Massive MIMO-OFDM physical channel fingerprint and channel map construction and channel acquisition method

Abstract

The present application proposes a large-scale MIMO-OFDM physical channel fingerprint and channel atlas construction and channel acquisition method. The present application first defines a new physical channel fingerprint, with the help of a cluster-based geometric channel model, a set of low-dimensional parameters are used to fully characterize the physical propagation characteristics of the channel; secondly, the channel atlas construction method is proposed, the relationship between the physical channel parameters and the space-frequency-time domain channel is used to extract the physical channel fingerprint from the channel measurement sample, and the physical channel atlas with position index is constructed; finally, the channel information acquisition method based on the physical channel atlas is also proposed, which can generate beam domain statistical channel information adapted to the current system configuration with low complexity to assist instantaneous channel estimation. The present application overcomes the limitations of existing channel fingerprint technology by using physical channel fingerprint representation; by replacing the traditional online detection method with a physical channel atlas, the system channel detection overhead is effectively reduced, and the channel information acquisition efficiency is improved.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology. It proposes a method for constructing physical channel fingerprints (PCF) and channel maps for large-scale MIMO-OFDM systems, and proposes a channel acquisition method based on physical channel maps. Background Technology

[0002] With the rapid development of 6G mobile communication technology, the system will achieve significant improvements in user terminal (UT) connection density, data transmission rate, and positioning accuracy. As a core technology of 5G networks, massive MIMO-OFDM technology, with its high spectral efficiency and strong anti-interference capabilities, will continue to play a crucial role in 6G communication systems. However, with the dramatic increase in UT connection density and the continuous expansion of antenna array size, the system faces new technical challenges, among which the difficulty and complexity of acquiring Channel State Information (CSI) are particularly prominent. Statistical CSI (sCSI), due to its slow time-varying characteristics, is often used as important prior information to assist in acquiring instantaneous CSI; however, using traditional long-term online detection methods to acquire sCSI leads to huge resource overhead and computational complexity. To address this technical bottleneck, existing research has begun to explore new approaches to acquire sCSI using location information, leveraging the technical advantages of 6G systems. Specifically, large-scale MIMO-OFDM systems can accurately acquire high-resolution channel parameters (including path gain, angle, and delay), and thanks to the dense deployment of UT nodes, the system can obtain more refined spatial sampling data. Based on this characteristic, by using base stations and UT devices as fixed and mobile sensors, a high-precision statistical parameter database, namely a channel fingerprint (CF) map, can be constructed. This location-indexed map can directly provide the statistical channel parameters of the current area, effectively avoiding the resource overhead of traditional online detection methods. However, existing CF maps still have significant shortcomings in terms of storage efficiency, parameter completeness, and system adaptability, which seriously restricts their practical application value in large-scale MIMO-OFDM systems. Summary of the Invention

[0003] Objective: To address the shortcomings of existing technologies, this invention aims to propose a method for constructing physical channel fingerprints and channel maps for large-scale MIMO-OFDM. By constructing a location-indexed physical channel map, it overcomes the limitations of existing channel fingerprints in terms of storage efficiency and practicality. Furthermore, this invention proposes a channel acquisition method based on physical channel maps, directly generating statistical channel information from physical channel fingerprints. This achieves efficient, low-overhead channel acquisition without online probing, providing reliable technical support for 6G and future mobile communication systems.

[0004] Technical Solution: To achieve the above-mentioned objectives, the present invention provides the following technical solution:

[0005] In a first aspect, the present invention provides a method for constructing physical channel fingerprints and channel maps for large-scale MIMO-OFDM, comprising the following steps:

[0006] Divide the target communication area into a grid and establish a location coordinate index;

[0007] For each location grid, physical feature parameters are extracted based on the acquired channel samples to construct a physical channel fingerprint. Specifically, the physical parameters of each multipath component in each channel sample are clustered to obtain physical propagation path clusters between the transceiver and receiver. The physical feature parameters of a single physical propagation path cluster include: cluster power, mean and standard deviation of the departure angle of the multipath components within the cluster, mean and standard deviation of the arrival angle of the multipath components within the cluster, and mean and standard deviation of the delay of the multipath components within the cluster. The physical channel fingerprint refers to the set of physical feature parameters corresponding to all physical propagation path clusters between the current user terminal location and the base station location.

[0008] Each location coordinate is associated with its corresponding physical channel fingerprint and stored to construct a physical channel map with a location index.

[0009] In some embodiments, the method for constructing the physical channel fingerprint includes:

[0010] Obtain multiple channel samples within the current location area;

[0011] Extract physical parameters such as path gain, departure angle, arrival angle, and delay of each multipath component in each channel sample;

[0012] Multipath components with similar angle-delay characteristics are clustered into propagation path clusters; the statistical characteristics of multipath parameters within each propagation path cluster are analyzed, and finally the characteristic parameters of all propagation path clusters are merged to form a complete physical channel fingerprint for that location.

[0013] In some embodiments, the channel samples are obtained through uplink probing, specifically including:

[0014] All antenna elements configured in the mobile terminal periodically send time-frequency two-dimensional pilot signals to the base station. After receiving the signals, the base station uses information geometry algorithms or sparse Bayesian methods to estimate the three-dimensional channel in the space-frequency-time domain, thereby obtaining channel samples.

[0015] In some embodiments, the extraction of physical parameters corresponding to the multipath components involves taking the physical channel parameters corresponding to the multipath components as the object to be solved after estimating the channel, constructing an optimization problem, and minimizing the mean square error between the spatial-frequency-time domain channel samples reconstructed by different multipath components and the detected spatial-frequency-time domain channel samples. The constructed optimization problem is then solved using a spatial alternation generalized expectation-maximization algorithm or a compressed sensing algorithm.

