A clustering-based channel estimation method, system, device, medium and terminal

Abstract

The present application belongs to but is not limited to the field of communication technology, and particularly relates to a clustering-based channel estimation method, system, device, medium and terminal, comprising: S1, transforming a target signal to a beam domain; S2, performing clustering processing on the obtained beam domain signal; S3, performing inter-cluster screening on the obtained clusters; S4, performing intra-cluster screening on the screened clusters; and S5, inversely reconstructing a channel according to the screened beams. The clustering-based channel estimation takes a cluster as a starting point, performs spatial beam domain filtering on the channel, can effectively filter out the influence of "false" peak points in practice, thereby enhancing channel estimation and avoiding performance loss.

Description

TECHNICAL FIELD

[0001] The present application belongs to but is not limited to the field of communication technology, and particularly relates to a clustering-based channel estimation method, system, device, medium and terminal. BACKGROUND

[0002] The effective operation of a wireless communication system, such as signal detection, precoding, wireless resource allocation, etc., depends on the acquisition of channel information. Therefore, the acquisition of channel state information (CSI) is an important content of the 3GPP standard protocol. From the 4G LTE standard (R8) released in 2009 to the 5G NR standard (R15) released in 2017, 3GPP has been continuously improving and enhancing the ability to acquire channel information. The quality of channel estimation and channel information acquisition directly determines the overall performance of the wireless communication system and is the key to the success of the wireless communication system.

[0003] The large-scale antenna array technology can greatly improve data throughput and spectral efficiency by equipping more antennas, and is widely used in various wireless communication systems. If each antenna is equipped with an independent radio frequency unit, the cost of the wireless communication system will be too high. Therefore, a simple, effective and low-cost channel estimation technology is particularly important for large-scale antenna array systems. The existing large-scale antenna array system fully utilizes the advantage of spatial sparsity to transform the received time-frequency domain channel to a more sparse beam domain. Then, the peak beams are selected in order of energy (amplitude) until the sum energy of the selected beams exceeds a preset threshold such as 85%. Next, the selected beams are retained and the beams at other positions are set to zero. Finally, the beam-filtered channel is inversely transformed in the beam domain to reconstruct and recover to the time-frequency domain to obtain the channel estimation result.

[0004] In view of the above analysis, the existing technical problems of the prior art are that the existing wireless communication system will greatly reduce the accuracy of channel estimation if the peak point selection is incorrect. SUMMARY

[0005] In view of the problems existing in the prior art, the present application provides a clustering-based channel estimation method, system, device, medium and terminal.

[0006] The present application is implemented as follows: a clustering-based channel estimation method, comprising:

[0007] S1, transforming a target signal to a beam domain;

[0008] S2, performing clustering processing on the obtained beam domain signal;

[0009] S3, performing inter-cluster screening on the obtained clusters.

[0010] S4, intra-cluster screening is performed on the screened clusters;

[0011] S5, the channel is reconstructed in reverse according to the screened beams.

[0012] Further, in S1, the target signal is a time-frequency domain signal or a time-frequency domain channel signal after preliminary estimation, and the preliminary estimated channel is: Maximal Ratio Combining (MRC), Least Squares (LS) channel estimation, and Least Minimum Mean Square Error (LMMSE) channel estimation.

[0013] Further, in S2, the obtained beam domain signal is subjected to clustering processing, and the parameters for clustering include but are not limited to: beam delay, beam angle of arrival, and beam energy. According to actual conditions, the specific parameters for clustering can be one or a combination of the listed reference parameters.

[0014] Further, the obtained beam clusters are screened to remove clusters with energy less than noise energy.

[0015] Further, the criteria for inter-cluster screening include:

[0016] ① According to the threshold to select the clusters to be retained. The clusters are arranged in descending order of energy, and clusters with a total energy greater than a predetermined threshold are selected, such as clusters with a total energy greater than 85% of the total energy.

[0017] ② According to the segmentation threshold to select the clusters to be retained. The received channel signal or the channel signal after preliminary estimation is segmented according to the energy size, and the clusters are selected according to the preset threshold for the corresponding segment.

[0018] ③ According to the adaptive threshold to select the clusters. As an embodiment, the adaptive threshold can be the total number of beams divided by the total number of clusters, multiplied by the noise energy.

