Mass multiple-input multiple-output system channel estimation method based on message passing

Abstract

The embodiment of the invention discloses a large-scale multiple-input-multiple-output system channel estimation method based on message passing. A specific embodiment of the method comprises the following steps: for each antenna in an antenna set included in the large-scale MIMO system, determining a receiving pilot signal corresponding to the antenna; establishing a signal linear model according to each determined receiving pilot signal; according to a singular value decomposition result of the measurement matrix, updating the signal linear model to obtain an updated signal model; determining a transmission probability, sparse feature representation, correlation, random variable distribution and jump probability distribution corresponding to the channel matrix; generating a joint probability density function; determining a factor graph corresponding to the joint probability density function; and generating a channel estimation value based on the mixed belief propagation average field message updating rule and the factor graph. According to the embodiment, the estimation error and the mean square error of the channel in the antenna-time delay domain are reduced.

Description

Technical Field

[0001] The embodiments of this disclosure relate to the field of computer technology, and more specifically to a channel estimation method for large-scale multiple-input multiple-output systems based on message passing. Background Technology

[0002] With the ever-changing application scenarios of 6G communication and the development of information technology, wireless communication is seeking to provide higher and more efficient transmission rates for enhanced mobile internet, higher reliability, accuracy, and integrity for military communications and cloud computing, and lower latency information exchange capabilities for autonomous driving and telemedicine. As a core technology of 6G, Massive MIMO (Multiple-Input Multiple-Output) has become a hot research topic. Compared to traditional MIMO systems, Massive MIMO systems have a greater number of antennas, which significantly improves spatial multiplexing gain. It can serve multiple users simultaneously on the same spectrum resource, significantly improving spectrum utilization. Based on the above discussion, for Massive MIMO systems to achieve better performance gains, they need to acquire efficient and accurate Channel State Information (CSI). Therefore, channel estimation is fundamental to communication system design.

[0003] To address the high sensitivity to noise and insufficient accuracy of traditional channel estimation methods, Minimum Mean Square Error (MMSE) estimation relies on the statistical properties of the channel and prior knowledge, offering strong interpretability and extensive mathematical derivation. This allows it to maintain high estimation accuracy even in noisy environments. Channel estimation in large-scale MIMO systems is essentially a statistical inference problem, estimating unknown variables using noisy observations and prior information about the variables. The core idea of ​​the message-passing method is to use probabilistic graphical models, such as Bayesian networks or Markov random fields, to represent the posterior distribution of the channel estimation problem. It updates the posterior probabilities of the variables by iteratively exchanging messages, which are passed between nodes in the graphical model until convergence to a stable state. Therefore, when the channel matrix has a sparse structure, this method can effectively handle high-dimensional and complex channel estimation problems.

[0004] In existing technologies, the HB-VBI channel estimation method based on Bayesian inference employs the 3GPP spatial channel model, models the channel as a Gaussian prior model, and uses the Mean Field (MF) message update rule to obtain the expected value of the estimated channel. The shortcomings of this method are: the MF message update rule is highly sensitive to priors and likelihoods, leading to difficulties in parameter tuning; secondly, the number of iterations is usually large, the convergence speed is significantly affected by the initial values, and the estimation error increases when the correlation between parameters is high.

[0005] Furthermore, channel estimation methods employing Sparse Bayesian Learning (SBL) and Pattern Coupled Sparse Bayesian Learning (PCSBL) only model the prior as a Gaussian prior model, assuming the parameters follow a gamma distribution. However, SBL requires storing the posterior covariance matrix, resulting in excessive memory consumption and computational burden as the number of antenna arrays increases. PCSBL utilizes adjacent coefficients to couple the prior, achieving higher estimation accuracy than traditional SBL, but it requires precise adjustment of coupling hyperparameters and thresholds. When the coupling parameters are mismatched, the estimation accuracy drops significantly.

[0006] The information disclosed in this background section is only intended to enhance the understanding of the background of the inventive concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0008] Some embodiments of this disclosure propose a message-passing-based channel estimation method for large-scale multiple-input multiple-output systems to address the technical problems mentioned in the background section above.

[0009] In a first aspect, some embodiments of this disclosure provide a channel estimation method for a large-scale multiple-input multiple-output (MLMI) system based on message passing. The method includes: for each antenna in an antenna set included in the MMI, determining the received pilot signal corresponding to the antenna, wherein the received pilot signal can be a combination of a measurement matrix, a composite channel impulse response, and additive white Gaussian noise; establishing a linear signal model based on the determined received pilot signals, wherein the linear signal model is a linear model composed of a received pilot signal matrix, a measurement matrix, a channel matrix, and a white Gaussian noise matrix; updating the linear signal model based on the singular value decomposition result of the measurement matrix to obtain an updated signal model, wherein the updated signal model includes the channel matrix; determining the transmission probability, sparse feature representation, correlation, random variable distribution, and jump probability distribution corresponding to the channel matrix based on a three-layer Bayesian inference technique; generating a joint probability density function based on the transmission probability, the sparse feature representation, the correlation, the random variable distribution, and the jump probability distribution; determining the factor graph corresponding to the joint probability density function; and generating a channel estimate based on a mixed belief propagation mean-field message update rule and the factor graph.

