is represented as:

$H={\left[\mathrm{PL}·\mathrm{SH}·\mathrm{BL}·\mathrm{AL}\right]}^{1/2}·H_s,$

where PL denotes path loss, SH denotes shadowing, BL denotes blockage loss, AL denotes atmospheric gas absorption loss,

$H_s={\left[h_{\mathrm{qp},f_c}\left(t,\tau \right)\right]}_{M_R\times M_T},$

where M_T (M_R) denotes the number of antenna elements in the transmitter (receiver) antenna array, h_qp,fc(t, τ) denotes a channel impulse response between A_p^T and A_q^R, which is represented by the superposition of an LoS component h_qp,fc^LoS(t, τ) and a NLoS component h_qp,fc^NLoS(t, τ):

$h_{qp,f_c}\left(t,\tau \right)=\sqrt{\frac{K_R\left(t\right)}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{LoS}\left(t,\tau \right)+\sqrt{\frac{1}{K_R\left(t\right)+1}}{h}_{\mathrm{qp},f_c}^{NLoS}\left(t,\tau \right)$

where K_R(t) denotes a Rice factor, h_qp,fc^LoS(t, τ) and h_qp,fc^NLoS(t, τ) are respectively represented as follows

$h_{\mathrm{qp},f_c}^{\mathrm{LoS}}\left(t,\tau \right)={\left[\begin{array}{c}{F}_{q,f_c,V}\left({\varphi }_{E,L}^R\left(t\right),{\varphi }_{A,L}^R\left(t\right)\right)\\ F_{q,f_c,H}\left({\varphi }_{E,L}^R\left(t\right),{\varphi }_{A,L}^R\left(t\right)\right)\end{array}\right]}^T\left[\begin{array}{cc}{e}^{j{\theta }_L^{\mathrm{VV}}}& 0\\ 0& -e^{j{\theta }_L^{\mathrm{HH}}}\end{array}\right]F_r\left[\begin{array}{c}{F}_{p,f_{c}V}\left({\varphi }_{E,L}^T\left(t\right),{\varphi }_{A,L}^T\left(t\right)\right)\\ F_{p,f_c,H}({\varphi }_{E,L}^T\left({\varphi }_{E,L}^T\left(t\right),{\varphi }_{A,L}^T\left(t\right)\right)\end{array}\right]e^{j2\pi f_c{\tau }_{\mathrm{qp}}^L\left(t\right)}\delta \left(\tau -{\tau }_{\mathrm{qp}}^L\left(t\right)\right)$ $h_{\mathrm{qp},f_c}^{\mathrm{NLoS}}\left(t,\tau \right)=\sum _{n=1}^{N_{\mathrm{qp}}\left(t\right)}\sum _{m=1}^{M_n\left(t\right)}{\left[\begin{array}{c}{F}_{q,f_c.V}\left({\varphi }_{E,m_n}^R\left(t\right),{\varphi }_{A,m_n}^R\left(t\right)\right)\\ F_{q,f_c,H}\left({\varphi }_{E,m_n}^R\left(t\right),{\varphi }_{A,m_n}^R\left(t\right)\right)\end{array}\right]}^T\left[\begin{array}{cc}{e}^{j{\theta }_{m_n}^{\mathrm{VV}}}& \sqrt{\mu {\kappa }_{m_n}^{-1}\left(t\right)}{e}^{j{\theta }_{m_n}^{\mathrm{VH}}}\\ \sqrt{{\kappa }_{m_n}^{-1}\left(t\right)}{e}^{j{\theta }_{m_n}^{\mathrm{HV}}}& \sqrt{\mu }{e}^{j{\theta }_{m_n}^{\mathrm{HH}}}\end{array}\right]F_r\left[\begin{array}{c}{F}_{p,f_c,V}\left({\varphi }_{E,m_n}^T\left(t\right),{\varphi }_{A,m_n}^T\left(t\right)\right)\\ F_{p,f_c,H}\left({\varphi }_{E,m_n}^T\left(t\right),{\varphi }_{A,m_n}^T\left(t\right)\right)\end{array}\right]\sqrt{P_{\mathrm{qp},m_n,f_c}\left(t\right)}{e}^{j2\pi f_c{\tau }_{\mathrm{qp},m_n}\left(t\right)}\delta \left(\tau -{\tau }_{\mathrm{qp},m_n}\left(t\right)\right),$