[0019] ④ According to artificial intelligence to select the clusters. As an embodiment, the ability of artificial intelligence neural network to recognize features can be used to learn the mapping relationship between the real clusters and the received channel signal or the preliminary estimated channel, and the clusters are screened through the mapping relationship.

[0020] ⑤ According to the delay to select the clusters. As an embodiment, the influence of the delay on the interference can be used for screening to remove clusters with excessive energy and delay exceeding the designed communication range.

[0021] ⑥According to the beam direction, the cluster is selected. As an embodiment, the influence of interference can be screened by means of the beam direction, and the cluster whose incoming wave direction is inconsistent with the beam direction of the communicated base station is removed.

[0022] ⑦According to the time delay and the direction, the cluster is selected. As an embodiment, the cluster whose time delay exceeds the designed communication range, whose energy is too large, and whose incoming wave direction is inconsistent with the beam direction of the communicated base station is removed.

[0023] The selected beam cluster is selected in the cluster, and the peak value beam of each cluster is selected according to the energy size of each beam in the cluster.

[0024] Further, the selected beam position is retained, the beams at other positions are set to zero, and the selected beam domain channel is inversely transformed to the time-frequency domain, so that the clustered enhanced channel estimation result is obtained.

[0025] Another object of the present application is to provide a clustered based channel estimation system for realizing the clustered based channel estimation method, comprising:

[0026] The target signal transformation module is configured to transform the target signal to the beam domain.

[0027] The clustered processing module is configured to perform clustered processing on the obtained beam domain signal.

[0028] The inter-cluster screening module is configured to screen the obtained clusters.

[0029] The intra-cluster screening module is configured to screen the selected clusters.

[0030] The inverse reconstruction module is configured to inversely reconstruct the channel according to the selected beams.

[0031] Another object of the present application is to provide a computer device, which comprises a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the clustered based channel estimation method and steps.

[0032] Another object of the present application is to provide a computer readable storage medium, which stores a computer program, and the computer program is executed by the processor to make the processor execute the clustered based channel estimation method and steps.

[0033] Another object of the present application is to provide an information data processing terminal, which comprises the clustered based channel estimation system.

[0034] In combination with the above technical solutions and the technical problems solved, the technical solutions to be protected by the present application have the following advantages and positive effects:

[0035] Firstly, the present application is based on cluster-based channel estimation, taking clusters as the breakthrough point, and performing spatial beam domain filtering on the channel, which can effectively filter out the influence of "false" peak points in practice, thereby enhancing channel estimation and avoiding performance loss.

[0036] The technical solutions of the present application mainly solve the following technical problems in the prior art and have made significant technical progress:

[0037] 1. Technical problems in the prior art

[0038] 1) Low channel estimation accuracy: In a multipath propagation environment, traditional beam domain channel estimation methods are often affected by signal noise and interference. The wrong selection of "false" peak points for channel reconstruction leads to low channel estimation accuracy. Existing methods usually cannot effectively distinguish between dominant paths and secondary paths, which easily leads to estimation errors and affects the overall performance of the communication system.

[0039] 2) Lack of robustness in complex environments: In complex wireless communication environments, signal propagation paths are complex and variable. Existing technologies often cannot maintain stable estimation performance in the face of multipath interference, signal fading, and other situations, resulting in insufficient system reliability.

[0040] 3) High computational complexity: Traditional channel estimation methods have high computational complexity when dealing with large-scale antenna arrays or complex scenarios, resulting in low processing efficiency and difficulty in meeting real-time processing requirements in practical applications. This high computational complexity not only increases the hardware requirements of the system but also prolongs the processing time, affecting the real-time performance of the system.

[0041] 2. Significant technical progress of the present application

[0042] 1) Improve channel estimation accuracy: By transforming the target signal to the beam domain and using clustering processing and multi-layer screening, the present application can effectively distinguish between dominant paths and noise interference, improving the accuracy of channel estimation. In particular, through inter-cluster screening and intra-cluster screening, the system can accurately extract useful signals in a multipath environment, reducing estimation errors.

[0043] 2) Enhance system robustness: The present application uses clustering processing technology to better adapt to complex scenarios such as multipath propagation and signal fading in complex wireless environments, significantly improving the robustness and stability of the system. This robustness enables the system to maintain high channel estimation performance in different propagation environments, thereby improving the overall reliability of the communication system.

