CN-121984809-A - Cross-domain channel rapid estimation method for high-altitude platform gesture sensing

CN121984809ACN 121984809 ACN121984809 ACN 121984809ACN-121984809-A

Abstract

The invention relates to the technical field of signal processing and channel estimation, and provides a high-altitude platform gesture-aware cross-domain channel rapid estimation method which comprises the following steps of S1, S2, constructing a UE-RIS-HAP uplink cascade channel model, S3, constructing an HAP receiving signal model based on the channel model established in S2, overlapping a plurality of time slots to obtain an integral receiving signal model, converting the integral receiving signal model into a virtual angle domain to obtain an angle domain cascade channel matrix and an angle domain receiving signal compressed sensing model corresponding to the angle domain cascade channel matrix, S4, constructing an HAP gesture dithering model, and determining an arrival angle interval and an optimal arrival angle estimated value of an HAP side through low-complexity dithering sensing arrival angle search. According to the three-stage progressive estimation architecture, the problem of beam mismatch caused by HAP gesture dithering is effectively solved, the calculation complexity of large-scale RIS channel estimation is obviously reduced, and the estimation accuracy is improved.

Inventors

YANG PENG
LI JINLONG
CAO XIANBIN
AN PUGUANG

Assignees

北京航空航天大学

Dates

Publication Date: 20260505
Application Date: 20260209

Claims (6)