where {*}^T denotes a transposition operation, f_c denotes a carrier frequency, F_p(q),fc,V and F_p(q),fc,H denote antenna patterns of the array element A_p^T (A_q^R) for vertical and horizontal polarizations at different frequency bands, κ_mn(t) denotes a cross polarization power ratio, μ denotes a co-polar imbalance, ϕ_A,mn^T(t) and ϕ_E,mn^T(t) denote an azimuth departure angle and an elevation departure angle corresponding to the m-th sub-path from A₁^T to C_n^A at the time instant t, ϕ_A,mn^R(t) and ϕ_E,mn^R(t) denote an azimuth arrival angle and an elevation arrival angle corresponding to the m-th sub-path from A₁^R to C_n^Z at the time instant t, ϕ_A,L^T(t) and ϕ_E,L^T(t) denote an azimuth departure angle and an elevation departure angle corresponding to the LoS path from A₁^T to A₁^R at the time instant t, ϕ_A,L^R(t) and ϕ_E,L^R(t) denote an azimuth arrival angle and an elevation arrival angle corresponding to the LoS path from A₁^T to A₁^R at the time instant t, θ_L^VV, θ_L^HH, θ_mn^VV, θ_mn^VH, θ_mn^HV and θ_mn^HH are random phases uniformly distributed over (0, 2π], P_qp,mn,fc(t) denotes a power of the m-th sub-path in the n-th path from A_p^T to A_q^R at the NLoS condition, τ_qp^L(t) denotes a delay of the LoS path at the time instant t,

${\tau }_{\mathrm{qp}}^L\left(t\right)=\frac{{\overrightarrow{d}}_{\mathrm{qp}}\left(t\right)}{c}$

{right arrow over (d)}_qp(t) denotes a vector distance between the transmitter antenna array A_p^T and the receiver antenna array A_q^R at the time instant t, c denotes a speed of light, τ_qp,mn(t) denotes a delay of the m-th sub-path in the n-th path from A_p^T to A_q^R at the time instant t, all of the above parameters are time-varying parameters.

The detailed steps for generating channel coefficients are specifically as follows.

In S1, propagation scenarios and propagation conditions are set; the carrier frequency, the antenna type, the layout of the transceiver and the motion trajectory of the transceiver are determined.

In S2, the path loss, the shadowing, the oxygen absorption and the large scale fading of blockage effect are generated; the method mainly focuses on a modeling for the small-scale fading, and standard channel models for large scale fading are referable to the calculation of this part.

In S3, according to positions and motion conditions of the transceiver, large-scale parameters with spatial consistency of the DS and 4 angle spreads are generated.

Except SH, other corresponding large-scale parameters include DS, ASA, ASD, ESA, ESD, K_R and XPR. The generation methods of large-scale parameters are the same. The delay extension DS generation is taken as an example herein and represented as the following formula:

$D{S}_{f_c}\left(P\right)=D{S}_{\mu ,f_c}+X^{DS}\left(P\right)·{\mathrm{DS}}_{\sigma ,f_c},$

where P=(P^T, P^R) is composed of transceiver position vectors, P^T(t)=(x^T(t), y^T(t), z^T(t)) and P^R(t)=(x^R(t), y^R(t), z^R(t)) denote a coordinate vector at the transmitter side and a coordinate vector at the receiver side, respectively, and initial values of which are generated according to simulation environments and requirements; X^DS(P) denotes a normal distribution variable generated by the sine wave superposition method and following the spatial consistency with the mean value of 0 and the variance of 1, DS_μ,fc denotes a mean value for DS in a frequency f_c, and DS_σ,fc denotes a variance of DS in the frequency f_c, configuration values for DS_σ,fc are divided into three types according to a height h_UT of a user terminal; the values in this embodiment can refer to Tables 7.5-6 in the 3GPP TR 38.901 standardization document. All 8 large-scale parameters are independently generated by this method, values for all large-scale parameters with spatial consistency in a logarithm domain can be obtained by multiplying a cross-correlation matrix among the large-scale parameters, subsequently values in the logarithm domain are required to be converted into a linear domain; so that the large-scale parameters of the channel are obtained.