[0044] 3) Reducing the computational complexity: By clustering technology to break down the channel estimation problem into multiple small-scale problems, the invention effectively reduces the computational complexity. The cluster screening reduces unnecessary computing, enabling the system to significantly improve processing efficiency while ensuring estimation accuracy, meeting the real-time processing needs of large-scale antenna arrays and complex scenarios.

[0045] In summary, the invention has made significant technical progress in solving the problems of channel estimation accuracy, robustness and computational complexity, providing a more efficient and reliable channel estimation method for modern wireless communication systems.

[0046] Second, the expected income and business value of the technical solution of the invention after transformation are: the channel estimation based on clustering can effectively filter out the influence of "false" peak points, improve the accuracy of channel estimation of wireless communication systems, especially in low signal-to-noise ratio scenarios. The source of "false" peak points usually has three forms: one is the random superposition of electromagnetic wave reflection, refraction and diffraction, the second is random noise, and the third is random interference. 3GPP commercial wireless communication systems generally use paid frequency bands, and the influence of "false" peak points caused by random interference is smaller, mainly the influence of random superposition of multipath effect and random noise. For 3GPP commercial wireless communication, better channel estimation performance in low signal-to-noise ratio means that the coverage of the base station is expanded, and the base station does not need to be deployed too densely. Therefore, the channel estimation technology based on clustering proposed in the invention can greatly save the cost of implementing 3GPP wireless communication systems. For wireless communication systems using public frequency bands, such as WiFi and Bluetooth, in addition to multipath effect and random noise, they are always faced with the influence of "false" peak points caused by random interference. The channel estimation based on clustering proposed in the invention can greatly improve the performance of these wireless communication systems using public frequency bands, and therefore has very high commercial value. In addition to wireless communication systems, wired communication systems are also affected by random crosstalk, and the channel estimation based on clustering proposed in the invention also has high potential commercial value in wired communication systems.

[0047] The technical solution of the invention fills the domestic and foreign industry technical blank: the existing beam domain channel estimation is directly selected for the beam, without considering the influence of "false" peak points caused by random superposition, random noise, random interference and other factors due to electromagnetic wave reflection, refraction and diffraction in the actual system. Once there is a "false" peak, the existing beam domain channel estimation often considers it as a real beam, and reconstructs the channel according to the position of the beam, resulting in a loss of channel estimation performance. The channel estimation based on clustering proposed in the invention effectively avoids the influence of "false" peak points while inheriting the advantages of existing beam domain channel estimation, which is the first in the domestic and foreign, filling the blank of channel estimation technology.

[0048] The technical scheme of the present application solves the technical problem that people have long been eager to solve but have failed to obtain success:

[0049] Wireless communication systems using public frequency bands, such as WiFi, have always been affected by co-frequency random interference, making the performance of these wireless communication systems face great uncertainty, and thus it is difficult to deploy them to industrial scenarios with high requirements for certainty and stability. The channel estimation based on clustering proposed in the present application can well filter out the influence of random interference, making the performance of wireless communication systems using public frequency bands more certain and robust, thereby bringing a lower-cost solution to many commercial scenarios such as industrial scenarios.

[0050] The technical scheme of the present application overcomes the technical prejudice: for a long time, because of the existence of random interference, people have a stereotyped impression that the performance of wireless communication systems using public frequency bands is uncertain, and they are rarely used in scenarios with high requirements for certainty, such as industrial production. The channel estimation based on clustering proposed in the present application well solves the influence of "false" peak points caused by random interference, providing a solution for the use of wireless communication systems using public frequency bands in scenarios with high requirements for certainty, and overcoming the traditional stereotype. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 is a flow chart of the channel estimation method based on clustering provided by the embodiment of the present application;

[0052] Figure 2 is a system structure diagram of the channel estimation based on clustering provided by the embodiment of the present application;

[0053] Figure 3 is a schematic diagram of the enhanced high-precision channel estimation based on clustering provided by the embodiment of the present application;

[0054] Figure 4 is a schematic diagram of the 3GPP UMi channel model, 64 antenna ULA, user speed 3KM / H provided by the embodiment of the present application;