[0010] Secondly, some embodiments of this disclosure provide a channel estimation apparatus for a large-scale multiple-input multiple-output system based on message passing. The apparatus includes: a first determining unit configured to determine a received pilot signal corresponding to each antenna in an antenna set included in the large-scale multiple-input multiple-output system, wherein the received pilot signal can be a combination of a measurement matrix, a composite channel impulse response, and additive white Gaussian noise; an establishing unit configured to establish a signal linear model based on each determined received pilot signal, wherein the signal linear model is a linear model composed of a received pilot signal matrix, a measurement matrix, a channel matrix, and a white Gaussian noise matrix; and an updating unit configured to update the measurement matrix based on the singular value decomposition of the measurement matrix. As a result, the above-mentioned linear signal model is updated to obtain an updated signal model, wherein the updated signal model includes the above-mentioned channel matrix; the second determining unit is configured to determine the transmission probability, sparse feature representation, correlation, random variable distribution, and jump probability distribution corresponding to the above-mentioned channel matrix based on three-layer Bayesian inference technology; the first generating unit is configured to generate a joint probability density function based on the above-mentioned transmission probability, the above-mentioned sparse feature representation, the above-mentioned correlation, the above-mentioned random variable distribution, and the above-mentioned jump probability distribution; the third determining unit is configured to determine the factor graph corresponding to the above-mentioned joint probability density function; the second generating unit is configured to generate a channel estimate based on the mixed belief propagation mean field message update rule and the above-mentioned factor graph.

[0011] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0012] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.

[0013] The various embodiments of this disclosure have the following beneficial effects: The message-passing-based channel estimation method for large-scale MIMO systems, based on some embodiments of this disclosure, adopts the TR38.901 channel model in the 3GPP standard to establish an accurate hierarchical prior model for the antenna-delay domain channel. Iterative parameter updates using Hybrid Message Passing (HMP) on the factor graph improve the estimation accuracy of the channel in large-scale MIMO systems. Considering that traditional message-passing algorithms often exhibit ill-conditioned characteristics in the measurement matrix during channel estimation and signal detection, leading to unstable solutions and error amplification, thus affecting the performance of the communication system, this application establishes an accurate prior model for the channel and likelihood, and performs singular value decomposition on the measurement matrix. This aims to improve the channel estimation performance under ill-conditioned measurement matrices, thereby enhancing the robustness and reliability of the communication system. Through the above operations, the estimation error of the antenna-delay domain channel in large-scale MIMO systems is effectively reduced, and a significant performance improvement is ensured. The sparsity of the channel is characterized by numerical features and sparse structure. A Bernoulli-Gaussian prior model is used to represent the numerical characteristics of the channel, and the correlation between antennas is represented by a Markov chain. This effectively improves the performance of large-scale MIMO systems in uplink communication scenarios, making the NMSE closer to the Cramer-Rao bound. Experimental results show that this application, by adopting a Bayesian inference framework to establish an accurate prior probability model of the channel, significantly reduces the estimation error of the channel in the antenna-delay domain. Furthermore, compared to other algorithms, the mean square error is minimized with increasing iteration count and pilot number. Attached Figure Description

[0014] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0015] Figure 1This is a flowchart of some embodiments of the message-passing-based channel estimation method for large-scale multiple-input multiple-output systems according to this disclosure;

[0016] Figure 2 This is a schematic diagram of a large-scale MIMO system according to some embodiments of the message-passing-based channel estimation method for large-scale multiple-input multiple-output systems disclosed herein;

[0017] Figure 3 This is an example diagram of the sparse representation of the channel in the antenna-delay domain according to some embodiments of the message-passing-based channel estimation method for large-scale multiple-input multiple-output systems according to this disclosure;

[0018] Figure 4 This is a factor graph of some embodiments of the message-passing-based channel estimation method for large-scale multiple-input multiple-output systems according to this disclosure;

[0019] Figure 5 These are experimental comparison figures of some embodiments of the message-passing-based channel estimation method for large-scale multiple-input multiple-output systems according to this disclosure;

[0020] Figure 6 This is a schematic diagram of the structure of some embodiments of a message-passing-based channel estimation apparatus for a large-scale multiple-input multiple-output system according to the present disclosure;

[0021] Figure 7 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0022] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0023] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0024] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0025] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0026] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0027] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0028] Figure 1 A flow 100 is shown illustrating some embodiments of a message-passing-based channel estimation method for large-scale multiple-input multiple-output systems according to this disclosure. This message-passing-based channel estimation method for large-scale multiple-input multiple-output systems includes the following steps:

[0029] Step 101: For each antenna in the antenna set included in the large-scale multiple-input multiple-output system, determine the receiving pilot signal corresponding to the antenna.