[0016] Secondly, the present invention provides a method for acquiring large-scale MIMO-OFDM channel information based on physical channel maps, comprising the following steps:

[0017] Based on the method for constructing physical channel fingerprints and channel maps for large-scale MIMO-OFDM described in the first aspect, a physical channel map is constructed.

[0018] The physical channel map is queried based on the location information of the user terminal to obtain the corresponding physical channel fingerprint; based on the physical channel fingerprint and combined with the terminal's mobility characteristics, statistical channel information is generated; the statistical channel information is used as prior information to assist in instantaneous channel estimation.

[0019] In some embodiments, the statistical channel information is a beam domain channel power distribution, including a one-dimensional spatial beam domain power distribution vector, a two-dimensional spatial-frequency beam domain power distribution matrix, and a triple spatial-frequency-time beam domain power distribution tensor, etc.; wherein, the spatial beam domain, frequency beam domain, and time beam domain can be regarded as quantized sampling of the angle domain, time delay domain, and Doppler domain, respectively, and the power distribution in the beam domain can also be regarded as the quantized form of the angle-time delay domain continuous power distribution function.

[0020] In some embodiments, the generation of the statistical channel information includes:

[0021] Based on the power distribution function in the angle-delay domain generated by the physical channel fingerprint, and combined with the mobility of the user terminal, the channel covariance matrix in the spatial domain / space-frequency domain / space-frequency-time domain is calculated.

[0022] Based on the mapping relationship between the spatial domain / space-frequency domain / space-frequency-time domain covariance matrix and the power distribution vector in the one-dimensional spatial beam domain / power distribution matrix in the two-dimensional spatial-frequency beam domain / power distribution tensor in the triple spatial-frequency-time beam domain, the acquisition of statistical channel information is established as an optimization problem. The goal is to minimize the difference between the channel covariance matrix in the spatial domain / space-frequency domain / space-frequency-time domain calculated from the physical channel fingerprint and the covariance matrix in the spatial domain / space-frequency domain / space-frequency-time domain calculated from the power distribution vector in the one-dimensional spatial beam domain / power distribution matrix in the two-dimensional spatial-frequency beam domain / power distribution tensor in the triple spatial-frequency-time beam domain to achieve optimal acquisition of statistical channel information.

[0023] Taking the triple-beam base channel model as an example, the generation of triple-beam domain statistical channel information includes:

[0024] Physical channel fingerprints are found based on physical channel maps, and the angle-delay domain power distribution function of the current position is reconstructed based on the channel fingerprints. A spatial-frequency-time domain covariance matrix is ​​generated by leveraging the relationship between the spatial-frequency-time domain channel covariance matrix and the angle-delay domain power distribution.

[0025] The power distribution tensor of the triple beam domain to be solved is decomposed into a dot product of a real tensor and itself. This real tensor is used as the solution object. Another space-frequency-time domain covariance matrix is ​​generated by using the mapping relationship between the space-frequency-time domain channel and the triple beam domain channel.

[0026] To measure the difference between two space-frequency-time domain covariance matrices, the two space-frequency-time domain covariance matrices are transformed using a rudder matrix in the space-frequency-time domain to obtain two tensors of the same dimension. An optimization problem is then established, with the goal of minimizing the Kullback-Leibler divergence between the two tensors of the same dimension. The optimization problem is solved using the gradient descent method to obtain the solution for the real tensor.

[0027] The solution of the real tensor is multiplied by itself to obtain the tensor, which is the desired triple beam domain statistical channel information.

[0028] Thirdly, the present invention provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when loaded onto the processor, implements the steps of the methods described in the first and / or second aspects above.

[0029] Fourthly, the present invention provides a computer program product, comprising a computer program that, when executed by a processor, implements the steps of the methods described in the first and / or second aspects above.

[0030] Beneficial effects: Compared with the prior art, the present invention has the following advantages: The physical channel fingerprint proposed in the present invention overcomes the limitations of existing channel fingerprints in terms of storage efficiency and practicality, can more comprehensively reflect the physical characteristics of the current propagation environment, and occupies relatively small storage space; The method of generating statistical CSI by means of physical channel map proposed in the present invention does not rely on the traditional online detection process, effectively avoids the additional overhead and data processing complexity caused by detection, and realizes efficient and low-overhead statistical CSI acquisition. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of a cluster-based geometric random channel model in an embodiment of the present invention.

[0032] Figure 2 This is a schematic diagram of the physical channel spectrum in an embodiment of the present invention.

[0033] Figure 3 This is a schematic diagram of the process for obtaining large-scale MIMO-OFDM channel information based on physical channel maps, as proposed in an embodiment of the present invention.

[0034] Figure 4 This is a performance diagram of the channel information acquisition method proposed in the embodiments of the present invention. Detailed Implementation

[0035] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0036] The method for constructing physical channel fingerprints and channel maps for large-scale MIMO-OFDM disclosed in this invention includes: in a large-scale MIMO-OFDM system, dividing the target communication area into several grids and establishing a location coordinate index; for each location grid, extracting the physical feature parameters of that location based on the acquired channel samples, and then constructing a physical channel fingerprint; associating and storing each location coordinate with its corresponding physical channel fingerprint to construct a physical channel map of the location index.

[0037] In a specific implementation, grid division ensures that the physical statistical characteristics of the corresponding channel at each location point remain approximately unchanged within the same grid area, and each grid has its corresponding location coordinates.