[0055] Figure 5 is a schematic diagram of the 3GPP UMi channel model, 64 antenna ULA, user speed 60KM / H provided by the embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical scheme and advantages of the present application clearer and more apparent, the present application will be further described in detail below in combination with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0057] For a multiple-input and multiple-output (MIMO) wireless communication system equipped with multiple antenna units, assume that the number of antenna units equipped at the base station side is M, the number of antenna units equipped at the user equipment side is N, and the number of transmission streams is S (S≤N). The uplink channel matrix can be expressed as H∈C MXS , where C represents a complex number field. In the training phase, a pre-designed reference signal x t ∈C SX1 is sent by the user side at time t, and the observed signal y t received at the base station side is expressed as

[0058] y t = H t x t + n (1)

[0059] , where n∈C MX1 represents noise. Since the mobile communication device of the user side is powered by a battery, the power consumption and computing power that it can provide are limited, and therefore, in practice, there is usually N=S=1. Since the reference signal x t is known, the least squares (LS) channel estimation can be expressed as

[0060]

[0061] , where H LS is the least squares estimation of the original channel signal H t . The reference signal x t usually satisfies orthogonality, and its conjugate transpose is often used instead. Since the least squares channel estimation is simple to implement, it has been widely used in the industry. However, its disadvantage is also obvious, since the noise term is not considered, it can greatly amplify the noise when the channel condition is poor, resulting in a significant decline in channel estimation performance. Therefore, the linear least mean square error (LMMSE) channel estimation is often used to improve the least squares (LS) channel estimation,

[0062]

[0063] , where R HH =E{HH H} is the autocorrelation matrix of the channel signal, β is a constant determined by the transmission signal constellation point (for example, β=17 / 9 for 16-QAM), and SNR represents the signal-to-noise ratio. The LMMSE-based method can improve the channel estimation performance, but its implementation complexity is high. In particular, it assumes that the channel autocorrelation matrix RHH It is known that this is not true in practice. Generally, R is approximated by long time collection and statistics of the estimated channel HH This makes the acquisition of the required channel statistics R HH for LMMSE very difficult and inefficient in practice. In order to reduce the computational and implementation complexity of LMMSE channel estimation, researchers often exploit the time-domain sparsity of wireless signals to approximate the implementation of LMMSE channel estimation. For example, only the first L (1 < L < N) elements of g are considered, and in its cross-correlation matrix R gg only the first L elements are retained, and other elements are set to zero, thereby reducing the computational complexity. The time-domain sparsity of the signal is also reflected in the eigenvalue space. R hh SVD singular value decomposition is performed on R hh = UAU H , only the eigenvalue vectors of rank p are retained, thereby obtaining a low-rank channel estimation

[0064]

[0065] where Δ p is a diagonal matrix, the first p diagonal elements δ k = λ k / (λ k + β / SNR) (λ k is the eigenvalue of R hh ), and the elements at other positions are zero. The low-rank channel estimation can approximate the theoretical optimal performance, but it still requires a singular value decomposition operation with high computational complexity. On the other hand, both LMMSE and low-rank channel estimation are based on the known channel statistics R hh In practice, the acquisition of channel statistics R hh is very difficult and inefficient, and long-time channel statistics are required to obtain a sufficiently accurate approximation of R hh , making the implementation complexity and cost of LMMSE and its low-rank channel estimation type algorithms still very high.

[0066] With further mining of the signal sparsity, Orthogonal Matching Pursuit (OMP) and Basis Pursuit (BP) get rid of the dependence on the channel statistical information. Thus, the traditional channel estimation problem is transformed into the problem of sparse recovery and reconstruction. The traditional time-domain channel estimation only collects the energy delay profile (PDP) of the channel, and estimates the time delay corresponding to the multipath peak value from the obtained PDP, and then reconstructs the channel according to the position corresponding to the main energy delay. With the increase of the number of base station antennas, array signal processing has been used. Taking a linear antenna array (ULA) equipped with M antenna units as an example, the channel model between the ULA and a single antenna user can be expressed as

[0067]