[0030] In some embodiments, the execution entity (e.g., a computing device) of the message-passing-based channel estimation method for a massive MIMO system can determine the received pilot signal corresponding to each antenna in the antenna set included in the massive MIMO system. The massive MIMO system can be used for uplink transmission between a user and a base station. The user can be a single-antenna user. The base station can use a uniform linear array equipped with multiple antennas to serve a single-antenna user. A schematic diagram of the massive MIMO system is shown below. Figure 2 As shown. Figure 2 In this context, BS stands for Base Station, and uplink transmission refers to uplink transmission. UE stands for User Equipment. The multiple antennas mentioned above can form an antenna set. Each antenna in this antenna set can receive pilot signals from the User Equipment, serving as a receive pilot signal. This receive pilot signal can be a combination of a measurement matrix, a composite channel impulse response, and additive white Gaussian noise.

[0031] In practice, the aforementioned large-scale multiple-input multiple-output system can employ orthogonal matching pursuit (OMP) technology to estimate the channel (composite channel impulse response) state based on the received pilot signal.

[0032] In some optional implementations of certain embodiments, the execution entity determining the received pilot signal corresponding to each antenna in the antenna set included in the large-scale multiple-input multiple-output system may include the following steps:

[0033] The first step, for each antenna in the antenna set included in the large-scale multiple-input multiple-output system, is to perform the following steps:

[0034] The first sub-step involves determining the set of pilot subcarriers for pilot symbol transmission based on Orthogonal Frequency Division Multiplexing (OFDM). OFDM is a multi-carrier modulation technique. In practice, firstly, based on OFDM, the user's transmitted data can be distributed across multiple mutually orthogonal subcarriers, resulting in a subcarrier set. Each subcarrier in this set can carry one data symbol. Secondly, at least two pre-selected subcarriers from this set can be designated as pilot subcarriers, resulting in a pilot subcarrier set. Each pilot subcarrier in this set can be used to transmit pilot symbols. Each pilot subcarrier in this set corresponds to a pilot symbol and a pilot position. The pilot positions can be a pre-selected complex number. The pilot positions corresponding to the above pilot subcarrier set can be represented as follows:

[0035] .

[0036] Where K represents the number of pilot subcarriers in the above pilot subcarrier cluster. i This represents the position of the i-th pilot. The pilot symbols corresponding to the above pilot subcarrier set can be represented as:

[0037] .

[0038] in, Representing each pilot symbol x(p) i A vector composed of ).

[0039] The second sub-step involves generating a pilot symbol matrix based on the pilot symbols corresponding to each pilot subcarrier in the aforementioned pilot subcarrier set. This pilot symbol matrix can be a diagonal matrix composed of the pilot symbols transmitted by each pilot subcarrier in the aforementioned pilot subcarrier set. In practice, the pilot symbol matrix can be generated using the following formula.

[0040] .

[0041] Where X represents the pilot symbol matrix. C represents the dimension, with a total of K rows and K columns.

[0042] The third sub-step involves generating a Discrete Fourier Transform (DFT) matrix based on the pilot position corresponding to each pilot subcarrier in the aforementioned pilot subcarrier set. In practice, the DFT matrix can be generated using the following formula.

[0043] .

[0044] Where F represents the Discrete Fourier Transform matrix. j represents the complex unit. N represents the number of sampling points in the Discrete Fourier Transform. d⁻¹ represents the sampling interval. D represents the channel length. ,in This indicates the preset maximum delay spread. This indicates the preset signal sampling period. This indicates rounding.

[0045] The fourth sub-step involves determining the measurement matrix by multiplying the pilot symbol matrix and the discrete Fourier transform matrix mentioned above.

[0046] The fifth sub-step involves determining the composite channel impulse response (CIR) corresponding to the antenna based on the channel length in the aforementioned large-scale multiple-input multiple-output system. This CIR can be D composite channel impulse responses (CIRs) of the antenna. These CIRs can be independent and identically distributed. The CIR can be a D-dimensional column vector. The elements of this column vector can be complex numbers. The specific values ​​of the elements in the CIR can be determined through the following steps.