[0038] The physical characteristic parameters in this embodiment include the statistical characteristic parameters of each physical propagation path cluster between the transceiver and the receiver in the angle-delay domain. Each physical propagation path cluster consists of several multipath components with similar angle-delay domain distribution characteristics. The power distribution of the multipath components within the same physical propagation path cluster in the angle and delay domains can be fitted using a Gaussian distribution function. The physical characteristic parameters of a single physical propagation path cluster include: cluster power, the mean and standard deviation of the departure angles of the multipath components within the cluster, the mean and standard deviation of the arrival angles of the multipath components within the cluster, and the mean and standard deviation of the delays of the multipath components within the cluster. The physical channel fingerprint refers to the set of physical characteristic parameters corresponding to all physical propagation path clusters between the current user terminal location and the base station location. This physical channel fingerprint allows for the reconstruction of the joint power distribution function in the angle-delay domain to fully characterize the channel statistical characteristics. Furthermore, this fingerprint depends only on the physical propagation environment characteristics of the current location and is independent of antenna configuration, system parameters, and terminal mobility.

[0039] In this embodiment, the physical channel map is a database composed of physical channel fingerprints with location indices. In some embodiments, the map content can be dynamically updated by adjusting the location grid division precision of the propagation area.

[0040] Based on the large-scale MIMO-OFDM physical channel fingerprint and channel map construction method disclosed in the above embodiments, the large-scale MIMO-OFDM channel information acquisition method based on physical channel map disclosed in this embodiment includes: querying the physical channel map according to the location information of the user terminal to obtain the corresponding physical channel fingerprint; generating beam domain statistical channel information adapted to the current system configuration based on the physical channel fingerprint and combined with the terminal's mobility characteristics; the statistical channel information can be used as prior information to assist in instantaneous channel estimation.

[0041] Assisted instantaneous channel estimation refers to using beam domain statistical channel information generated from physical channel fingerprints as prior information in the instantaneous channel estimation process to obtain an instantaneous channel estimate with the minimum mean square error. Existing research can be consulted for details on using prior statistical channel information to assist instantaneous channel estimation. The key point of this invention is generating beam domain statistical channel information based on physical channel fingerprints.

[0042] Figure 1A schematic diagram of a propagation scenario according to an embodiment of the present invention is provided, wherein a base station (BS) communicates with multiple user terminals (UTs) on the ground. Both the BS and UT are configured with uniform linear arrays (ULAs), which represent different propagation paths between the BS and UT using different physical propagation path clusters. Each cluster contains multipath components with similar propagation characteristics.

[0043] Figure 2 A schematic diagram of the physical channel map in an embodiment of the present invention is provided. First, the current propagation scenario is divided into several grids. This grid division ensures that the physical statistical characteristics of the channel remain stable within the same grid region. Each grid has its corresponding location coordinates. Each grid region has a corresponding PCF (Physical Channel Function). All grids and their corresponding PCFs together constitute the physical channel map describing the current scenario.

[0044] Figure 3 A schematic flowchart of the large-scale MIMO-OFDM channel information acquisition method based on physical channel maps proposed in this invention is provided. The target communication area is divided into several grids, and a location coordinate index is established. For each location point, a set of physical characteristic parameters of the channel is obtained, generating a physical channel fingerprint for that location. The coordinates of each location point are associated and stored with their corresponding physical channel fingerprints to construct a physical channel map based on the location index. The physical channel map is queried based on the real-time location information of the user terminal. After obtaining the corresponding physical channel fingerprint, it is converted into the required statistical channel information, achieving probe-free acquisition of statistical channel information.

[0045] The following describes in detail the specific implementation process of the large-scale MIMO-OFDM physical channel fingerprint and channel map construction and channel acquisition method involved in this invention, using specific communication system examples. It should be noted that the method of this invention is not only applicable to the specific system model mentioned in the example below, but also applicable to other system models with different configurations.

[0046] I. Physical Channel Fingerprint

[0047] Consider as Figure 1The propagation environment shown depicts communication between a base station (BS) and several user terminals (UTs) on the ground. Assuming both the user and base station sides are configured with uniform linear arrays (ULAs), the propagation environment can be mapped to a two-dimensional space. As can be seen from the figure, the propagation environment between the BS and UTs can be described by several physical propagation path clusters, each cluster containing several multipath components (MPCs) with similar characteristics. Each cluster corresponds to a propagation path in the environment, and each MPC corresponds to a sub-path within that path. Figure 2 In this example, the current scene is divided into several grids. For ease of analysis, this embodiment assumes that the propagation environment is the same within each grid region. In the following description, s will be used. BS s and s represent the grid positions of BS and UT, respectively. This represents the set of all grids.

[0048] Based on the above configuration, assuming that the motion area of ​​the UT is located within the same grid during the analysis time, the Channel Impulse Response (CIR) between the a-th antenna element on the BS side and the b-th antenna element on the UT side can be expressed as:

[0049]

[0050] Where L(s) represents the number of MPCs between BS and UT. and f represents the complex gain and delay corresponding to the l-th MPC, respectively. c This represents the carrier frequency. Furthermore, MPCs with similar characteristics are grouped together, (1) which can be further written as...

[0051]

[0052] in and Let L represent the complex gain and delay corresponding to the l-th MPC in the p-th cluster, respectively, where P(s) represents the number of clusters between BS and s, and L represents the time delay. p (s) represents the number of MPCs in the current cluster, and has Established. For the uplink, the departure angle (AoD, Angel of Departure) and arrival angle (AoA, Angel of Arrival) of the current subpath are respectively represented by... and This is expressed as follows. Under the far-field assumption, the time delay can be approximated as...

[0053]

[0054] Where d BS and d UT These represent the element spacing on both sides of BS and UT, respectively. It is the propagation delay between position BS and position s, where c represents the speed of light, and v is the propagation delay between position BS and position s. move and α move Let (3) represent the terminal's moving speed and moving direction, respectively. Substituting (3) into (2), CIR can be further expressed as:

[0055]

[0056] Among them, Doppler frequency shift Where λ is the carrier wavelength of the system.