[0068] where a0u(θ0) is the line-of-sight (LOS) component, a0is the path gain, and the components corresponding to 1≤p≤P are non-line-of-sight (NLOS) path components, P is the total number of multipaths, θ p is the angle of arrival (AoA) of path p, u(θ p ) is usually referred to as the steering vector of the antenna array, and its elements can be expressed as m e [q-(M-1) / 2, q = 0, 1,..., M-1], λ is wavelength, d is antenna spacing, usually satisfies d = λ / 2. In typical 3GPP environment, the angle spread is relatively limited, usually between 2°-5°, similar to time domain sparsity, wireless signal also shows spatial sparsity. Thus, the channel estimation problem is transformed to the spatial direction estimation of the main path. The estimation of the angle of arrival is a classic problem of array signal processing, such as MUSIC and ESPRIT algorithms. These classic array signal processing algorithms, originally designed for radar perception, are based on the assumption of blind estimation, unlike commercial wireless communication systems which have pre-designed training sequences, and the complexity of these algorithms is high, making it difficult to apply to commercial wireless communication systems. Similarly, researchers have proposed a series of spatial low-rank channel estimation algorithms, the basic idea is to transform the received signal to the beam domain with the help of fast Fourier transform, find the beam direction containing the main energy, zero the beams in other directions, and then transform the zeroed channel back through inverse Fourier transform, thereby achieving fast estimation and reconstruction of the wireless channel. These beam domain low-rank channel estimation algorithms take advantage of the spatial sparsity of wireless signals to reduce the dimensionality of the originally large channel, reducing the demand for radio frequency resources of large-scale antenna array systems, while also effectively reducing the data pressure between the radio frequency end and the digital end of the edge antenna Fronthaul.

[0069] The existing beam domain low-rank channel estimation all involves a key operation: the selection of the beam energy peak point. The selection criterion is through the way of the Karman threshold, generally arranging the beams in descending order of energy, and selecting all beams whose beam energy sum is greater than a certain threshold such as 85%. Then, the other beams are zeroed, and the channel is reconstructed in reverse through the selected peak points. If the peak point selection is wrong, it will greatly reduce the accuracy of channel estimation. Taking the indoor wireless environment as an example, various furniture, production equipment, materials and other scatterers are placed in a limited space, and electromagnetic waves are reflected, refracted, diffracted and other by numerous scatterers, which may be superimposed in the positive direction, thereby forming a peak value at the receiving end. For channel estimation, this is a "false" peak point, which will cause serious channel estimation distortion. Furthermore, the wireless communication system represented by WiFi uses a public frequency band, which is easily disturbed by other electronic devices operating on the public frequency band, which manifests as a burst random interference, which will also cause "false" peak points, thereby affecting the quality of channel estimation. In addition, random noise of communication systems everywhere will also be the source of "false" peak points.

[0070] As shown in Figure 1 , a channel estimation method based on clustering includes:

[0071] S1, transforming the target signal to the beam domain;

[0072] S2, performing clustering on the obtained beam domain signals;

[0073] S3, performing inter-cluster screening on the obtained clusters;

[0074] S4, performing intra-cluster screening on the screened clusters;

[0075] S5, performing reverse reconstruction of the channel based on the screened beams.

[0076] Step S1:

[0077] Firstly, the target signal is transformed into the beam domain. The beam domain is formed by angular decomposition of the spatial signal, resulting in multiple beams in different directions, so that the signal energy is concentrated in a few directions. This step is achieved by transforming the received time domain signal or frequency domain signal through a beamforming matrix to obtain the beam domain signal. This conversion helps to highlight the main propagation path of the signal, thereby simplifying the subsequent processing steps.

[0078] Step S2:

[0079] Next, the obtained beam domain signal is subjected to clustering processing. In the beam domain, the signal energy is usually concentrated in a few beams, which represent the main propagation path of the signal. The purpose of clustering processing is to gather these high-energy beams together to form multiple clusters. Each cluster represents a main propagation path or reflection path. Clustering processing can be achieved by setting an energy threshold or clustering algorithm, and beams with similar direction or energy level are allocated to the same cluster.

[0080] Step S3:

[0081] Then, inter-cluster screening is performed on the obtained clusters. The purpose of inter-cluster screening is to screen out the clusters that contribute most to channel estimation among all clusters. By analyzing the energy, position or other statistical characteristics of each cluster, the cluster with the largest channel gain is screened out, and those clusters with smaller contribution to channel estimation are removed. This step can significantly reduce the computational complexity while retaining the most important signal components in channel estimation.