[0047] The sixth sub-step involves determining the additive white Gaussian noise corresponding to the aforementioned antenna. This additive white Gaussian noise follows a Gaussian distribution. , Indicates the noise variance. K Let represent the K-order identity matrix. Only the structure of the additive white Gaussian noise described above is determined here; the numerical values ​​corresponding to the additive white Gaussian noise can be estimated using the following steps.

[0048] The seventh sub-step involves adding the product of the aforementioned measurement matrix and the aforementioned composite channel impulse response to the aforementioned additive white Gaussian noise to obtain the received pilot signal. In practice, the received pilot signal can be obtained using the following formula.

[0049] .

[0050] in, This represents the received pilot signal of the m-th antenna. This represents the measurement matrix. This represents the composite channel impulse response of the m-th antenna. Let represent the additive white Gaussian noise of the m-th antenna. The above formula can characterize the signal model of a single antenna.

[0051] Step 102: Establish a linear model of the signal based on the determined received pilot signals.

[0052] In some embodiments, the aforementioned execution entity can establish a signal linear model based on the determined received pilot signals. This signal linear model is a linear model composed of the received pilot signal matrix, the measurement matrix, the channel matrix, and the Gaussian white noise matrix.

[0053] In some optional implementations of certain embodiments, the execution entity establishes a signal linear model based on the determined individual received pilot signals, which may include the following steps:

[0054] The first step is to convert the determined received pilot signals into a received pilot signal matrix. In practice, the vectors corresponding to each received pilot signal can be combined to obtain the received signal matrix. The vector form of the received pilot signal can be defined as:

[0055] .

[0056] The second step is to convert the composite channel impulse responses into a channel matrix. In practice, the vectors corresponding to the composite channel impulse responses can be combined to form a channel matrix.

[0057] The third step is to convert each additive Gaussian white noise into a Gaussian white noise matrix. In practice, the vectors corresponding to each additive Gaussian white noise can be combined to form a Gaussian white noise matrix. The vector form of additive Gaussian noise can be defined as:

[0058] .

[0059] Fourth, based on the received pilot signal matrix, the measurement matrix, the channel matrix, and the Gaussian white noise matrix, a signal linear model is generated. The dependent variable of this signal linear model is the received pilot signal matrix. The received pilot signal matrix is ​​equal to the product of the measurement matrix and the channel matrix plus the Gaussian white noise matrix. The signal linear model can be expressed as:

[0060] .

[0061] Where Y represents the received pilot signal matrix, H represents the channel matrix, and n represents the Gaussian white noise matrix.

[0062] In practice, due to the potential sparsity of the channel in the antenna-delay domain, Only a few non-zero elements exist in it, that is An example diagram of the sparse representation of the channel in the antenna-delay domain is shown below. Figure 3 As shown, it contains both zero and non-zero elements. Therefore, the main objective of this application is to design an efficient channel estimation algorithm based on the received signal matrix Y and the measurement matrix. Reconstruct H.

[0063] Step 103: Update the linear model of the signal based on the singular value decomposition results of the measurement matrix to obtain the updated signal model.

[0064] In some embodiments, the execution entity may update the linear signal model based on the singular value decomposition result of the measurement matrix to obtain an updated signal model. The updated signal model includes the channel matrix.

[0065] In some optional implementations of certain embodiments, the execution entity updates the linear model of the signal based on the singular value decomposition results of the measurement matrix to obtain an updated signal model, which may include the following steps:

[0066] The first step is to determine the first unitary matrix, the second unitary matrix, and the singular value diagonal matrix corresponding to the aforementioned measurement matrix. In practice, the first unitary matrix, the second unitary matrix, and the singular value diagonal matrix corresponding to the aforementioned measurement matrix can be determined using the following formula.

[0067] .

[0068] Where U represents the first unitary matrix and V represents the second unitary matrix. This represents a singular value diagonal matrix.

[0069] In practice, the measurement matrix often exhibits ill-conditioned characteristics during algorithm iteration, resulting in a large condition number and potential convergence issues and decreased estimation accuracy. To address this problem, this application performs singular value decomposition on the aforementioned measurement matrix, making it more stable for channel estimation and effectively suppressing the error amplification effect caused by the ill-conditioned matrix.

[0070] The second step is to determine the conjugate matrix corresponding to the first unitary matrix mentioned above.

[0071] The third step is to multiply the conjugate matrix above with the linear signal model above to obtain the updated signal model. In practice, the updated signal model can be obtained using the following formula.

[0072] .

[0073] in, , , .

[0074] Step 104: Based on the three-layer Bayesian inference technique, determine the transmission probability, sparse feature representation, correlation, random variable distribution and jump probability distribution corresponding to the channel matrix.