[0057] Considering the uplink in a MIMO-OFDM architecture, using N c N g These represent the number of subcarriers and the length of the cyclic prefix, respectively, with Δf representing the subcarrier spacing. Assume that within one symbol time T... sym The channel state information remains unchanged within the symbol set, but changes between symbols due to the Doppler effect. Under this assumption, the channel response on the k-th subcarrier of the n-th OFDM symbol can be obtained as follows:

[0058]

[0059] For any Define vector in The definition is given in (4), representing the current MPC gain. Then, when the cluster number P(s) and the number of MPCs within the cluster L... p (s) and the corresponding MPC When the information is known, the channel response corresponding to the current location can be reconstructed. Let Θ α =[θ AoD ,θ AoA ,τ] T Represents an arbitrary angle-time delay domain vector, and ν = v move cos(α move -θ AoD ) / λ represents the Doppler frequency shift, and (5) can be written as

[0060]

[0061] in The channel function in the angle-delay domain is represented.

[0062] make This represents the power distribution function of the channel between BS and s in the angle-delay domain. As a type of statistical CSI, this function systematically describes the physical propagation characteristics of the current location and can also be converted into various statistical CSIs for channel transmission. Therefore, the set of all location tags... S corresponding to all positions s (Θ α This can form a channel map corresponding to the current scenario, storing physical channel information at different locations.

[0063] In practical applications, S is a type of multidimensional function. s (Θ α It cannot be directly stored. Therefore, this embodiment approximates it by using S. s (Θ α Approximately represented as

[0064]

[0065] in This represents the average power of the p-th cluster. Let AoD, AoA, and latency represent the probability distribution functions of the current cluster's MPC, respectively. These can typically be fitted using a Gaussian distribution. For example, this function can be approximated as

[0066]

[0067] in and Let represent the mean and variance of the current cluster AoD, respectively. Similarly, It can also be represented in a similar form, with the characteristic function using... and Let each represent a vector. Thus, the vectors are defined as follows:

[0068] It contains the physical statistical properties of the p-th cluster at the current location. Let the set... but This summarizes all the physical statistics between the current location s and BS. When the mobility of the UT is known, according to... The provided information can be used to infer the channel's distribution characteristics in the Doppler domain. (The last part, "S," appears to be a typo and can be omitted.) s (Θ α )use Alternatively, using the Physical Channel Fingerprint (PCF) as the current location can effectively compress the storage space of the channel fingerprint.

[0069] The PCF in this embodiment of the invention is for Figure 2 The scenario shown is specifically defined, but this method can also be extended to more complex scenarios. For example, in a 3D propagation scenario, the elevation angle parameter can also be incorporated. In its definition, regardless of the scenario, the design philosophy of PCF is to fully characterize the physical characteristics of complex channel environments through the most concise set of parameters.

[0070] II. PCF Generation Methods under Large-Scale MIMO-OFDM Architecture

[0071] consider Figure 2 In the propagation scenario, assume the antenna dimensions on both sides of BS and UT are A×1 and B×1, respectively. Select K effective subcarriers along the subcarrier domain, and let their subcarrier index set be... Along the time dimension, the time resource is divided into several subframes, each subframe containing N. s 10 OFDM symbols, and further subdivide these subframes into N p There are N time slots, each time slot consisting of N b Composed of OFDM symbols, then there are N s =N p N b This is valid. Since the first OFDM in each time slot is used to transmit pilot signals, during channel sounding, the total number of OFDM symbols in the pilot portion of each frame is N. p indivual.

[0072] To study the generation of PCF, this embodiment of the invention first presents the relationship between the spatial-frequency-time (SFT) domain channel and parameters in PCF based on the aforementioned frame structure. Frame-by-frame analysis is then performed, assuming the current time slot number is... Then the current time slot and its predecessor (N) p -1) time slots together constitute the current frame. Let The OFDM symbol number contained in the pilot band of this frame is: Therefore, A×K×N can be defined. p ×B-dimensional tensor This indicates the SFT domain channel between the UTs at positions BS and s, along all valid subcarriers on the OFDM symbol of the current frame pilot band, where...

[0073]

[0074] make To represent the vector form of the SFT domain channel, we have:

[0075]

[0076] The rudder vector in the SFT domain satisfy

[0077]

[0078] When the spatial broadband effect can be ignored, the above rudder vector can be simplified to:

[0079]

[0080] The space rudder vector on the AoA side is: and Frequency domain rudder vector is and The space rudder vector on the AoD side is and Time rudder vector is and These rudder vectors represent the mapping relationships from the angle domain and time delay domain to the space-frequency-time domain, respectively.

[0081] The aforementioned channel model establishes the relationship between the SFT domain channel and MPC parameters. Specifically, through mapping between the angle domain, time domain, and SFT domain, the effects of different MPCs in the angle-delay domain are mapped to the SFT domain, and the effects of different MPCs in the SFT domain are summed to obtain the final representation of the SFT domain channel. Since PCF is a statistical characteristic of MPC parameters, sufficient SFT domain channel samples need to be obtained through multiple measurements. The effects of each MPC are then separated from the channel samples, and the MPC parameters and their statistical characteristics are extracted. First, this embodiment of the invention provides a method for obtaining SFT domain channel samples. This embodiment selects a two-dimensional time-frequency joint phase-shifted pilot (TFPSP) as the probe pilot signal. When the antenna array at the UT end simultaneously transmits the probe pilot to the BS side, the received signal model can be expressed as follows:

[0082]

[0083] in and This represents the received signal of the a-th antenna element on the BS side in the k-th effective subcarrier of the n-th pilot band OFDM symbol in the current frame. (Operator ×) n Represents the n-modular product between a matrix and a tensor. and It is a diagonal matrix. Let Then the pilot signal transmitted by the b-th antenna element on the n-th pilot band OFDM symbol on the k-th effective subcarrier is: The TFPSP design uses phase-shifted pilots to modulate the Zadoff-Chu sequence, i.e.