[0082] Steps S4 and S5:

[0083] Finally, intra-cluster screening is performed on the screened clusters, and the channel is reconstructed in reverse based on the screened beams. Intra-cluster screening further refines the beams within each cluster, eliminating those beams that may be affected by noise or multipath effects. Through intra-cluster screening, the clearest signal path in each cluster can be extracted, reducing errors in channel estimation. Then, using the screened beam information, the channel is reconstructed in reverse to restore the time domain or frequency domain characteristics of the channel, thereby completing the entire process of channel estimation. Through this clustering-based processing method, high-precision channel estimation results can be obtained while reducing complexity.

[0084] As shown in Figure 2 , the clustering-based channel estimation system includes:

[0085] a target signal transformation module for transforming the target signal into a beam domain;

[0086] a clustering processing module for performing clustering processing on the obtained beam domain signal;

[0087] an inter-cluster screening module for performing inter-cluster screening on the obtained clusters;

[0088] an intra-cluster screening module for performing intra-cluster screening on the screened clusters;

[0089] a reverse reconstruction module for performing reverse reconstruction of the channel according to the screened beams.

[0090] The clustering-based channel estimation method mainly completes the processing and estimation of the signal through a series of modules. First, in the target signal transformation module, the system receives the target signal and converts it into a beam domain. This step is achieved by performing a specific mathematical transformation (such as Fourier transform or discrete cosine transform) on the signal, which aims to convert the signal from time domain or frequency domain to beam domain, making the channel matrix more sparse, so as to more effectively extract the directional information in the signal, thereby improving the accuracy of subsequent processing.

[0091] Next, the clustering processing module performs clustering processing on the signal converted into the beam domain. The purpose of clustering processing is to gather beams with similar characteristics (such as spatial direction and channel condition) together to form multiple clusters. Each cluster represents a group of similar propagation paths in the signal space. Through this clustering processing, the system can effectively reduce noise interference and improve the accuracy of channel estimation.

[0092] Subsequently, the system enters the operation of the inter-cluster screening module and the intra-cluster screening module. The inter-cluster screening module analyzes the characteristics of the clustered signal and screens out the most representative clusters in the global range. These clusters usually correspond to the dominant propagation paths of the signal. The screened clusters then enter the intra-cluster screening module, which further refines and screens the signals within each cluster, removing noise and other interference signals, thereby retaining the most beneficial signal components for channel estimation.

[0093] Finally, the system uses the screened beam signals for reverse reconstruction of the channel through the reverse reconstruction module. Reverse reconstruction is the reconstruction of the processed beam signal back to the time domain or frequency domain representation of the channel. Through the reconstruction of these screened beam signals, the system can accurately estimate the channel characteristics of the target signal. This clustering-based channel estimation method can significantly improve the accuracy and reliability of channel estimation, especially in complex multipath propagation environments.

[0094] To further improve the robustness and accuracy of channel estimation, the application proposes a high-precision channel estimation enhancement scheme based on clustering. Specifically, by investigating the time delay, angle of arrival, amplitude, etc. of each point of the multipath channel, the application clusters each point, finds the cluster with the maximum energy in each iteration, and then obtains the peak point of each cluster. This can effectively filter out "false" peak points, thereby improving the accuracy and robustness of channel estimation. The principle of the application is shown in Figure 3 .

[0095] The implementation steps of the application are as follows:

[0096] Step 1: Beam domain transformation. Transform the received time-frequency domain signal y t as shown in equation (1) or the preliminary estimated time-frequency domain channel signal to the beam domain. Further, the preliminary estimated channel can be but is not limited to:

[0097] ① Maximal Ratio Combining (MRC).

[0098] ② Least Squares (LS) channel estimation.

[0099] ③ Least Minimum Mean Square Error (LMMSE) channel estimation.

[0100] Step 2: Beam clustering. Cluster the beam domain channel, and the reference parameters for clustering include but are not limited to: beam time delay, beam angle of arrival, and beam energy. According to actual conditions, the specific clustering parameters can be one or a combination of the listed reference parameters.

[0101] Step 3: Cluster selection. Select the obtained beam clusters and remove clusters with energy less than the noise energy.

[0102] Further, the remaining beam clusters are subjected to a second selection, and the selection criteria include but are not limited to:

[0103] ① According to the threshold, select the clusters that should be retained. Arrange the clusters in descending order of energy, and select clusters with a total energy greater than a certain threshold, such as selecting clusters with an energy proportion greater than 85% of the total energy.