In S4, scatterers following an ellipsoid Gaussian scattering distribution are generated, delays, angles and powers of the clusters are calculated according to geographical location information of the transceiver and the scatterers, and channel coefficients are generated.

In S401, positions of the scatterers are obtained by using an ellipsoid Gaussian scattering distribution, the scatterers in the n-th cluster centered on (d_n^X, ϕ_E,n^X, ϕ_A,n^X) follow a Gaussian distribution with standard deviations of σ_x^X, σ_y^X and σ_z^X on three axes respectively; after obtaining the positions of the scatterers, the positions of the scatterers are converted into spherical coordinates; positions {right arrow over (C)}_mn^A(t₀) and {right arrow over (C)}_mn^Z(t₀) of the scatterers in the n-th cluster corresponding to a position of a first transmitter antenna {right arrow over (A)}₁^T(t₀) and a position of a first receiver antenna {right arrow over (A)}₁^R(t₀) are represented as {right arrow over (C)}_mn^A(t₀)=(d_mn^T(t₀), ϕ_A,mn^T(t₀), ϕ_E,mn^T(t₀)) and {right arrow over (C)}_mn^Z(t₀)=(d_mn^R(t₀), ϕ_A,mn^R(t₀), ϕ_E,mn^R(t₀)), where d_mn^X(t₀), ϕ_A,mn^X(t₀) and ϕ_E,mn^X(t₀) denote a distance, an azimuth angle, and an elevation angle of the m-th sub-path of the n-th cluster at the transmitter side or the receiver side, respectively, X∈{T, R} denotes the transmitter side and the receiver side.

In S402, in the multi-bounce channel model, delays of sub-paths in the cluster at the initial time instant are calculated by

${\tau }_{qp,m_n}\left(t_0\right)=\frac{\left(d_{p,m_n}^T\left(t_0\right)+d_{q,n_n}^R\left(t_0\right)\right)}{c}+{\tilde{\tau }}_{m_n}\left(t_0\right),$

where {tilde over (τ)}_mn denotes a delay of virtual links between {right arrow over (C)}_mn^A and {right arrow over (C)}_mn^Z, d_p,mn^T(t₀) denotes a distance between A_p^T and C_mn^A at the time instant t₀, and d_q,mn^R(t₀) denotes a distance between A_q^R and C_mn^Z at the time instant t₀,

${\tilde{\tau }}_{m_n}\left(t_0\right)=\frac{{\tilde{d}}_{m_n}\left(t_0\right)}{c}+{\tau }_{\mathrm{link}}\left(t_0\right),{\tilde{d}}_{m_n}\left(t_0\right)$

denotes a distance between the first-bounce cluster and the last-bounce cluster, τ_link denotes a non-negative variable following an exponential distribution.

In S403, in (ultra-)massive MIMO scenarios, a sub-paths power P_qp,mn,fc(t) in the clusters is varied along a time axis, a frequency axis and an array axis, and the sub-paths power is commonly modeled as a lognormal process varying with time and a lognormal process varying with the array, a non-normalized sub-paths power P′_qp,mn,fc(t) in the clusters is:

$P_{\mathrm{qp},m_n,f_c}^{\prime }\left(t\right)=\underset{\mathrm{The}\mathrm{time}\mathrm{domain}}{\underbrace{\exp \left(-{\tau }_{\mathrm{qp},m_n}\left(t\right)\frac{r_{\tau }-1}{r_{\tau }\mathrm{DS}}\right){10}^{-\frac{Z_n}{10}}}}·\underset{\mathrm{The}\mathrm{space}\mathrm{domain}}{\underbrace{{\xi }_n\left(p,q\right)}},$

where Z_n denotes a per cluster shadowing term in dB, r_τ denotes a delay distribution proportionality factor, ξ_n(p, q) denotes a two-dimensional spatial lognormal process for simulating smooth power variations over antenna arrays.