[0084]

[0085] Where φ b and γ represents the phase shift modulation factor in the frequency domain and the time domain, respectively. f and γ t Represent the roots of the Zadoff-Chu sequence, respectively, and K and N. p Coprime. Let ν max =2v move / λ、 as well as When B≤K f K t At that time, the phase shift modulation factor can be set to

[0086]

[0087] The operators <> and These represent the modulo operation and the floor operation, respectively. When B > K f K t At this point, Zadoff-Chu sequences with different roots can be generated as root sequences to increase pilot capacity. Then, estimation algorithms based on information geometry or sparse Bayesian methods can be used to estimate the pilot capacity based on the received signal at the BS. estimate

[0088] In the estimated Next, sufficient MPC parameter samples need to be extracted using channel data for statistical analysis to obtain the physical statistical characteristics of the channel. The problem of extracting MPC parameters can be solved by constructing an optimization problem. Specifically, by searching for MPC parameters, the mean square error between the SFT domain channel obtained by accumulating these MPCs in the SFT domain effect and the SFT domain channel samples obtained by detection is minimized, thus obtaining the required MPC parameters.

[0089] make Indicates to be estimated The optimization problem for finding a set of MPC parameters can be expressed as: finding a set of MPC parameters such that...

[0090]

[0091] Where the function Vector Θ=(α s ,θ AoD ,θ AoALet , τ) represent any set of MPC parameters. This problem can be solved using the Space Alternating Generalized Expectation-maximization (SAGE) algorithm or compressed sensing-based algorithms. Since the number of MPCs is unknown beforehand in actual parameter estimation, it is usually set according to the type of environment in which the channel exists; in scenarios with abundant multipath scattering, this value is typically set larger. During the solution process, the original optimization problem can be decomposed into several sub-problems using the aforementioned algorithms, thereby transforming the parallel multipath component parameter extraction problem into a sequential extraction of multipath components, reducing the dimensionality of the optimization problem.

[0092] This section uses SAGE as an example to illustrate the solution of the optimization problem, transforming the original parallel multidimensional parameter search problem into a sequential estimation problem. When estimating the l-th group of MPC parameters, let

[0093]

[0094] The optimization problem is then:

[0095]

[0096] After performing steps (19)-(20) of the SAGE algorithm, until the estimated value is obtained... After grouping the MPC parameters, the Kernel-Power-Means (KPM) algorithm can be used to cluster the MPC parameters, obtaining MPC parameter samples for each cluster. The number of clusters can be determined using the Davies-Bouldin (DB) criterion. After obtaining the MPC parameter samples for each cluster, a Gaussian distribution function can be used to fit the parameter distributions of angle and time delay to obtain the PCF. Within the coverage area of ​​the current cell, the PCF corresponding to different grid locations and the grid coordinates are associated and stored to obtain the PCF map.

[0097] III. Channel Information Acquisition Method Based on PCF Map

[0098] After obtaining the PCFs at different locations and constructing the PCF map, it can be used as reference information to generate the required sCSI for channel acquisition and precoding under a given system configuration and UT mobility. In practical applications, sCSI usually refers to the sCSI under the Beam Based Channel Model (BBCM), including the power distribution vector in the one-dimensional spatial beam domain, the power distribution matrix in the two-dimensional spatial-frequency beam domain, or the power distribution tensor in the triple spatial-frequency-time beam domain. This embodiment presents a modeling method for triple BBCM in the scenario where B=1, and thereby details the method for generating the triple beam domain power distribution tensor. Specifically, by quantizing the angle cosine, time delay, and Doppler domain, the spatial, frequency, and time beam domains are obtained respectively, and then the SFT domain channel between the BS and UT is represented as an A×K×N p The tensor, i.e.

[0099]

[0100] Where V s V f and The rudder matrices represent the spatial, frequency, and time domains, respectively. Under the beam-based model, these rudder matrices can be generated using elementary transformations of the DFT matrix. It is the channel tensor in the triple beam (TB) domain, satisfying in

[0101]

[0102] Indicates the current number Channel gain within a beam range Represents the first TB field One beam interval.

[0103] In real-world environments, because the number of propagation paths is finite, channels typically exhibit sparse characteristics in the beam domain.

[0104] Define the power distribution tensor in the TB domain Right now Given its sparsity and the fact that sCSI varies slowly over time, This is an important class of channel covariance matrix (sCSI), commonly used in various stages such as channel acquisition, precoding design, and power allocation. According to the definition of the covariance matrix, the channel covariance matrix in the SFT domain... In large-scale MIMO-OFDM systems, assuming that the channel gains of different beams satisfy mutually independent cyclic symmetric complex Gaussian distributions, then we have:

[0105]

[0106] Established, among which It is important to note that both the power distribution vector in the spatial beam domain and the two-dimensional power distribution matrix in the space-frequency domain can serve as channel fingerprints, but the sCSI under triple BBCM, i.e. It cannot be directly used as a channel fingerprint because it also includes Doppler domain channel characteristics, which are jointly determined by the physical channel characteristics of the current location and the mobility of the UT. Therefore, given that the PCF and UT mobility at location s are known, it is possible to... Perform derivation to avoid online detection acquisition The resulting expenses.