[0104] ② According to the segmentation threshold, select the clusters that should be retained. According to the energy of the received channel signal or the preliminary estimated channel signal, divide the threshold into segments. For the corresponding segments, select the clusters according to the preset threshold.

[0105] 3. Selecting clusters according to an adaptive threshold. As an embodiment, the adaptive threshold can be the total beam energy divided by the total number of clusters, multiplied by the noise energy.

[0106] 4. Selecting clusters according to artificial intelligence. As an embodiment, the mapping relationship between the real clusters and the received channel signals or the preliminary estimated channel can be learned by the ability of artificial intelligence neural network to identify features, and the clusters are filtered through the mapping relationship.

[0107] 5. Selecting clusters according to time delay. As an embodiment, the impact of interference can be filtered by time delay, and clusters with time delay exceeding the designed communication range and excessive energy are removed.

[0108] 6. Selecting clusters according to beam direction. As an embodiment, the impact of interference can be filtered by beam direction, and clusters with inconsistent incoming wave direction and beam direction of the communication base station are removed.

[0109] 7. Selecting clusters according to the combination of time delay and direction. As an embodiment, clusters with time delay exceeding the designed communication range, excessive energy and inconsistent incoming wave direction and beam direction of the communication base station are removed.

[0110] Step 4, intra-cluster filtering. The selected beam clusters are selected within the cluster, and the peak beam of each cluster is selected according to the energy size of each beam within the cluster.

[0111] Step 5, channel reconstruction. The selected beam positions are retained, the beams at other positions are set to zero, and the filtered beam domain channel is inversely transformed to the time-frequency domain, and the clustered enhanced channel estimation result is obtained.

[0112] The clustered channel estimation takes the cluster as the starting point, filters the spatial beam domain of the channel, effectively filters out the influence of the "false" peak point in practice, thereby enhancing the channel estimation and avoiding performance loss.

[0113] The present application establishes a channel model for simulation according to 3GPP TR 38.901, the frequency band used is 5GHz, the antenna array is a uniform linear array ULA (Unitary Linear Array), the number of antenna elements is 64, and the channel scene is UMi (Urban Micro). Figure 4 and Figure 5 As shown in the drawings, the channel estimation based on cluster selection (labeled as ClusterSel in the drawings) achieves more robust performance than the traditional beam domain channel estimation based on threshold selection (labeled as BmSelThrd in the drawings). Especially in the low signal-to-noise ratio scene, the random fluctuations of noise are more likely to cause "false" peaks.

[0114] Figures 4 to 5 The simulation of the present application only simulates the influence of urban multipath and noise on channel estimation, and does not add the influence of interference. In an actual system, random interference is an important source of "false" peaks.

[0115] It is worth noting that the present application transforms the time-frequency domain channel into the beam domain in order to obtain a more sparse channel, more accurately extract the direction information of the signal, and improve the accuracy of channel estimation. This step of beam domain transformation is not necessary, and the clustering-based channel estimation proposed in the present application can also be directly applied to time-frequency domain channel estimation.

[0116] The application embodiment of the present application provides a computer device, which comprises a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of the clustering-based channel estimation method.

[0117] The application embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of the clustering-based channel estimation method.

[0118] The application embodiment of the present application provides an information data processing terminal, which comprises a clustering-based channel estimation system.

[0119] The "false" peak point is generally caused by the random superposition of electromagnetic wave multipath effect, random noise, and random interference. The abnormal beam cluster formed by the "false" peak point has a large difference from the real beam cluster in terms of beam delay, beam direction, and beam energy, etc. Therefore, the clustering-based channel estimation proposed in the present application can well distinguish the real cluster and the abnormal beam cluster by clustering processing the channel signal according to the beam delay, the beam direction, and the beam energy, etc. Figure 4 and Figure 5 As shown in FIGS. 1 and 2, the present application applies the clustering-based channel estimation and the traditional threshold-based beam domain channel estimation to the received channel signal respectively, and the clustering-based channel estimation achieves more robust performance, especially in a low signal-to-noise ratio scenario.