1. The method for quickly estimating the cross-domain channel of the attitude sensing of the high-altitude platform is characterized by comprising the following steps of: s1, constructing a large-scale RIS-assisted single-antenna UE-to-HAP communication system model; s2, constructing a UE-RIS-HAP uplink cascade channel model; s3, constructing an HAP (hybrid automatic repeat request) receiving signal model based on the channel model established in the S2, overlapping a plurality of time slots to obtain an overall receiving signal model, and converting the overall receiving signal model into a virtual angle domain to obtain an angle domain cascade channel matrix and a corresponding angle domain receiving signal compressed sensing model; s4, constructing an HAP posture jitter model, and determining an arrival angle interval and an optimal arrival angle estimated value of an HAP side through low-complexity jitter sensing arrival angle search; Step S5, based on the HAP side arrival angle estimated value obtained in the step S4, carrying out preliminary dimension reduction on the received signal model, introducing hierarchical sparse prior, detecting an effective angle area at the RIS, and realizing further reduction on the dimension of the angle domain cascade channel matrix; and S6, carrying out channel parameter fine estimation in the angle subspace after dimension reduction based on the detected effective angle area.
2. The method for quickly estimating the cross-domain channel perceived by the attitude of the high-altitude platform according to claim 1, wherein the step S2 specifically comprises: Step S2-1, constructing a UE-to-RIS channel model, wherein the channel model takes the sparsity of the angle domain of the UE-to-RIS channel into consideration Expressed as: ; Wherein, the Representing the number of principal propagation paths, Is the first The complex gain of the strip path is used, Representing the kronecker product, which corresponds to the response of the UPA in two dimensions, And Respectively represent Shaft and method for producing the same An array steering vector in the axial direction; a joint array steering vector representing the RIS; And Respectively represent Shaft and method for producing the same Spatial frequency in the axial direction; step S2-2 constructing a channel model of RIS to HAP taking into account that the channel is dominated by the line of sight component, channel Expressed as: ; Wherein, the Is the path complex gain, superscript Represents the conjugate transpose of the object, Is the joint steering vector of the HAP receiving end, Is the joint guiding vector of the RIS transmitting end; , Respectively represent the HAP sites Shaft and method for producing the same The spatial frequency in the axial direction is, And Respectively represent the RIS sites Shaft and method for producing the same Spatial frequency in the axial direction.
3. The method for quickly estimating the cross-domain channel perceived by the attitude of the high-altitude platform according to claim 1, wherein the step S3 specifically comprises: S3-1, constructing an HAP receiving signal model for channel estimation based on the channel model constructed in the step S2; ; Wherein, the Is a known per power pilot; Is additive white gaussian noise; And Respectively represent time slots HAP receive vector sum of RIS phase shift vectors at the time, and , ; S3-2, obtaining an overall received signal model by superposing a plurality of time slots of the received channel model in the step S-31; Will be Stacking observation data of each time slot to obtain an integral receiving signal model : ; Wherein, the Represents the line-wise Khatri-Rao product, , In order to facilitate low complexity estimation, And All have a Cronecker structure, i.e And ; ; Is the effective noise vector, the first The individual elements are ; Step S3-3, converting the integral received signal model in the step S3-2 into a virtual angle domain representation to obtain an angle domain cascade channel matrix and a corresponding angle domain received signal compressed sensing model: in view of the sparsity of physical channels in the angular domain, the channels are represented in the virtual angular domain as Wherein Is an angular domain sparse concatenated channel matrix and since RIS-HAP is a LoS channel, Approximately only one row is non-zero, and since the UE-RIS is a sparse multipath channel, only a few elements in this row are non-zero, a dictionary matrix And From two directions The spatial frequency is uniformly sampled, and then the receiving model can be rewritten into a standard compressed perceptual form: 。
4. The method for quickly estimating the cross-domain channel perceived by the attitude of the high-altitude platform according to claim 1, wherein the step S4 specifically comprises: S4-1, constructing an HAP attitude shake model based on the statistical distribution of HAP attitude angle shake components; ; Wherein, the Representing the actual attitude angle vector of the HAP, Representing a desired attitude angle vector configured by the HAP flight control system, Representing the attitude angle fluctuation vector caused by the jitter effect; Representing the desired yaw, pitch and roll angles, respectively; jitter components representing yaw, pitch and roll angles, respectively, and all obey Gaussian distribution, i.e , , Wherein The corresponding variances; s4-2, deducing a search interval of a real arrival angle of the HAP side based on the HAP posture jitter model; Under the assumption of small-angle dithering, And A first order taylor expansion approximation can be utilized, the distribution of which can be modeled as a gaussian distribution And Based on the statistical characteristics of Gaussian distribution, we construct a method of using one to And The confidence interval of three times standard deviation of the center is used as the searching interval of the real arrival angle of the HAP side, namely And ; Step S4-3, searching in the real arrival angle searching interval to obtain the optimal HAP side arrival angle estimated value : In the interval obtained in step S4-2, the following problem can be solved by a gradient-increasing or exhaustive search (e.g., grid search) method to obtain an optimal HAP-side arrival angle estimation value : ; Wherein the method comprises the steps of 。
5. The method for quickly estimating the cross-domain channel perceived by the attitude of the high-altitude platform according to claim 1, wherein the step S5 specifically comprises: S5-1, obtaining a received signal model after dimension reduction based on an accurate arrival angle of the HAP side, and constructing a compressed sensing model of a Kroneck structure; ; Wherein the method comprises the steps of Is an angle domain sparse matrix of the effective angle at RIS; S5-2, introducing a hierarchical sparse prior to capture the clustering sparsity of the RIS angle domain sparse matrix in each direction; Specifically, let the (Or ) And (Or ) Respectively represent At the position of (Or ) The joint posterior distribution of all hidden variables can be decomposed into: ; the prior distribution of the specific layers is as follows: (1) Channel vector Obeying complex gaussian distribution, the variance of which is defined by And Precision parameters of two directions And Coupling control: ; Wherein the method comprises the steps of ; (2) Precision vector Obeying Bernoulli-gamma distribution to Direction is exemplified by the introduction of binary support vectors To indicate sparsity: ; Wherein the method comprises the steps of And Respectively the conditions are And Lower part(s) Parameters of (2); Expression of (2) Similarly; (3) Support vector Obeying Bernoulli distribution to The following are examples: ; Wherein the method comprises the steps of Is that The sparsity of the direction is noted And Robust to mismatch, both values set to To obtain good performance; (4) Noise accuracy Obeys the gamma distribution: ; S5-3, iteratively solving the approximate posterior distribution of the hidden variable by utilizing a Cronecker-variability decibel leaf inference algorithm; All hidden variables are processed Is approximated as the product of the independent distributions of the variables: By minimizing approximate distribution And true posterior distribution The Kullback-Leibler divergence between the two, alternately and iteratively updating the distribution parameters of each variable, wherein the specific iteration process comprises the following substeps: (1) Updating channel vectors Is an approximate posterior of (a) : Obeys complex gaussian distribution Cronecker structure using a perceptual matrix Singular value decomposition is performed on the obtained product: And According to the above-mentioned decomposition, the method, Can be re-expressed as Wherein Reconstruction-based , Mean of (2) Sum covariance diagonal element Can be obtained by low complexity operations: ; ; Wherein the method comprises the steps of , ; (2) Updating precision vectors , The approximate posterior of the precision vector obeys the gamma distribution product form, and by utilizing symmetry, And (3) with Is in the same form as the updated formula of (c), only exchange is required And (3) with Subscript of (2) to The following are examples: ; The distribution parameters are updated as follows: ; ; (3) Updating support vectors , The approximate posterior of the support vector obeys the Bernoulli distribution product form, and the same is just to The following are examples: ; wherein posterior probability The updating is as follows: ; (4) Updating noise accuracy Approximate posterior compliance gamma distribution for noise accuracy The parameters are updated as follows: ; ; repeating the updating step until the variation of the posterior probability of the support vector meets the convergence criterion Wherein Is a preset tolerance value; s5-4, judging an effective angle area of the RIS side based on posterior probability of the support vector; Obtaining approximate posterior probabilities of support vectors through steps S5-1 through S5-3 And Thereafter, the following rules are used to determine respectively Shaft and method for producing the same Axial effective index set And : ; ; Wherein the threshold value Controlling the severity of the selection based on the estimate And Obtaining and constructing final effective angle area by adopting union strategy 。
6. The method for quickly estimating the cross-domain channel perceived by the attitude of the high-altitude platform according to any one of claims 1 to 5, wherein the method for quickly estimating the cross-domain channel perceived by the attitude of the high-altitude platform is applicable to a high-altitude platform communication system in which a large-scale RIS is deployed in an air-space integrated network, and is used for solving the problem of beam mismatch caused by mechanical vibration and airflow disturbance of the platform.