At the mmWave band, in wide bandwidth scenarios, the power value is multiplied by

${\left(\frac{f}{f_c}\right)}^{{\gamma }_{m_n}}$

in the frequency domain by taking frequency domain non-stationary characteristic into account, where γ_mn is a frequency-dependent constant factor, eventually, the ultimate power P_qp,mn,fc(t) of the sub-paths in the clusters is obtained by normalizing the powers of all clusters; if the cluster are newly generated, τ_qp,mn(t) is substituted with τ_qp,mn(t₀) to obtain the initial power of the m-th sub-path in the n-th cluster between A_p^T and A_q^R.

In S404, for the survived clusters, small-scale parameters such as the powers and the delays of the sub-paths in the clusters at different time instants are required to be updated. For the trajectory segment at the time instant t₁, that is, at the subsequent time instant after the clusters are generated, a coordinate of the p-th transmitter antenna A_p^T is:

${\overrightarrow{A}}_p^T\left(t_1\right)={\overrightarrow{A}}_p^T\left(t_0\right)+v^T\left(t_1-t_0\right)·{\left[\begin{array}{c}\cos {\alpha }_A^T·\cos {\alpha }_E^T\\ \sin {\alpha }_A^T·\cos {\alpha }_E^T\\ \sin {\alpha }_E^T\end{array}\right]}^T,$

where a coordinate {right arrow over (A)}_p^T(t₀) of the p-th transmitter antenna at the initial time instant is calculated by

${\overrightarrow{A}}_p^T\left(t_0\right)={\overrightarrow{A}}_1^T\left(t_0\right)+\left(p-1\right)·{{\delta }_T\left[\begin{array}{c}\cos {\beta }_A^T·\cos {\beta }_E^T\\ \sin {\beta }_A^T·\cos {\beta }_E^T\\ \sin {\beta }_E^T\end{array}\right]}^T,{\overrightarrow{C}}_{m_n}^A\left(t_1\right)$

can be calculated by

${\overrightarrow{C}}_{m_n}^A\left(t_1\right)={\overrightarrow{C}}_{m_n}^A\left(t_0\right)+{v^{A_n}\left(t_1-t_0\right)\left[\begin{array}{c}\cos {\alpha }_A^{A_n}·\cos {\alpha }_E^{A_n}\\ \sin {\alpha }_A^{A_n}·\cos {\alpha }_E^{A_n}\\ \sin {\alpha }_E^{A_n}\end{array}\right]}^T.$

The distance from A_p^T to the first-bounce cluster C_mn^A can be obtained by calculating d_p,mn^T(t₁)=∥{right arrow over (C)}_mn^A(t₁)−{right arrow over (A)}_p^T(t₁)∥ at the time instant t₁, similarly, the distance d_q,mn^R(t₁) from A_q^Z to C_mn^Z is obtained. A delay of the sub-path in the clusters at the time instant t₁ is τ_qp,mn(t₁)=(d_p,mn^T(t₁)+d_q,mn^R(t_q))/c+{tilde over (τ)}_mn; τ_qp,mn(t) and P_qp,mn(t) are obtained by using geographical locations of the transmitter, the receiver, and the scatterer at a previous time instant, (t=t₂, t₃, . . . ).

In S5, the large-scale parameters and the small-scale parameters are updated according to the movements of the transceiver and the birth-death processes of the clusters; and new channel coefficients are generated.