[0120] It should be noted that embodiments of the present application can be realized by hardware, software, or a combination of software and hardware. The hardware portion can be realized by a special logic; the software portion can be stored in a memory and executed by a proper instruction execution system, such as a microprocessor or a specially designed hardware. A person of ordinary skill in the art can understand that the above-mentioned apparatus and method can be realized by computer executable instructions and / or included in processor control codes, such as a carrier medium, such as a magnetic disk, CD or DVD-ROM, a programmable memory, such as a read-only memory (firmware), or a data carrier, such as an optical or electronic signal carrier. The apparatus of the present application and its modules can be realized by a hardware circuit, such as a very large scale integrated circuit or a gate array, a semiconductor, such as a logic chip, a transistor, or a programmable hardware device, such as a field programmable gate array, a programmable logic device, or the like, by software executed by various types of processors, or by a combination of the above-mentioned hardware circuit and software, such as firmware.

[0121] The above description is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any modification, equivalent replacement, and improvement within the technical range disclosed by the present application, and within the spirit and principle of the present application, should be covered within the protection scope of the present application.

Claims

1. A clustering-based channel estimation method, characterized in that, Comprising: S1, transforming the target signal into the beam domain; S2, performing clustering processing on the obtained beam domain signal; S3, performing inter-cluster screening on the obtained clusters; S4, performing intra-cluster screening on the screened clusters; S5, performing reverse reconstruction on the channel according to the screened beams. The parameters referred to in the S2 clustering include one or more combinations of beam delay, beam angle of arrival, and beam energy. In the S3, the obtained beam clusters are screened to remove clusters with energy less than noise energy. The remaining beam clusters are subjected to secondary screening, and the screening criteria include: selecting clusters that should be retained according to a threshold; selecting clusters according to artificial intelligence, using the ability of artificial intelligence neural network to identify features, learning the mapping relationship between the real clusters and the received channel signals or the preliminary estimated channel, and screening the clusters through the mapping relationship; selecting clusters according to the joint of delay and direction, removing clusters with delay exceeding the designed communication range, energy being too large, and incoming direction being inconsistent with the beam direction of the communication base station. In the S4, the selected beam clusters are subjected to intra-cluster selection, and the peak beam of each cluster is selected according to the energy of each beam in the cluster. In the S5, the selected beam positions are retained, the beams at other positions are set to zero, and the screened beam domain channel is inversely transformed into the time-frequency domain to obtain a clustered enhanced channel estimation result.

2. The cluster-based channel estimation method of claim 1, wherein, In the S1, the target signal is a time-frequency domain signal or a time-frequency domain channel signal after preliminary estimation, and the channel signal after preliminary estimation is: maximum ratio combining MRC, least squares LS channel estimation, and linear minimum mean square error LMMSE channel estimation.

3. The cluster-based channel estimation method as claimed in claim 1, wherein, The criteria for inter-cluster screening include: Criterion 1: selecting clusters that should be retained according to a threshold, arranging the clusters in descending order of energy, and selecting clusters with energy greater than a predetermined threshold; Criterion 2: selecting clusters that should be retained according to a segmented threshold, dividing the threshold according to the energy of the received channel signal or the channel signal after preliminary estimation, and selecting clusters according to the preset threshold for each segment; Criterion 3: selecting clusters according to an adaptive threshold, the adaptive threshold being the total number of beams divided by the total number of clusters, and then multiplied by the noise energy; Criterion 4: selecting clusters according to artificial intelligence, using the ability of artificial intelligence neural network to identify features, learning the mapping relationship between the real clusters and the received channel signals or the preliminary estimated channel, and screening the clusters through the mapping relationship.

4. A clustering-based channel estimation system for implementing the clustering-based channel estimation method according to any one of claims 1 to 3, characterized in that, Comprising: A target signal transformation module for transforming the target signal into the beam domain; A clustering processing module for performing clustering processing on the obtained beam domain signal; An inter-cluster screening module for performing inter-cluster screening on the obtained clusters; An intra-cluster screening module for performing intra-cluster screening on the screened clusters; A reverse reconstruction module for performing reverse reconstruction on the channel according to the screened beams.

5. A computer device, comprising a memory and a processor, the memory storing a computer program, and the computer program being executed by the processor to make the processor execute the steps of the channel estimation method based on clustering according to any one of claims 1-3.

6. A computer readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the cluster-based channel estimation method according to any one of claims 1-3.

7. An information data processing terminal, the information data processing terminal comprising the cluster-based channel estimation system according to claim 4.

Images