[0107] The following describes the generation of mobility using PCF and UT. The method, whose main steps include: generating an SFT-domain covariance matrix by utilizing the relationship between the SFT-domain channel covariance matrix and the angle-delay domain power distribution; and applying the solution to the... The problem is decomposed into a dot product of a real tensor and itself. This real tensor is used as the solution object. By leveraging the mapping relationship between the SFT domain channel and the TB domain channel, another SFT domain covariance matrix is ​​generated. To measure the difference between the two types of SFT domain covariance matrices, the two covariance matrices are transformed using the rudder matrix V to obtain two tensors of the same dimension. An optimization problem is established, with minimizing the Kullback-Leibler (KL) divergence between the two tensors as the optimization objective. The gradient descent method is used to solve the optimization problem, and the solution corresponding to the convergence of the algorithm is obtained. This solution is a real tensor. The dot product of this real tensor with itself is the tensor obtained, which is the TB domain sCSI.

[0108] The method described above will be explained below with reference to specific embodiments. (Regarding location...) When its corresponding PCF, i.e. Given that, then the different clusters correspond to as well as All of these can be obtained, and thus the power distribution function S in the AAD domain can be obtained. s (Θ α By leveraging the relationship between the channel parameters in the SFT domain and the AAD domain, and the approximate assumption that different paths in a large-scale MIMO-OFDM system are uncorrelated, an S... s (Θ α )and The following relationship exists between them.

[0109]

[0110] According to (23) and (24), the beam domain sCSI can be obtained by means of PCF in the following steps: First, based on the UT position information s, the PCF corresponding to the current position is obtained by looking up the PCF map, that is... Based on PCF, S is calculated. s (Θ α Then, substituting into (24), we obtain the covariance matrix. With the help of (23) mid-beam domain sCSI and The relationship between them, find a non-negative real tensor Make Established.

[0111] To solve for the nonnegative real tensor that meets the requirements We transform the original problem as follows: Based on its nonnegativity, It can be decomposed into the dot product of two real tensors, i.e. Solving Convert to solution Make it satisfy

[0112]

[0113] In order to make the values ​​at both ends of (25) as close as possible, this embodiment will make the values ​​at both ends of (25) as close as possible. The vectors are rearranged into two The tensors, respectively, are used as follows and Indication. For measurement and To address the similarity, this embodiment introduces the Kullback-Leibler (KL) divergence. and The KL divergence between them can be expressed as

[0114]

[0115] The smaller the KL divergence mentioned above, the better. and The higher the similarity, the better. Therefore, solving... The problem ultimately transforms into the following optimization problem.

[0116]

[0117] During implementation, It can be calculated as follows:

[0118]

[0119] in

[0120] T s=((V) s ) H V s )⊙((V s ) H V s ) * (29)

[0121] T f =((V) f ) H V f )⊙((V f ) H V f ) * (30)

[0122]

[0123] Problem (27) is a convex problem, therefore it can be solved by gradient descent, that is, in the d-th iteration, we have

[0124]

[0125] Where δ (d) This represents the step size of the iteration. To prevent divergence caused by excessively fast convergence, it is usually decreased as the number of iterations increases. The gradient calculation formula is:

[0126]

[0127] tensor satisfy When the iterations converge, the real tensor that satisfies the conditions can be obtained. The estimated value of sCSI in the TB domain is

[0128] The above method can also be extended to other types of sCSI acquisition. Based on the mapping relationship between the spatial domain / space-frequency domain covariance matrix and the power distribution vector of the one-dimensional spatial beam domain / power distribution matrix of the two-dimensional spatial-frequency beam, sCSI acquisition is established as an optimization problem similar to (27). The optimal statistical channel information acquisition is achieved by minimizing the difference between the spatial domain / space-frequency domain channel covariance matrix calculated by the physical channel fingerprint and the spatial domain / space-frequency domain covariance matrix calculated by the power distribution vector of the one-dimensional spatial beam domain / power distribution matrix of the two-dimensional spatial-frequency beam domain. The above differences can be measured by Kullback-Leibler divergence. The optimization problem is solved by gradient descent.

[0129] IV. Implementation Results

[0130] To enable those skilled in the art to better understand the present invention, the performance results of the channel information acquisition method in this embodiment under specific configurations are given below.

[0131] Consider a large-scale MIMO-OFDM system with the following system parameters: carrier frequency f c = 5.8GHz, number of subcarriers N c = 2048, cyclic prefix length N g =144, subcarrier spacing Δf = 15kHz, effective subcarrier number K = 360, base station antenna element number M = 128, element spacing is half wavelength, number of users in the system U = 300, number of time slots N in one frame along the time domain p =8, N is the number of OFDM symbols in a single time slot. b =14, user terminal moving speed 15km / h. Oversampling factor F ν =4. This embodiment of the invention uses a 6G Universal Channel Model (6GPCM) generator to generate channel data. The specific implementation steps are as follows: First, a PCF map corresponding to the current cell is constructed based on the PCF generation method; then, according to the proposed technical solution, the PCF is converted into sCSI under BBCM under a specific system configuration and UT mobility state. To verify the accuracy of the generated sCSI, it is compared and analyzed with the sCSI obtained by the traditional online detection method. Both sCSIs are applied to the channel estimation process. If their channel estimation performance is comparable, it confirms that the sCSI generated based on PCF can effectively replace the online detection sCSI.

[0132] The specific simulation process includes two key steps: extracting the SFT domain channel from the channel measurement results and generating the PCF, and generating the corresponding sCSI from the PCF based on the UT's current location and movement state to assist in channel acquisition. It should be noted that the PCF is configuration-independent and can support sCSI generation under any antenna configuration; during the channel measurement phase, to improve spatial resolution and ensure angle estimation accuracy, both the BS and UT are configured with multi-antenna arrays. In this embodiment, a 32×1 uniform linear array is used at the UT end during PCF generation; after the PCF database is constructed, this embodiment verifies the effectiveness of sCSI generation based on the PCF in scenarios where multiple single-antenna UTs communicate with the BS.