A space-time-frequency non-stationarity of the model is mainly reflected in two aspects, one is parameters of space-time-frequency variations, and another is the birth-death processes of the clusters in a space-time-frequency domain, the number of clusters at the time instant t is calculated as follows:

$N_{qp}\left(t\right)=N_{surv}\left(t\right)+N_{new}\left(t\right),$

where N_qp(t) denotes the number of the clusters, N_surv(t) denotes the number of survived clusters, determined by a survived probability P_surv(Δt, Δr, Δf) of the clusters, N_new(t) denotes the number of newly generated clusters following the Poisson distribution with a mean value E[N_new(t)], λ_G is defined as a birth rate of the clusters, λ_R is defined as a combination rate (death rate) of the clusters. In order to model a space-time-frequency evolution process of the clusters more accurately, two types of sampling intervals are introduced, one type is a time domain sampling interval Δt, a frequency domain sampling interval Δf and a space domain (array domain) sampling interval Δr, and channel parameters are updated continuously, another type is described by Δt_BD, Δf_BD and Δr_BD that are integer multiples of corresponding Δt, Δf and Δr, and during the birth-death processes and the evolution processes of the clusters occurred at sampling points, survived probabilities of the transmitter side and receiver side clusters along the array axis and time axis are as follows:

$P_{\mathrm{surv}}^T\left(\Delta t_{\mathrm{BD}},{\delta }_p\right)=e^{-{{\lambda }_R\left({\left({ϵ}_1^T\right)}^2+{\left({ϵ}_2^T\right)}^2+2{ϵ}_1^T{ϵ}_2^T\mathrm{co}s\left({\alpha }_A^T-{\beta }_A^T\right)\right)}^{1/2}}$ $P_{\mathrm{surv}}^R\left(\Delta t_{\mathrm{BD}},{\delta }_q\right)=e^{-{{\lambda }_R\left({\left({ϵ}_q^R\right)}^2+{\left({ϵ}_2^R\right)}^2+2{ϵ}_1^R{ϵ}_2^R\mathrm{co}s\left({\alpha }_A^R-{\beta }_A^R\right)\right)}^{1/2}}$ $\mathrm{where}{ϵ}_1^T=\frac{{\delta }_p\cos {\beta }_E^T}{D_c^A}\left({ϵ}_1^R=\frac{{\delta }_q\cos {\beta }_E^R}{D_c^A}\right)\mathrm{and}{ϵ}_2^T=\frac{v^T\Delta t_{\mathrm{BD}}}{D_c^s}\left({ϵ}_2^R=\frac{v^R\Delta t_{\mathrm{BD}}}{D_c^s}\right)$

denote position differences of the transmitter antenna element and the receiver antenna element on the array axis and the time axis, respectively, D_c^A and D_c^S denote scenario-dependent factors on the array axis and the time axis, respectively, a joint survived probability of the transmitter side and receiver side clusters is represented as follows:

P _surv(Δt_BD, δ_p, δ_q)=P_surv^T(Δt_BD, δ_p)P_surv^R(Δt_BD, δ_q).

The average number of the newly generated clusters is:

$E\left(N_{new}\right)=\frac{{\lambda }_G}{{\lambda }_R}\left(1-P_{surv}\left(\Delta t_{\mathrm{BD}},\Delta r_{BD}\right)\right).$

When wide bandwidth scenarios are studied, the birth-death processes of the clusters also exist on a frequency axis, and a survived probability of the clusters on the frequency axis is:

$P_{surv}\left(\Delta f_{BD}\right)=e^{-{\lambda }_R\frac{F\left(\Delta f_{\mathrm{BD}}\right)}{D_c^f}},$

where F(Δf_BD) and D_c^f are determined by channel measurements, D_c^f denotes a scenario-dependent factor on the frequency axis. In summary, when the birth-death processes of the space-time-frequency domain clusters are taken into account, the survived probability of the clusters is:

P _surv(Δt_BD, Δr_BD, Δf_BD)=P_surv^T(Δt_BD, δ_p)P_surv^R(Δt_BD, δ_q)P_surv(Δf_BD).

The average number of the newly generated clusters is:

$E\left(N_{new}\right)=\frac{{\lambda }_G}{{\lambda }_R}\left(1-P_{surv}\left(\Delta t_{\mathrm{BD}},\Delta r_{\mathrm{BD}},\Delta f_{BD}\right)\right).$

Based on the above method and the geometric relationship between the transmitter, the receiver and the scatterers, the small-scale parameters of different antenna pairs can be obtained, so that all parameter values in the channel matrix can be obtained.