[0075] In some embodiments, the aforementioned execution entity may determine the transmission probability, sparse feature representation, correlation, random variable distribution, and jump probability distribution corresponding to the aforementioned channel matrix based on three-layer Bayesian inference technology.

[0076] In some optional implementations of certain embodiments, the execution entity determines the transmission probability, sparse feature representation, correlation, random variable distribution, and jump probability distribution corresponding to the channel matrix based on three-layer Bayesian inference technology, which may include the following steps:

[0077] The first step is to determine the transmission probabilities corresponding to the aforementioned channel matrix based on a Bayesian framework. In practice, the transmission probabilities corresponding to the channel matrix can be determined using the following formula.

[0078] .

[0079] in, This indicates the transmission probability. This represents conditional probability. This represents the conditional probability of the first layer of the Bayesian framework. Indicates precision. Indicates variance.

[0080] The second step is to determine the sparse characteristic representation of the aforementioned channel matrix in the antenna delay domain based on the Bernoulli-Gaussian model. In practice, the sparse characteristic representation of the aforementioned channel matrix in the antenna-delay domain can be determined using the following formula based on the Bernoulli-Gaussian model.

[0081] .

[0082] in, The sparse feature representation is given by S, where S indicates whether the channel is sparse and V indicates the sparse structure of the channel. Both S and V can be D×M dimensional matrices. Used to represent Markov state values, representing the channel It is zero It is still non-zero. The reciprocal of an element in V It can represent the variance of a Gaussian distribution.

[0083] In practice, the above formula can characterize that each element in the channel matrix H is either zero with a certain probability or has a certain probability of having a variance of 0. The complex Gaussian random variable. The above formula conforms to the sparsity characteristics of the channel in the antenna-delay domain.

[0084] The third step, based on a first-order Markov chain, is to determine the correlation between the aforementioned channel matrix and the various antennas in the aforementioned antenna set. In practice, a first-order Markov chain can be used to characterize the correlation between different antennas. Therefore, in the third layer of Bayesian inference, S can be modeled as:

[0085] .

[0086] in, Indicates correlation. This represents the initial probability. This is used to characterize the transition probability of the state change between the m-th antenna and the (m-1)-th antenna. express The probability of jumping from 0 to 1. express The probability of jumping from 1 to 0. Here, , pr() represents probability.

[0087] The fourth step is to generate a jump probability distribution based on the above correlations.

[0088] In practice, the above can be used as a reference. and As a variable, it is modeled as a Beta distribution, as shown in the following formula:

[0089] , .

[0090] in, and This represents the jump probability distribution. , Represents probability Parameters that follow a Beta distribution. , Represents probability Parameters that follow a Beta distribution.

[0091] The fifth step is to generate a random variable distribution based on the above sparse characteristic representation.

[0092] In practice, the above can be used as a reference. The model is based on a Gamma distribution, as shown in the following formula:

[0093] .

[0094] in, This represents the distribution of a random variable. and yes Parameters that follow a Gamma distribution.

[0095] Step 105: Generate a joint probability density function based on the transmission probability, sparse feature representation, correlation, random variable distribution, and jump probability distribution.

[0096] In some embodiments, the executing entity can generate a joint probability density function based on the transmission probability, the sparse feature representation, the correlation, the random variable distribution, and the jump probability distribution. In practice, the joint probability density function can be generated using the following formula.

[0097] .

[0098] in, , , , , , , , Both represent factor nodes.

[0099] Step 106: Determine the factor graph corresponding to the joint probability density function.

[0100] In some embodiments, the aforementioned executing entity can determine the factor graph corresponding to the joint probability density function. In practice, this can be achieved by factoring the joint probability density function, and each factor node obtained can serve as a node in the factor graph. The edges of the factor graph then connect these variables and factors, such as... Figure 4 As shown. Figure 4 In the diagram, circles represent variable nodes, black boxes represent factor nodes, solid lines indicate the use of the Belief Propagation (BP) edge, and dashed lines indicate the use of the Mean Field (MF) edge.

[0101] Step 107: Generate channel estimates based on the hybrid belief propagation mean field message update rule and the factor graph described above.

[0102] In some embodiments, the aforementioned execution entity can generate channel estimates based on the hybrid belief propagation mean-field message update rule and the aforementioned factor graph. In practice, channel estimates can be generated based on the aforementioned factor graph using a preset hybrid belief propagation mean-field message update algorithm. The aforementioned hybrid belief propagation mean-field message update algorithm can be a message update algorithm combining the hybrid belief propagation (BP) algorithm and the mean-field (MF) algorithm.

[0103] Specifically, firstly, the Noncoherent Approximate Message Passing (UAMP) algorithm can be used to process the measurement matrix, thereby improving channel estimation performance and obtaining... The likelihood of the message. As shown in the following formula:

[0104] .