[0133] This embodiment follows the method described above, performing channel estimation and extracting MPC parameters in the SFT domain to generate PCFs with a grid density of 1m×1m. After generating PCFs for all location points, a PCF database based on location indexes is constructed. To verify the practical value of PCFs and the effectiveness of the proposed sCSI generation method, the sCSIs generated by PCFs are compared with those obtained by traditional online detection methods. Traditional online detection-based sCSI acquisition methods (such as sparse Bayesian methods based on compressed sensing (CS)) model sCSIs as sparse signals and have been proven to have high accuracy; therefore, this method is used to generate SCFs as a comparison benchmark.

[0134] This embodiment considers a large-scale MIMO-OFDM channel acquisition scenario with U=180. The initial position of UT is randomly distributed within the scenario area, the moving speed is 15km / h, and the moving direction is randomly selected within the range of 0 to π. Simulation verification is performed under two typical scenarios: "3GPP_38.901_UMa_NLOS" and "3GPP_38.901_RMa_NLOS". The feasibility of generating sCSI by PCF is verified by applying the online detection method and the PCF-based method proposed in this patent to the channel acquisition process.

[0135] In the channel acquisition process, both pilot design and channel estimation stages require sCSI as prior information. Since the performance of channel acquisition depends on the accuracy of sCSI, and sCSI directly reflects its precision, Normalized Mean Square Error (NMSE) is used as the evaluation metric. Figure 4 This paper presents a comparison of the NMSE performance of different sCSI-assisted channel acquisition methods in two scenarios: UMa and RMa. These include: SF Probing: sCSI in the space-frequency (SF) beam domain acquired through online sounding; SFPCF-based: SF beam domain sCSI generated from the PCF using the proposed method; TB Probing: TB domain sCSI acquired through online sounding; and TBPCF-based: TB domain sCSI generated from the PCF using the proposed method. The NMSE comparison of the four sCSI methods fully verifies the feasibility of generating multi-dimensional sCSI from the PCF. Simulation results show that, regardless of whether it is SF beam domain or TB domain sCSI, the sCSI generated from the PCF is similar in channel acquisition NMSE performance to that based on online sounding. Furthermore, triple BBCM exhibits stronger Doppler domain channel feature capture capability compared to dual BBCM. Figure 4 As shown, since TB domain sCSI can more comprehensively characterize the characteristics of multi-domain channels, it can achieve better NMSE performance than SF beam domain sCSI in channel acquisition.

[0136] In summary, this invention first proposes a method for defining physical channel fingerprints, using a cluster-based geometric channel model to comprehensively characterize the physical propagation characteristics of the channel through a set of low-dimensional parameters. Second, it proposes a method for constructing physical channel fingerprints for large-scale MIMO-OFDM systems, utilizing the relationship between physical channel parameters and the spatial-frequency-time domain channel to extract physical channel fingerprints from channel measurement samples in the spatial-frequency-time domain. These physical channel fingerprints are then associated and stored with location information to construct a location-indexed physical channel map. Finally, it proposes a channel information acquisition method based on the physical channel map. By querying the physical channel fingerprint corresponding to the user terminal's location in real time and combining it with terminal mobility characteristics, it generates beam domain statistical channel information adapted to the current system configuration with low complexity, and uses this as prior information to achieve efficient instantaneous channel estimation. Experiments show that this invention overcomes the limitations of existing channel fingerprinting technologies through physical channel fingerprint characterization; by replacing traditional online detection methods with physical channel maps, it effectively reduces system channel detection overhead and improves the efficiency of channel information acquisition.

[0137] Based on the same inventive concept, an embodiment of the present invention discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded onto the processor, it implements the method for constructing physical channel fingerprints and channel maps for large-scale MIMO-OFDM, and / or the method for acquiring large-scale MIMO-OFDM channel information based on physical channel maps.

[0138] In a specific implementation, the device includes a processor, a communication bus, a memory, and a communication interface. The processor can be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present invention. The communication bus may include a path for transmitting information between the aforementioned components. The communication interface, using any transceiver-like device, is used for communicating with other devices or communication networks. The memory can be read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions, random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), read-only optical disc (CD-ROM) or other optical disc storage, disk storage media, or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto. The memory can exist independently and be connected to the processor via a bus. The memory can also be integrated with the processor.

[0139] The memory stores application code that executes the present invention and is controlled by the processor. The processor executes the application code stored in the memory to implement the method provided in the above embodiments. The processor may include one or more CPUs, or multiple processors, each of which may be a single-core processor or a multi-core processor. Here, "processor" may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0140] This invention also discloses a computer program product, including a computer program that, when executed by a processor, implements the method for constructing physical channel fingerprints and channel maps for large-scale MIMO-OFDM, and / or a method for acquiring large-scale MIMO-OFDM channel information based on physical channel maps.

[0141] The program code used to implement the method of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the steps of the method of the present invention to be performed. The program code can be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a standalone software package, or entirely on a remote machine or server. All aspects not detailed in this invention are well-known to those skilled in the art.

[0142] In the embodiments provided in this application, it should be understood that the disclosed methods can be implemented in other ways without departing from the spirit and scope of this application. The current embodiments are merely exemplary examples and should not be considered limiting, nor should the specific content given limit the purpose of this application. For example, some features may be omitted or not implemented.

[0143] The technical means disclosed in this invention are not limited to those disclosed in the above embodiments, but also include technical solutions composed of any combination of the above technical features. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications are also considered within the scope of protection of this invention.

Claims

1. A method for constructing physical channel fingerprints and channel maps for large-scale MIMO-OFDM, characterized in that, Includes the following steps: Divide the target communication area into a grid and establish a location coordinate index; For each location grid, physical feature parameters are extracted based on the acquired channel samples to construct a physical channel fingerprint. Specifically, the physical parameters of each multipath component in each channel sample are clustered to obtain physical propagation path clusters between the transceiver and receiver. The physical feature parameters of a single physical propagation path cluster include: cluster power, mean and standard deviation of the departure angle of the multipath components within the cluster, mean and standard deviation of the arrival angle of the multipath components within the cluster, and mean and standard deviation of the delay of the multipath components within the cluster. The physical channel fingerprint refers to the set of physical feature parameters corresponding to all physical propagation path clusters between the current user terminal location and the base station location. Each location coordinate is associated with its corresponding physical channel fingerprint and stored to construct a physical channel map with a location index. The channel samples are obtained through uplink probing, including: all antenna array elements configured by the mobile terminal periodically send time-frequency two-dimensional pilot signals to the base station; after receiving the signals, the base station uses an estimation algorithm based on information geometry or a sparse Bayesian method to estimate the three-dimensional channel in the space-frequency-time domain, thereby obtaining the channel samples. The extraction of physical parameters corresponding to the multipath components involves taking the physical channel parameters corresponding to the multipath components as the objects to be solved after estimating the channel, constructing an optimization problem, and minimizing the mean square error between the spatial-frequency-time domain channel samples reconstructed by different multipath components and the detected spatial-frequency-time domain channel samples. The constructed optimization problem is then solved using a spatial alternation generalized expectation-maximization algorithm or a compressed sensing algorithm.

2. The method for constructing physical channel fingerprints and channel maps for large-scale MIMO-OFDM according to claim 1, characterized in that: The construction of the physical channel fingerprint includes: Obtain multiple channel samples within the current location area; Extract the path gain, departure angle, arrival angle, and delay of each multipath component in each channel sample; cluster multipath components with similar angle-delay characteristics into propagation path clusters; Analyze the statistical characteristics of multipath parameters within each propagation path cluster, and merge the characteristic parameters of all propagation path clusters to form a complete physical channel fingerprint for that location.

3. A method for acquiring large-scale MIMO-OFDM channel information based on physical channel maps, characterized in that, Includes the following steps: The physical channel map is constructed according to the method for constructing physical channel fingerprints and channel maps of large-scale MIMO-OFDM as described in claim 1 or 2. Based on the location information of the user terminal, query the physical channel map to obtain the corresponding physical channel fingerprint; Based on physical channel fingerprinting and combined with terminal mobility characteristics, statistical channel information is generated; the statistical channel information is used as prior information to assist in instantaneous channel estimation.

4. The method for acquiring large-scale MIMO-OFDM channel information based on physical channel maps according to claim 3, characterized in that: The statistical channel information is the channel power distribution in the beam domain, including the power distribution vector in the one-dimensional spatial beam domain, the power distribution matrix in the two-dimensional spatial-frequency beam domain, or the power distribution tensor in the triple spatial-frequency-time beam domain; wherein, the spatial beam domain, frequency beam domain, and time beam domain are respectively regarded as quantized samples of the angle domain, time delay domain, and Doppler domain, and the power distribution in the beam domain is regarded as the quantized form of the continuous power distribution function in the angle-time delay domain.

5. The method for acquiring large-scale MIMO-OFDM channel information based on physical channel maps according to claim 3, characterized in that: The generation of the statistical channel information includes: Based on the power distribution function in the angle-delay domain generated by the physical channel fingerprint, and combined with the mobility of the user terminal, the channel covariance matrix in the spatial domain / space-frequency domain / space-frequency-time domain is calculated. Based on the mapping relationship between the spatial domain / space-frequency domain / space-frequency-time domain covariance matrix and the power distribution vector in the one-dimensional spatial beam domain / power distribution matrix in the two-dimensional spatial-frequency beam domain / power distribution tensor in the triple spatial-frequency-time beam domain, the acquisition of statistical channel information is established as an optimization problem. The goal is to minimize the difference between the channel covariance matrix in the spatial domain / space-frequency domain / space-frequency-time domain calculated from the physical channel fingerprint and the covariance matrix in the spatial domain / space-frequency domain / space-frequency-time domain calculated from the power distribution vector in the one-dimensional spatial beam domain / power distribution matrix in the two-dimensional spatial-frequency beam domain / power distribution tensor in the triple spatial-frequency-time beam domain to achieve optimal acquisition of statistical channel information.

6. The method for acquiring large-scale MIMO-OFDM channel information based on physical channel maps according to claim 5, characterized in that, The generation of triple beam domain statistical channel information includes: Physical channel fingerprints are found based on physical channel maps, and the angle-delay domain power distribution function of the current position is reconstructed based on the channel fingerprints. A spatial-frequency-time domain covariance matrix is ​​generated by leveraging the relationship between the spatial-frequency-time domain channel covariance matrix and the angle-delay domain power distribution. The power distribution tensor of the triple beam domain to be solved is decomposed into a dot product of a real tensor and itself. This real tensor is used as the solution object. Another space-frequency-time domain covariance matrix is ​​generated by using the mapping relationship between the space-frequency-time domain channel and the triple beam domain channel. To measure the difference between two space-frequency-time domain covariance matrices, the two space-frequency-time domain covariance matrices are transformed using a rudder matrix in the space-frequency-time domain to obtain two tensors of the same dimension. An optimization problem is then established, with the goal of minimizing the Kullback-Leibler divergence between the two tensors of the same dimension. The optimization problem is solved using the gradient descent method to obtain the solution for the real tensor. The solution of the real tensor is multiplied by itself to obtain the tensor, which is the desired triple beam domain statistical channel information.

7. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is loaded into the processor, it implements the steps of the method according to any one of claims 1-6.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-6.

Images