Description

Cross-domain channel rapid estimation method for high-altitude platform gesture sensing Technical Field The invention relates to the technical field of signal processing and channel estimation, in particular to a cross-domain channel rapid estimation method for high-altitude platform gesture sensing. Background With the development of sixth generation (Thesixthgeneration, 6G) mobile communication technology, an air-to-ground integrated network is considered as a key architecture for realizing global seamless coverage. Emerging high-altitude platform (HighAltitudePlatform, HAP) networks are envisioned as key components in an aerospace-ground integrated network. The high-altitude platform (HighAltitudePlatform, HAP) is usually deployed on a stratosphere with the height of about 20 km, has the characteristics of wide coverage, flexible deployment, low operation cost, long residence time and the like, and is considered as an important component of an air-space-ground integrated network. HAPs are capable of providing high directivity, high capacity communication services for terrestrial users by combining Multiple-input Multiple-Output (MIMO) and beamforming technologies. In the HAP communication link with terrestrial users, line-of-Sight (LoS) propagation dominates. Aiming at the problem of LoS path shielding possibly caused by complex ground environments such as urban building dense areas or valleys, a reconfigurable intelligent surface (ReconfigurableIntelligentSurface, RIS) has become an important technical example for improving propagation conditions. RIS is generally composed of a plane composed of a large number of electromagnetic wave passive reflecting elements, and deployment of RIS can establish virtual LoS links, thereby improving coverage capability of communication system. To overcome the multiplicative fading caused by RIS reflections and to ensure link quality, large-scale passive RIS arrays are typically employed. In an actual running environment, the HAP is affected by unstable airflow of a stratosphere and mechanical vibration of the HAP, and the attitude of a platform of the HAP can be changed unstably, namely the platform shake is generated. Such dithering can result in changes in the direction of the signal transmit and receive beams. In a HAP communication system based on large-scale RIS assistance, obtaining accurate channel state information (ChannelStateInformation, CSI) is the basis for achieving efficient communication. The channel estimation of the system involves high-dimensional channel parameter acquisition for large-scale arrays and capture of dynamic spatial geometry due to platform attitude changes. Disclosure of Invention (One) solving the technical problems Aiming at the defects of the prior art, the invention provides a method for quickly estimating the cross-domain channel of the gesture perception of a high-altitude platform, which solves the problems in the background art. (II) technical scheme The invention specifically adopts the following technical scheme that the method for quickly estimating the cross-domain channel of the attitude sensing of the high-altitude platform comprises the following steps: s1, constructing a large-scale RIS-assisted single-antenna UE-to-HAP communication system model; s2, constructing a UE-RIS-HAP uplink cascade channel model; s3, constructing an HAP (hybrid automatic repeat request) receiving signal model based on the channel model established in the S2, overlapping a plurality of time slots to obtain an overall receiving signal model, and converting the overall receiving signal model into a virtual angle domain to obtain an angle domain cascade channel matrix and a corresponding angle domain receiving signal compressed sensing model; s4, constructing an HAP posture jitter model, and determining an arrival angle interval and an optimal arrival angle estimated value of an HAP side through low-complexity jitter sensing arrival angle search; Step S5, based on the HAP side arrival angle estimated value obtained in the step S4, carrying out preliminary dimension reduction on the received signal model, introducing hierarchical sparse prior, detecting an effective angle area at the RIS, and realizing further reduction on the dimension of the angle domain cascade channel matrix; and S6, carrying out channel parameter fine estimation in the angle subspace after dimension reduction based on the detected effective angle area. Further, the step S2 specifically includes: Step S2-1, constructing a UE-to-RIS channel model, wherein the channel model takes the sparsity of the angle domain of the UE-to-RIS channel into consideration Expressed as: Wherein, the Representing the number of principal propagation paths,Is the firstThe complex gain of the strip path is used,Representing the kronecker product, which corresponds to the response of the UPA in two dimensions,AndRespectively representShaft and method for producing the sameAn array steering vector in t