[0105] in, express Passed to The iterative update formula.

[0106] Regarding the message , and Let these represent the expectation and variance, respectively, and continue iterative updates. The final output message can be represented as:

[0107] .

[0108] Initialize confidence Received news Update:

[0109] .

[0110] in, , , .

[0111] Secondly, the numerical features can be updated, including the following steps:

[0112] First, initialize the confidence. and The Markov chain passes messages down:

[0113] .

[0114] in, Indicates the parameter when m=1. The updated value. Specifically, the parameters in the above formula represent:

[0115] ,

[0116] ,

[0117] .

[0118] In the formula, c1, c2, c3, and c4 represent:

[0119] , ,

[0120] , .

[0121] Variable Node To factor node A message update can be represented as:

[0122] ,

[0123] .

[0124] The parameters represent:

[0125] ,

[0126] ,

[0127] .

[0128] Confidence and The update can be represented by the following formula:

[0129] ,

[0130] .

[0131] Wherein, the parameter represents:

[0132] ,

[0133] ,

[0134] ,

[0135] ,

[0136] ,

[0137] .

[0138] Finally, a first-order Markov chain can be used to characterize the sparse structure of the channel and complete message updates and variable nodes. To factor node News update:

[0139] ,

[0140] ,

[0141] .

[0142] in, It indicates a direct proportion.

[0143] Confidence update obtained and ,in They represent:

[0144] , ,

[0145] , .

[0146] Variable Node To factor node News update:

[0147] .

[0148] Wherein, the parameter represents:

[0149] .

[0150] Confidence renew:

[0151] .

[0152] Wherein, the parameter represents:

[0153] ,

[0154] , .

[0155] Factor Node To variable node News update:

[0156] .

[0157] confidence Parameter update:

[0158] ,

[0159] .

[0160] Factor Node To variable node News update:

[0161] .

[0162] Variable Node confidence renew:

[0163] .

[0164] Finally, confidence can be obtained. The expectation and variance are used to generate channel estimates:

[0165] ,

[0166] .

[0167] Among them, the above This represents the channel estimate.

[0168] Optionally, the aforementioned execution entity can also determine the deviation between the estimated channel value and the actual channel value based on the normalized mean square error (NMSE). The actual channel value can be a preset channel value, generated by a channel model. In practice, NMSE can be used as a performance indicator to determine the deviation between the estimated channel value and the actual channel value, as shown in the following formula:

[0169] .

[0170] NMSE represents the aforementioned deviation. Indicates the first Real channel vector in sub-Monte Carlo simulation The estimate, It is the number of Monte Carlo trajectories.

[0171] In practice, the method of this application was compared with existing SBL, PCSBL, and HB-VBI algorithms under signal-to-noise ratio (SNR) conditions of 5dB to 30dB using normalized mean square error comparison. The experimental comparison figures are shown below. Figure 5 As shown. Figure 5In the diagram, (a) shows the NMSE curve as a function of the number of iterations. (b) shows the NMSE curve as a function of the signal-to-noise ratio (SNR). (c) shows the NMSE curve as a function of the number of pilots. Specifically, (a) shows the normalized mean square error convergence curves of each algorithm at a 15 dB SNR, with the number of iterations (Iter), indicating that the proposed algorithm has a lower NMSE. The reduction in NMSE reflects the algorithm's stronger ability to characterize channel sparsity and correlation, thus becoming a key factor in improving the performance of large-scale MIMO systems. (b) shows that the proposed algorithm maintains a significant advantage under all SNR conditions, consistently exhibiting the lowest NMSE. This robustness over a wide SNR range ensures reliable system operation in variable environments. (c) examines the relationship between NMSE and the number of pilots at a 15 dB SNR. When the number of pilots varies between 16 and 64, the proposed algorithm consistently maintains optimal performance, with significantly better channel characterization accuracy than other comparative schemes. The results demonstrate that the proposed algorithm can achieve high-precision estimation with very few pilots under conditions of limited pilot resources, which is particularly important for systems where pilots are scarce. Furthermore, maintaining estimation accuracy while reducing the number of pilots also implies more efficient utilization of system resources and bandwidth.

[0172] Further reference Figure 6 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a channel estimation apparatus for a large-scale multiple-input multiple-output system based on message passing. These apparatus embodiments are similar to... Figure 1 Corresponding to the method embodiments shown, this message-passing-based large-scale multiple-input multiple-output system channel estimation apparatus can be specifically applied to various electronic devices.