Claims

1. A method for pervasively modeling 6G channels for all frequency bands and all scenarios, wherein, in the modeling method, massive uniform linear arrays are adopted at both a transmitter side and a receiver side in a 6GPCM, and the model is a multi-bounce propagation model, where denotes a p-th array element of a transmitter antenna array, AqR denotes a q-th array element of a receiver antenna array, and a distance between the transmitter antenna array and the receiver antenna array is δT(δR); βAT(R) denotes an azimuth angle of the transmitter antenna array and the receiver antenna array in an xy plane, and βET(R) denotes an elevation angle of the transmitter antenna array and the receiver antenna array; for a n-th propagation path from ApT to AqR, n=1, 2, 3, . . . , Nqp(t), where CnA denotes a first-bounce cluster of the n-th path proximity to the transmitter side, CnZ denotes a last-bounce cluster proximity to the receiver side, and a propagation path between the two clusters is modeled as a virtual link; when a delay of the virtual link between the first-bounce cluster and the last-bounce cluster is zero, the model is reduced to a single-bounce model; besides, Nqp(t) is a number of paths from ApT to AqR at a time instant t corresponding to Nqp(t) cluster pairs in a double-cluster model with the first-bounce cluster and the last-bounce cluster in one-to-one correspondence with each other, and corresponding to Nqp(t) clusters in a single-cluster model; on a microscopic level, analyzing clusters CnA and CnZ on the n-th path, and Mn(t) scatterers are existed in the clusters, CmnA denotes an m-th scatterer in CnA, CmnZ denotes an m-th scatterer in CnZ; from a view point of the path, CmnA is understood as a scatterer connected by an m-th sub-path from ApT to CnA, and CmnZ is understood as a scatterer connected by an m-th sub-path from AqR to CnZ; besides, ϕA,mnT(t) and ϕE,mnT(t) denote an azimuth departure angle and an elevation departure angle corresponding to an m-th sub-path from A1T to CnA at the time instant t, ϕA,mnR(t) and ϕE,mnR(t) are an azimuth arrival angle and an elevation arrival angle corresponding to an m-th sub-path from A1R to CnZ at the time instant t; besides, motion conditions at the transmitter side, the receiver side and the clusters are modeled by the model respectively, and three-dimensional motions of an arbitrary speed and an arbitrary trajectory of a transceiver and the clusters are supported, where νT(t), νR(t), νAn(t), νZn(t) denote motion speeds at the transmitter side, the receiver side, the first-bounce cluster and the last-bounce cluster respectively, αAT(t), αAR(t), αAAn(t), αAZn(t), denote azimuth angles of the motor directions at the transmitter side, the receiver side, the first-bounce cluster and the last-bounce cluster respectively, αET(t), αER(t), αEAn(t), αEZn(t) denote elevation angles of the motor direction at the transmitter side, the receiver side, the first-bounce cluster and the last-bounce cluster respectively; a channel matrix of the 6GPCM is represented as:

2. The method for pervasively modeling 6G channels for all frequency bands and all scenarios according to claim 1, wherein the small-scale fading Hs is represented as:

3. (canceled)

4. The method for pervasively modeling 6G channels for all frequency bands and all scenarios according to claim 1, wherein when the method for pervasively modeling 6G channels is utilized in RIS scenarios, channels are divided into a sub-channel HTI from the transmitter side to the RIS, a sub-channel HIR from the RIS to the receiver side and a sub-channel HTR from the transmitter side to the receiver side, the three sub-channels are modeled respectively and a phase shift diagonal matrix Φ is introduced to implement an intelligent control for channel environments, calculation methods of HIR, HTI and HTR are the same as that of Hs merely with different parameter values and different distributions of clusters.