[0173] like Figure 6As shown, a message-passing-based channel estimation apparatus 600 for a large-scale multiple-input multiple-output system (MLMI) includes, in some embodiments, a first determining unit 601, an establishing unit 602, an updating unit 603, a second determining unit 604, a first generating unit 605, a third determining unit 606, and a second generating unit 607. The first determining unit 601 is configured to determine the received pilot signal corresponding to each antenna in the antenna set included in the MLMI, wherein the received pilot signal can be a combination of a measurement matrix, a composite channel impulse response, and additive white Gaussian noise. The establishing unit 602 is configured to establish a signal linear model based on the determined received pilot signals, wherein the signal linear model is a linear model composed of a received pilot signal matrix, a measurement matrix, a channel matrix, and a white Gaussian noise matrix. The updating unit 603 is configured to update the signal linear model based on the singular value decomposition result of the measurement matrix to obtain an updated signal model. The aforementioned updated signal model includes the aforementioned channel matrix; the second determining unit 604 is configured to determine the transmission probability, sparse feature representation, correlation, random variable distribution, and jump probability distribution corresponding to the aforementioned channel matrix based on three-layer Bayesian inference technology; the first generating unit 605 is configured to generate a joint probability density function based on the aforementioned transmission probability, the aforementioned sparse feature representation, the aforementioned correlation, the aforementioned random variable distribution, and the aforementioned jump probability distribution; the third determining unit 606 is configured to determine the factor graph corresponding to the aforementioned joint probability density function; and the second generating unit 607 is configured to generate a channel estimate based on the mixed belief propagation mean-field message update rule and the aforementioned factor graph.

[0174] It is understandable that the units described in the message-passing-based large-scale multiple-input multiple-output system channel estimation device 600 are related to the reference... Figure 1 The steps in the described method correspond accordingly. Therefore, the operations, features, and beneficial effects described above for the method are also applicable to the message-passing-based large-scale multiple-input multiple-output system channel estimation device 600 and the units contained therein, and will not be repeated here.

[0175] The following is for reference. Figure 7 It illustrates a schematic diagram of the structure of an electronic device (such as a computing device) suitable for implementing some embodiments of the present disclosure. Figure 7 The electronic device shown is merely an example and should not be construed as limiting the functionality or scope of the embodiments of this disclosure. Figure 7As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The memory may include a non-volatile storage medium and internal memory. The non-volatile storage medium may store an operating system and a computer program. The computer program includes program instructions that, when executed, cause the processor to perform any of the methods described above. The processor provides computational and control capabilities to support the operation of the entire computer device. The internal memory provides an environment for the execution of the computer program in the non-volatile storage medium; when executed by the processor, the computer program causes the processor to perform any of the methods described above. The network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present disclosure and does not constitute a limitation on the computer device to which the present disclosure is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0176] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0177] In one embodiment, the processor is configured to run a computer program stored in a memory to perform the following steps: for each antenna in the antenna set included in the large-scale multiple-input multiple-output system, determine the received pilot signal corresponding to the antenna, wherein the received pilot signal can be a combination of a measurement matrix, a composite channel impulse response, and additive white Gaussian noise; establish a signal linear model based on the determined received pilot signals, wherein the signal linear model is a linear model composed of the received pilot signal matrix, the measurement matrix, the channel matrix, and the white Gaussian noise matrix; update the signal linear model based on the singular value decomposition result of the measurement matrix to obtain an updated signal model, wherein the updated signal model includes the channel matrix; determine the transmission probability, sparse feature representation, correlation, random variable distribution, and jump probability distribution corresponding to the channel matrix based on three-layer Bayesian inference technology; generate a joint probability density function based on the transmission probability, the sparse feature representation, the correlation, the random variable distribution, and the jump probability distribution; determine the factor graph corresponding to the joint probability density function; and generate a channel estimate based on the mixed belief propagation mean-field message update rule and the factor graph.

[0178] This disclosure also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, and the method implemented when the program instructions are executed can be referred to the various embodiments of the methods described above.

[0179] The aforementioned computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. Alternatively, the aforementioned computer-readable storage medium may be an external storage device of the computer device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.

[0180] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0181] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A channel estimation method for a large-scale multiple-input multiple-output system based on message passing, characterized in that, include: For each antenna in the antenna set included in a large-scale multiple-input multiple-output system, determine the receiving pilot signal corresponding to the antenna, wherein the receiving pilot signal can be a combination of a measurement matrix, a composite channel impulse response, and additive white Gaussian noise; Based on the determined received pilot signals, a linear model of the signal is established, wherein the linear model of the signal is a linear model composed of the received pilot signal matrix, the measurement matrix, the channel matrix and the Gaussian white noise matrix; Based on the singular value decomposition results of the measurement matrix, the linear model of the signal is updated to obtain an updated signal model, wherein the updated signal model includes the channel matrix; Based on three-layer Bayesian inference technology, the transmission probability, sparse feature representation, correlation, random variable distribution and jump probability distribution corresponding to the channel matrix are determined. A joint probability density function is generated based on the transmission probability, the sparse feature representation, the correlation, the random variable distribution, and the jump probability distribution. Determine the factor graph corresponding to the joint probability density function; Channel estimates are generated based on the hybrid belief propagation mean-field message update rule and the factor graph.

2. The method according to claim 1, characterized in that, For each antenna in the antenna set included in a large-scale multiple-input multiple-output system, determining the corresponding received pilot signal for that antenna includes: For each antenna in the antenna set included in a large-scale multiple-input multiple-output system, perform the following steps: Based on orthogonal frequency division multiplexing (OFDM) technology, a set of pilot subcarriers for pilot symbol transmission is determined, wherein each pilot subcarrier in the set corresponds to a pilot symbol and a pilot position. Generate a pilot symbol matrix based on the pilot symbol corresponding to each pilot subcarrier in the pilot subcarrier set; A discrete Fourier transform matrix is ​​generated based on the pilot position corresponding to each pilot subcarrier in the pilot subcarrier set. The product between the pilot symbol matrix and the discrete Fourier transform matrix is ​​determined as the measurement matrix; The composite channel impulse response corresponding to the antenna is determined based on the channel length in the large-scale multiple-input multiple-output system. Determine the additive white Gaussian noise corresponding to the antenna; The product of the measurement matrix and the composite channel impulse response is added to the additive white Gaussian noise to obtain the received pilot signal.

3. The method according to claim 1, characterized in that, The step of establishing a linear model of the signal based on the determined received pilot signals includes: The determined individual receive pilot signals are converted into a receive pilot signal matrix; Convert the impulse responses of each composite channel into a channel matrix; Convert each additive Gaussian white noise into a Gaussian white noise matrix; A linear signal model is generated based on the received pilot signal matrix, the measurement matrix, the channel matrix, and the Gaussian white noise matrix, wherein the dependent variable of the linear signal model is the received pilot signal matrix, which is equal to the product of the measurement matrix and the channel matrix plus the Gaussian white noise matrix.

4. The method according to claim 1, characterized in that, The step of updating the linear model of the signal based on the singular value decomposition results of the measurement matrix to obtain an updated signal model includes: Determine the first unitary matrix, the second unitary matrix, and the singular value diagonal matrix corresponding to the measurement matrix; Determine the conjugate matrix corresponding to the first unitary matrix; The conjugate matrix is ​​multiplied by the linear signal model to obtain the updated signal model.

5. The method according to claim 1, characterized in that, The method for determining the transmission probability, sparse feature representation, correlation, random variable distribution, and jump probability distribution corresponding to the channel matrix based on three-layer Bayesian inference technology includes: Based on the Bayesian framework, the transmission probability corresponding to the channel matrix is ​​determined. Based on the Bernoulli Gaussian model, the sparse feature representation of the channel matrix in the antenna delay domain is determined; Based on a first-order Markov chain, the correlation between the channel matrix and each antenna in the antenna set is determined. Based on the aforementioned correlation, a jump probability distribution is generated; Based on the sparse feature representation, a random variable distribution is generated.

6. The method according to claim 1, characterized in that, The method further includes: Based on the normalized mean square error, the deviation between the estimated channel value and the true channel value is determined.

7. A channel estimation device for a large-scale multiple-input multiple-output system based on message passing, characterized in that, include: The first determining unit is configured to determine the received pilot signal corresponding to each antenna in the antenna set included in the large-scale multiple-input multiple-output system, wherein the received pilot signal can be a combination of a measurement matrix, a composite channel impulse response, and additive white Gaussian noise. The establishment unit is configured to establish a signal linear model based on each determined received pilot signal, wherein the signal linear model is a linear model composed of the received pilot signal matrix, the measurement matrix, the channel matrix, and the Gaussian white noise matrix; An update unit is configured to update the linear model of the signal based on the singular value decomposition result of the measurement matrix to obtain an updated signal model, wherein the updated signal model includes the channel matrix; The second determining unit is configured to determine the transmission probability, sparse feature representation, correlation, random variable distribution and jump probability distribution corresponding to the channel matrix based on three-layer Bayesian inference technology. The first generation unit is configured to generate a joint probability density function based on the transmission probability, the sparse feature representation, the correlation, the random variable distribution, and the jump probability distribution. The third determining unit is configured to determine the factor graph corresponding to the joint probability density function; The second generation unit is configured to generate channel estimates based on the hybrid belief propagation mean field message update rule and the factor graph.

8. An electronic device, characterized in that, include: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 6.

9. A computer-readable medium, characterized in that, It stores a computer program thereon, wherein the computer program, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images