5. The method for pervasively modeling 6G channels for all frequency bands and all scenarios according to claim 1, wherein when the method for pervasively modeling 6G channels is utilized in modeling for VLC channels, on one hand, wavelengths of optical signals are extremely short, a size of the receiver is commonly multi-million wavelengths with no rapid signal fading on multi-wavelengths; on another hand, due to an incoherent light emitted by an LED light in a VLC communication system, the optical signals has no phase information, and no rapid signal fading is caused after a superposition of real-valued multipath signals at the receiver side with an exhibition on a slow-varying shadowing, therefore, although a current VLC model representation is a channel impulse response form of a multipath superposition, the representation is essentially a large-scale model of modeling PL and SH, that is, Hs=1, PL·SH=hpVpHLoS(t, τ)+hpVpHNLoS(t, τ)=PpVpHLoS(t)·δ(τ−τpVpHLoS(t))+PpVpH,mnNLoS(t)·δ(τ−τpVpH,mn(t)), pH, pV denote a number of rows and a number of columns in an LED array.

6. The method for pervasively modeling 6G channels for all frequency bands and all scenarios according to claim 1, wherein when the method for pervasively modeling 6G channels is utilized in multi-link scenarios: assuming that a number of base stations is NBS and a number of users is NMS, a channel transmission matrix of a multi-link channel model is represented as a following formula:

7. The method for pervasively modeling 6G channels for all frequency bands and all scenarios according to claim 1, wherein in the method for pervasively modeling 6G channels, detailed steps for generating the channel matrix H are specifically as follows: S1, setting propagation scenarios and propagation conditions, and determining a carrier frequency, an antenna type, a layout of the transceiver and a motion trajectory of the transceiver;S2, generating a path loss, a shadowing, an oxygen absorption and blockage effect loss; wherein the method mainly focuses on a modeling for a small-scale fading, and standard channel models for large scale fading are referable to the calculation of this part;S3, generating, according to positions and motion conditions of the transceiver, large-scale parameters with spatial consistency for a DS and 4 angle spreads;wherein, except SH, other corresponding large-scale parameters include DS, ASA, ASD, ESA, ESD, KR and XPR, a generation of the DS is represented as a following formula:

8. The method for pervasively modeling 6G channels for all frequency bands and all scenarios according to claim 7, wherein S4 is specifically as follows: S401, obtaining, by using an ellipsoid Gaussian scattering distribution, positions of the scatterers, wherein the scatterers in a n-th cluster centered on (dnX, ϕE,nX, ϕA,nX)) follow a Gaussian distribution with standard deviations of σxX, σyX and σzX on three axes respectively; converting, after obtaining the positions of the scatterers, the positions of the scatterers into spherical coordinates; positions {right arrow over (C)}mnA(t0) and {right arrow over (C)}mnZ(t0) of the scatterers in the n-th cluster corresponding to a position of a first transmitter antenna {right arrow over (A)}1T(t0) and a position of a first receiver antenna {right arrow over (A)}1R(t0) are represented as {right arrow over (C)}mnA(t0)=(dmnT(t0), ϕA,mnT(t0), ϕE,mnT(t0)) and {right arrow over (C)}mnZ(t0)=(dmnR(t0), ϕA,mnR(t0), ϕE,mnR(t0)), where dmnX(t0), ϕA,mnX(t0) and ϕE,mnX(t0) denote a distance, an azimuth angle, and an elevation angle of a m-th sub-path of the n-th cluster at the transmitter side or the receiver side, respectively, X∈{T, R} denotes the transmitter side and the receiver side;S402, calculating, in a multi-bounce channel model, delays of sub-paths in the cluster at an initial time instant by

9. The method for pervasively modeling 6G channels for all frequency bands and all scenarios according to claim 7, wherein in S5: in order to model a space-time-frequency evolution process of the clusters more accurately, two types of sampling intervals are introduced, one type is a time domain sampling interval Δt, a frequency domain sampling interval Δf and a space domain (array domain) sampling interval Δr, and channel parameters are updated continuously, another type is described by ΔtBD, ΔfBD and ΔrBD that are integer multiples of corresponding Δt, Δf and Δr, and during the birth-death processes and the evolution processes of the clusters occurred at sampling points, survival probabilities of the transmitter side and receiver side clusters along the array axis and time axis are as follows: