Signal processing method and device for cellular-free large-scale MIMO (Multiple Input Multiple Output) system

Abstract

The invention provides a cellular-free large-scale multiple-input-multiple-output system signal processing method and device, and relates to the technical field of communication, and the method comprises the steps: obtaining a first Gaussian parameter transmitted from a factor node to a variable node; wherein the factor node is determined according to a receiving signal of an access point in the cellular-free large-scale multiple-input multiple-output system, and the variable node is determined according to a sending symbol of the cellular-free large-scale multiple-input multiple-output system; the first Gaussian parameter is calculated by each access point according to an initialized Gaussian parameter, a received signal and channel information; target extrinsic information is determined according to the first Gaussian parameter, and a second Gaussian parameter transmitted from the variable node to the factor node is calculated according to the target extrinsic information; and according to the second Gaussian parameter, updating the first Gaussian parameter, and re-determining the extrinsic information, thereby updating the second Gaussian parameter until a preset iteration condition is satisfied, and obtaining a target sending symbol.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and in particular, to a signal processing method and apparatus for a cell-free massive multiple-input multiple-output system. Background Art

[0002] In order to overcome the inter-cell interference problem of traditional cellular networks, cell-free massive multiple-input multiple-output (CF-mMIMO) has been proposed and widely concerned. Compared with the traditional co-located MIMO system that relies on a fixed base station (BS) to serve multiple users in a predefined cell, CF-mMIMO adopts a distributed architecture and densely deploys a large number of access points (APs) in the coverage area. Specifically, the access points are connected to a central processing unit (CPU) through high-speed optical fibers or wireless fronthaul links, so as to achieve coordinated signal processing. This distributed MIMO system uses spatial multiplexing and beamforming technologies to serve multiple users simultaneously, making full use of the rich spatial diversity provided by the dense AP deployment. Since more APs are closer to the edge users, CF-mMIMO can reduce inter-cell interference, thereby improving the signal-to-interference-plus-noise ratio (SINR) and increasing the spectral efficiency (SE).

[0003] An efficient data detection algorithm, which can effectively and accurately recover the transmitted signals of users from the signals received by multiple APs, is crucial for maximizing the advantages of the CF-mMIMO system. In the related technologies, there is a lack of an effective way to perform signal recovery in a cell-free massive multiple-input multiple-output system. Therefore, how to efficiently perform signal recovery in a cell-free massive multiple-input multiple-output system has become an urgent problem in the industry. Summary of the Invention

[0004] The present invention provides a signal processing method and apparatus for a cell-free massive multiple-input multiple-output system, so as to solve the problem of how to efficiently perform signal processing in a cell-free massive multiple-input multiple-output system in the prior art.

[0005] The present invention provides a signal processing method for a cell-free massive multiple-input multiple-output system, including: Obtain the first Gaussian parameter passed from the factor node to the variable node; wherein, the factor node is determined according to the received signal of the access point in the cell-free massive multiple-input multiple-output system, and the variable node is determined according to the transmitted symbol of the cell-free massive multiple-input multiple-output system; the first Gaussian parameter is calculated by each of the access points based on the initialized Gaussian parameter, the received signal, and the channel information; Determine the target extrinsic information according to the first Gaussian parameter, and calculate the second Gaussian parameter passed from the variable node to the factor node according to the target extrinsic information; Update the first Gaussian parameter according to the second Gaussian parameter, re-determine the extrinsic information, and thus update the second Gaussian parameter until a preset iteration condition is satisfied to obtain the target transmitted symbol.

[0006] According to a signal processing method for a cell-free massive multiple-input multiple-output system provided by the present invention, updating the first Gaussian parameter according to the second Gaussian parameter, re-determining the extrinsic information, and thus updating the second Gaussian parameter until a preset iteration condition is satisfied to obtain the target transmitted symbol includes: For each of the second Gaussian parameters, send the second Gaussian parameter back to each of the access points so that each access point updates the first Gaussian parameter according to the received second Gaussian parameter; Re-determine the target extrinsic information according to the updated first Gaussian parameter, and update the second Gaussian parameter according to the re-determined target extrinsic information; When a preset iteration stop condition is satisfied, stop updating the second Gaussian parameter; Calculate the posterior probability of the transmitted symbol according to the first target Gaussian parameter to obtain the target transmitted symbol; wherein, the first target Gaussian parameter is the first Gaussian parameter obtained when the preset iteration stop condition is satisfied.

[0007] According to a signal processing method for a cell-free massive multiple-input multiple-output system provided by the present invention, calculating the second Gaussian parameter passed from the variable node to the factor node according to the target extrinsic information includes: Calculate the posterior probability of each transmitted symbol according to the mean and variance of the target extrinsic information of the transmitted symbol corresponding to each variable node, and the prior probability information of the transmitted symbol; Determine the second Gaussian parameter from the variable node to the factor node according to the mean and variance of each of the posterior probabilities.

[0008] According to a signal processing method for a cell-free massive multiple-input multiple-output system provided by the present invention, the first Gaussian parameter includes: the mean and variance of the message passed from the factor node to the variable node.

[0009] According to a signal processing method for a cell-free massive multiple-input multiple-output system provided by the present invention, the target extrinsic information includes: the weighted average of the means of the messages transmitted from multiple factor nodes to a variable node, and the weighted average of the variances of the messages transmitted from multiple factor nodes to a variable node.

[0010] According to a signal processing method for a cell-free massive multiple-input multiple-output system provided by the present invention, the calculation method of the second Gaussian parameter specifically includes: Based on the mean and variance of the probability distribution from a factor node to a variable node, obtain the mean and variance of the approximate posterior probability distribution of each user; According to the approximate posterior probability variance and the variance of the message transmitted from a factor node to a variable node, update the variance of the message transmitted from the variable node to the factor node; According to the approximate posterior probability mean, the approximate posterior probability variance, the variance and mean of the message transmitted from a factor node to a variable node, update the mean of the message transmitted from the variable node to the factor node; According to the updated mean and variance of the message transmitted from the variable node to the factor node, obtain the second Gaussian parameter.

[0011] The present invention also provides a signal processing device for a cell-free massive multiple-input multiple-output system, including: An acquisition module, configured to acquire a first Gaussian parameter transmitted from a factor node to a variable node; wherein, the factor node is determined according to the received signal of an access point in the cell-free massive multiple-input multiple-output system, and the variable node is determined according to the transmitted symbol of the cell-free massive multiple-input multiple-output system; the first Gaussian parameter is calculated by each of the access points according to the initialized Gaussian parameter, the received signal, and the channel information; A calculation module, configured to determine target extrinsic information according to the first Gaussian parameter, and calculate a second Gaussian parameter transmitted from the variable node to the factor node according to the target extrinsic information; An update module, configured to update the first Gaussian parameter according to the second Gaussian parameter, re-determine the extrinsic information, thereby update the second Gaussian parameter until a preset iteration condition is met, and obtain a target transmitted symbol.

[0012] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the signal processing method for a cell-free massive multiple-input multiple-output system as described in any one of the above.

[0013] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the signal processing method for a cell-free massive multiple-input multiple-output system as described in any one of the above.

[0014] The present invention also provides a computer program product, including a computer program, which when executed by a processor, implements the signal processing method for the cell-free massive multiple-input multiple-output system as described in any one of the above.

[0015] The signal processing method and device for the cell-free massive multiple-input multiple-output system provided by the present invention model the transmitted symbols and received signals as factor nodes and variable nodes, so that the signal detection problem can be transformed into a message passing problem on a factor graph, thereby more accurately estimating the transmitted symbols, and by transmitting Gaussian parameters between the access point and the central processor, complex matrix inversion operations are avoided. Especially in the cell-free massive multiple-input multiple-output system, the complexity of matrix inversion operations is very high, while the transmission of Gaussian parameters greatly reduces the computational complexity. At the same time, by determining the target extrinsic information according to the first Gaussian parameter, the channel information and the statistical characteristics of the received signals can be better utilized, thereby optimizing the system performance; finally, the transmitted symbols are recovered through the second Gaussian parameter, effectively performing cell-free massive multiple-input multiple-output system detection, thus helping the cell-free massive multiple-input multiple-output system to better exert its advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0017] Figure 1 is one of the flow schematic diagrams of the signal processing method for the cell-free massive multiple-input multiple-output system provided by the present invention; Figure 2 is the second of the flow schematic diagrams of the signal processing method for the cell-free massive multiple-input multiple-output system provided by the present invention; Figure 3 is the factor graph schematic diagram of an AP in the CF-mMIMO system in the embodiment of the present application; Figure 4 is one of the schematic diagrams of the simulation results provided by the embodiment of the present application; Figure 5 is the second of the schematic diagrams of the simulation results provided by the embodiment of the present application; Figure 6 is one of the analysis schematic diagrams provided by the embodiment of the present application; Figure 7 is the second of the analysis schematic diagrams provided by the embodiment of the present application; Figure 8Schematic structural diagram of a signal processing device for a cell-free massive multiple-input multiple-output system provided by an embodiment of the present application; Figure 9 Schematic structural diagram of an electronic device provided by the present invention. Detailed implementation manners

[0018] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts shall fall within the protection scope of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0019] Figure 1 One of the flow diagrams of a signal processing method for a cell-free massive multiple-input multiple-output system provided by the present invention, as Figure 1 shown, includes: Step 110, obtaining first Gaussian parameters transmitted from factor nodes to variable nodes; wherein, the factor nodes are determined according to received signals of access points in the cell-free massive multiple-input multiple-output system, and the variable nodes are determined according to transmitted symbols in the cell-free massive multiple-input multiple-output system; the first Gaussian parameters are calculated by each of the access points according to initialized Gaussian parameters, received signals and channel information; In the present invention, in a cell-free massive multiple-input multiple-output system, factor nodes and variable nodes are used to model the statistical probabilities of transmitted symbols and received signals. Factor nodes are abstractions of received signals, while variable nodes are abstractions of transmitted symbols. By constructing factor nodes and variable nodes, the joint probability distribution of transmitted symbols and received signals can be accurately characterized. This modeling method can precisely depict the probability correlation between received signals and transmitted symbols.

[0020] Specifically, first, transmitted symbols, received signals, channel information and noise information can be obtained from the physical layer of the system to effectively perform signal detection.

[0021] More specifically, with the collected data, a transmitted symbol vector x and a received signal vector y are constructed. These vectors contain signals transmitted by all user devices in the system and signals received by each access point.

[0022] Extract a channel matrix and a noise variance from the channel information ; is a measure of the background noise present in the system. H describes the propagation characteristics of signals during transmission and reception. The channel matrix is the key to understanding how signals propagate between user devices and access points.

[0023] Gaussian parameter initialization, specifically, it can refer to initializing the Gaussian parameters from variable nodes to factor nodes. To obtain the initialized Gaussian parameters, where the superscript represents the number of iterations.

[0024] Each access point uses the signals it receives and the known channel state information, combined with the initialized Gaussian parameters, to calculate the first Gaussian parameters transmitted from factor nodes to variable nodes. These parameters include the mean and variance, which reflect the preliminary estimates of the received signals and channel state information. These calculated first Gaussian parameters are then sent to the central processor for further processing and global information integration, so as to achieve accurate detection and estimation of the transmitted symbols.

[0025] Step 120, determine the target extrinsic information according to the first Gaussian parameters, and calculate the second Gaussian parameters transmitted from the variable nodes to the factor nodes according to the target extrinsic information; The central processor first collects the first Gaussian parameters calculated by each access point according to the received signals and channel information. These parameters include the mean and variance, reflecting the preliminary estimates of the received signals.

[0026] The central processor uses these parameters to determine the target extrinsic information, that is, in the iterative process, the new information provided by the system except for the prior information.

[0027] The determination of the extrinsic information is based on the data provided by all access points, which contains global information and helps to more accurately estimate the transmitted symbols.

[0028] The central processor calculates the second Gaussian parameters transmitted from the variable nodes to the factor nodes according to the target extrinsic information. This calculation process involves updating the first Gaussian parameters to reflect the new information and constraints. The updated second Gaussian parameters will be sent back to each access point so that they can further optimize their signal estimates.

[0029] Step 130, update the first Gaussian parameters according to the second Gaussian parameters, re-determine the extrinsic information, and thus update the second Gaussian parameters until the preset iteration condition is met to obtain the target transmitted symbols.

[0030] In the present invention, the Gaussian parameters of the factor nodes and variable nodes are initialized according to the transmitted symbols and received signals. The Gaussian parameters generally include the mean and variance of the messages transmitted from the factor nodes to the variable nodes, and are used to describe the statistical characteristics of the transmitted symbols and received signals. In the present invention, each access point calculates the first Gaussian parameters transmitted from the factor nodes to the variable nodes according to the initialized Gaussian parameters, received signals and channel information. The first Gaussian parameters reflect the preliminary estimates of the received signals by each AP.

[0031] Based on the first Gaussian parameter, the central processing unit determines extrinsic information outside the target, which is the product of multiple Gaussian distributions and is used to update the posterior probability distribution of the variable nodes.

[0032] Based on the extrinsic information outside the target, the second Gaussian parameter transmitted from the variable nodes to the factor nodes is calculated. These parameters are used to update the information of the factor nodes and further optimize the estimation of the transmitted symbols.

[0033] In the present invention, based on the second Gaussian parameter, the first Gaussian parameter is updated, and by recalculating the extrinsic information, the second Gaussian parameter is further optimized. The above steps are repeated until a preset iteration condition is met, such as reaching the maximum number of iterations or the convergence condition.

[0034] In the present invention, when the iteration process meets the preset condition, the first Gaussian parameter obtained in the last iteration process is used to calculate the posterior probability of the transmitted symbol, thereby obtaining the target transmitted symbol and realizing the recovery of the transmitted symbol.

[0035] Optionally, based on the second Gaussian parameter, the first Gaussian parameter is updated, the extrinsic information outside the target is re-determined, and thereby the second Gaussian parameter is updated until a preset iteration condition is met to obtain the target transmitted symbol, including: For each of the second Gaussian parameters, the second Gaussian parameter is sent back to each of the access points so that each access point updates the first Gaussian parameter according to the received second Gaussian parameter; Based on the updated first Gaussian parameter, the extrinsic information outside the target is re-determined, and the second Gaussian parameter is updated according to the re-determined extrinsic information outside the target; When the preset iteration stop condition is met, the update of the second Gaussian parameter is stopped; Based on the first target Gaussian parameter, the posterior probability of the transmitted symbol is calculated to obtain the target transmitted symbol; wherein, the first target Gaussian parameter is the first Gaussian parameter obtained when the preset iteration stop condition is met.

[0036] In the present invention, each second Gaussian parameter is sent back to each access point. These parameters contain the feedback information of the variable nodes to the factor nodes and are used to update the state of the factor nodes. Each access point updates its first Gaussian parameter according to the received second Gaussian parameter. The update process usually involves fusing the existing parameters and the received feedback information to obtain a more accurate estimate.

[0037] After the access point updates the first Gaussian parameter, the variable nodes need to re-determine the extrinsic information outside the target. This includes collecting the new first Gaussian parameters of all factor nodes connected to the variable nodes and calculating the extrinsic information outside the target.

[0038] Based on the reconfirmed off-target information, the variable node updates the second Gaussian parameters it passes to the factor node. This process is similar to the previous step, but uses the updated first Gaussian parameters.

[0039] Repeat the above steps until the preset iteration stop condition is met. The stop condition can be reaching the maximum number of iterations or meeting the preset training time.

[0040] When the iteration stop condition is met, the second Gaussian parameter is stopped from being updated. At this point, the first Gaussian parameter of the factor node has converged to a stable state and can be used as the first target Gaussian parameter.

[0041] In the present invention, the posterior probability distribution of each transmitted symbol is calculated according to the first target Gaussian parameter; the posterior probability distribution includes the most likely value of the transmitted symbol under a given received signal and system model and its uncertainty, and for each transmitted symbol, the value that maximizes the posterior probability is found. This value is considered to be the optimal detection value of the symbol, and the optimal detection values ​​of all transmitted symbols are combined to obtain a complete transmitted symbol.

[0042] In the present invention, the system can gradually optimize the estimation of the transmitted symbol through an iterative process, thereby improving the detection accuracy and system performance. The system makes full use of the statistical characteristics of the channel information and the received signal, and gradually approaches the real transmitted symbol by iteratively updating the Gaussian parameters and external information, thereby achieving efficient signal detection in a complex environment. The target transmitted symbol finally obtained is the result of multiple rounds of iterative optimization, and has higher accuracy and reliability.

[0043] Figure 2 FIG. 2 is a flow chart of a method for processing a signal in a non-cellular large-scale multi-input multi-output system provided by the present invention. Figure 2 As shown, the method includes the following: Step S1, obtaining a first Gaussian parameter transmitted from a factor node to a variable node; wherein the factor node is determined according to a received signal of an access point in the non-cellular large-scale multi-input multi-output system, and the variable node is determined according to a transmitted symbol of the non-cellular large-scale multi-input multi-output system; the first Gaussian parameter is calculated by each of the access points according to an initialization Gaussian parameter, a received signal and channel information; Step S2, determining target external information according to the mean and variance of the first Gaussian parameter, so as to calculate a second Gaussian parameter transmitted from the variable node to the factor node according to the external information; Step S3, sending the second Gaussian parameters back to each access point, so that each access point updates the first Gaussian parameters according to the received second Gaussian parameters; Repeat the above steps S2 - S3 according to the updated first Gaussian parameter until a preset iteration stop condition is met, and output the detection result of signal estimation.

[0044] In the present invention, the central processing unit determines the target extrinsic information according to the mean and variance of the first Gaussian parameter, and calculates the second Gaussian parameter transmitted from the variable node to the factor node based on this. These second Gaussian parameters contain new estimates and uncertainty information of the transmitted symbols.

[0045] In step S3, these second Gaussian parameters are sent back to each access point. After receiving the second Gaussian parameters, each access point updates its local first Gaussian parameter using this information. This update process enables each access point to adjust the estimate of the transmitted symbol according to the global information, thereby improving the accuracy and reliability of detection.

[0046] The updated first Gaussian parameter will be used for the next round of iterative calculation. Specifically, the system will repeat steps S2 and S3, that is, recalculate the target extrinsic information and the second Gaussian parameter according to the updated first Gaussian parameter, and send these parameters back to the access points for update again. This iterative process will continue until a preset iteration stop condition is met.

[0047] The preset iteration stop condition can be reaching the maximum number of iterations, that is, the difference between the updated parameter and the parameter before update is very small, indicating that the system has reached a stable state. Once the stop condition is met, the system will output the final signal estimation detection result, and these results represent the best estimate of the symbols transmitted by the user equipment.

[0048] In the present invention, by modeling the transmitted symbol and the received signal as factor nodes and variable nodes, the signal detection problem can be transformed into a message passing problem on the factor graph, so as to more accurately estimate the transmitted symbol, and by transmitting Gaussian parameters between the access point and the central processing unit, complex matrix inversion operations are avoided. Especially in a cell - free massive multiple - input multiple - output system, the complexity of matrix inversion operations is very high, while the transmission of Gaussian parameters greatly reduces the computational complexity. At the same time, through the process of determining the target extrinsic information according to the first Gaussian parameter, the channel information and the statistical characteristics of the received signal can be better utilized to optimize the system performance; finally, the transmitted symbol is recovered through the second Gaussian parameter, and the cell - free massive multiple - input multiple - output system detection is effectively carried out, thereby helping the cell - free massive multiple - input multiple - output system to better play its advantages. Optionally, calculating the second Gaussian parameter transmitted from the variable node to the factor node according to the target extrinsic information includes: Calculate the posterior probability of each of the transmitted symbols based on the mean and variance of the extrinsic information corresponding to the transmitted symbols of each of the variable nodes, and the prior probability information of the transmitted symbols; Determine the second Gaussian parameter from the variable node to the factor node according to the mean and variance of each of the posterior probabilities.

[0049] In the present invention, for the transmitted symbol corresponding to each variable node, the mean and variance of the extrinsic information are utilized, combined with the prior probability information of the transmitted symbol, to calculate the posterior probability of each transmitted symbol. Among them, the posterior probability distribution reflects the most likely value of each transmitted symbol given the received signal and channel information.

[0050] More specifically, extract the mean and variance from the calculated posterior probability distribution, and determine the second Gaussian parameter transmitted from the variable node to the factor node according to these mean and variance. These parameters will be used to update the information of the factor node to further optimize the estimation of the transmitted symbol.

[0051] In the present invention, by utilizing the statistical characteristics of the channel information and the received signal, the Gaussian parameters and extrinsic information are iteratively updated to gradually approximate the true transmitted symbol, thereby achieving efficient signal detection in a complex environment. The finally obtained target transmitted symbol is the result of multiple rounds of iterative optimization, with higher accuracy and reliability.

[0052] Optionally, the first Gaussian parameter includes: the mean and variance of the message transmitted from the factor node to the variable node.

[0053] In the present invention, the first Gaussian parameter includes the mean and variance of the message transmitted from the factor node to the variable node. These parameters are set in the initialization stage of the algorithm and updated according to the received signal and channel information in each round of iteration. The mean and variance jointly describe the prior probability distribution of the transmitted symbol. The mean represents the expected value of the transmitted symbol, and the variance reflects the uncertainty of the transmitted symbol.

[0054] Optionally, the target extrinsic information includes: the weighted average of the means of the messages transmitted from multiple factor nodes to the variable node, and the weighted average of the variances of the messages transmitted from multiple factor nodes to the variable node.

[0055] In the present invention, the target extrinsic information is determined by the weighted average of the means and variances of the messages transmitted from multiple factor nodes to the variable node.

[0056] The mean and variance of the out-of-target information are obtained by weighted averaging the means and variances of the messages passed from multiple factor nodes to variable nodes. The setting of the weights reflects the relative importance of each factor node in the overall information, usually based on the quality of the channel, the strength of the received signal, or other relevant factors. The calculation method of weighted averaging enables the out-of-target information to synthesize the observations of multiple access points, thereby more accurately reflecting the statistical characteristics of the transmitted symbols.

[0057] Optionally, the calculation method of the second Gaussian parameter specifically includes: Based on the mean and variance of the probability distribution from the factor node to the variable node, obtain the mean and variance of the approximate posterior probability distribution of each user; According to the variance of the approximate posterior probability and the variance of the message passed from the factor node to the variable node, update the variance of the message passed from the variable node to the factor node; According to the mean of the approximate posterior probability, the variance of the approximate posterior probability, the variance and mean of the message passed from the factor node to the variable node, update the mean of the message passed from the variable node to the factor node; According to the updated mean and variance of the message passed from the variable node to the factor node, obtain the second Gaussian parameter.

[0058] In the present invention, first starting from the mean and variance of the message passed from the factor node to the variable node, these parameters reflect the preliminary estimation of the received signal by each access point. By combining these messages and the prior probability information of the transmitted symbol, the mean and variance of the approximate posterior probability distribution of each user are calculated, and this step usually involves the product and normalization processing of multiple Gaussian distributions.

[0059] Then, using the variance of the approximate posterior probability distribution and the variance of the message passed from the factor node to the variable node, update the variance of the message passed from the variable node to the factor node. Specifically, by combining the variance of the approximate posterior probability and the variance of the message passed from the factor node to the variable node, for example, using weighted averaging or more complex statistical processing methods, to obtain the updated variance.

[0060] More specifically, further using the mean and variance of the approximate posterior probability distribution, as well as the mean and variance of the message passed from the factor node to the variable node, update the mean of the message passed from the variable node to the factor node. This step also uses weighted averaging or more complex statistical processing methods, combining all relevant parameters, to obtain the updated mean.

[0061] In the present invention, the updated mean and variance can be used to effectively determine the second Gaussian parameter passed from the variable node to the factor node, and the second Gaussian parameter will be used in the next iteration to update the information of the factor node, further optimizing the estimation of the transmitted symbol.

[0062] In the present invention, the Gaussian parameters can be iteratively updated to gradually approximate the true transmitted symbols, thereby achieving efficient signal detection in a complex environment and improving the detection accuracy and system performance.

[0063] More specifically, in the embodiments of the present application, the cell-free massive multiple-input multiple-output system may specifically be an uplink CF-mMIMO system composed of APs and single-antenna users. Each AP is equipped with antennas. All APs are connected to a CPU with high computing power through a fronthaul link. Importantly, the CPU enables all APs to serve all users within the same time-frequency resources.

[0064] During the uplink payload data transmission phase, all users transmit information to the access points simultaneously. Within each coherence period, the channel between the th user and the th AP is modeled as: (1) where represents the spatial correlation matrix. Therefore, the baseband signal received by the th AP is obtained by the following formula: (2) where, is the transmit power of the th user, represents the transmission information symbol mapped from -QAM constellation, is the additive white Gaussian noise (AWGN) at the receiving end. To represent the channel matrix in a more concise form, the channel matrix between the th AP and all users is represented as . Therefore, the uplink data detection problem of CF-mMIMO can be formulated as: Given the channel matrix between all users and all APs, the received signal , and the noise variance , infer the transmitted signal .

[0065] It is assumed that the channel state information (CSI) is estimated at each AP. Therefore, the posterior distribution of (3) where the joint prior distribution is discretely and uniformly distributed on the constellation points and is the -th likelihood probability distribution.

[0066] The EP-based approximate message passing algorithm is an application of the sum-product algorithm. Its basic idea is to use the expectation propagation idea to approximate the messages transmitted on the factor graph (FG, factor graph) so as to calculate the joint distribution of the marginal function. Based on the above CF-mMIMO model, the joint distribution can be factorized as (4) where is the -th element in the channel vector .

[0067] Figure 3 is a schematic diagram of the factor graph of an AP in the CF-mMIMO system in the embodiment of the present application. As shown in Figure 3 , the expression in formula (4) can be described by the factor graph, thereby explaining the information transmission process in the approximate message passing algorithm. Define and as the messages transmitted from the factor graph variable node (VN, variable node) to the factor node (FN, factor node) and vice versa, respectively. Then, the message update rule in the factor graph can be derived as follows: (5) (6) where is the prior information transmitted from the mapping node to VN . Therefore, the marginal posterior distribution of the -th symbol can be calculated in the following way: (7) To avoid the exhaustive traversal of all constellation points every time messages are transmitted through the factor graph, the EP-based AMP algorithm uses the expectation propagation technique to treat the discrete probability as a continuous probability distribution. Specifically, the posterior probability and the message are both approximated as complex Gaussian distributions and , respectively. Under the moment matching condition, is obtained by the following formula: (8) Naturally, since , at this time also follows a Gaussian distribution . The factor, at this time, formula (6) can be rewritten in integral form as: (9) According to formulas (5) and (9), the messages passed back and forth in the factor graph at this time are as follows: (10) (11) (12) (13) Optionally, the transmitted symbol, received signal, channel information, and noise information can be obtained from the physical layer of the system to effectively perform signal detection.

[0068] More specifically, with the collected data, a transmitted symbol vector x and a received signal vector y are constructed. These vectors contain the signals transmitted by all user devices in the system and the signals received by each access point.

[0069] The channel matrix H and the noise variance are extracted from the channel information ; is a measure of the background noise present in the system. H describes the propagation characteristics of the signal between transmission and reception. The channel matrix is the key to understanding how the signal propagates between user devices and access points.

[0070] Gaussian parameter initialization, specifically, it can refer to initializing the Gaussian parameters from VN to FN , where the superscript represents the number of iterations.

[0071] In the embodiments of the present application, each access point (AP) calculates the first Gaussian parameter passed from the factor node (FN) to the variable node (VN) according to the received signal and channel information.

[0072] In the embodiments of the present application, for the signal processing method of an expectation propagation based distributed approximate message passing (EP-dAMP) detector, the calculation method of the first Gaussian parameter can specifically include: The AP first calculates the Gaussian parameter according to formulas (10) and (11) . Among them can be calculated and completed in the preprocessing stage before iteration. It can simplify the traversal process and avoid traversing messages, and rewrite the update process as: (14) (15) Among them, and .

[0073] In the embodiment of the present application, for the signal processing method based on a low-complexity distributed approximate message passing (LC-dAMP) detector, according to the antenna ratio characteristics in the cell-free network , the average amplitude of is much larger than can be simply approximated as . Substituting into formula (13), then at this time can be calculated as: (16) And, among them can be omitted as a high-order term. Therefore, can be directly approximated as . According to the above series of approximation methods, the AP side and can be directly calculated as: (17) (18) In the embodiment of the present application, for the signal processing method based on an expectation propagation based distributed approximate message passing (EP-dAMP) detector, calculating the second Gaussian parameter specifically includes: After updating all the messages of APs, the CPU collects the messages that have been updated from each AP to obtain the extrinsic information of the target. As the mean and variance of the product distribution of multiple Gaussian distributions, the extrinsic information parameters and are calculated as follows: (19) (20) According to the calculated mean and variance of the extrinsic information of each transmitted symbol, and the prior probability , the The approximate posterior probability of a symbol can be calculated as: (21) where is the normalization coefficient.

[0074] Next, for each user, iterate through the receive antennas of each AP, and use equations (8), (12), and (13) to calculate the message transmitted from VN to FN. At this point, the round of detection iteration is completed, and are reassigned to each AP from the CPU and enter the next iteration until the maximum number of iterations is reached.

[0075] In the embodiment of the present application, for a signal processing method based on a low-complexity distributed approximate message passing (LC-dAMP) detector, if the number of users is large, the high-order terms in equation (15) can be directly ignored. According to , the extrinsic information of each symbol in the CPU can be calculated as: (22) It can be seen that the fronthaul overhead in EP-dAMP is large, and each AP needs to transmit complex scalars to the CPU for each channel use. To reduce the fronthaul overhead, the channel statistics data is transmitted from the AP to the CPU in the first iteration. Therefore, is rewritten as: (23) At this time, the fronthaul overhead is reduced to complex scalars.

[0076] In the embodiment of the present application, after each AP receives the second Gaussian parameter, it uses this information to update its local first Gaussian parameter. This process is repeated between the AP and the CPU until a preset iteration stop condition is met.

[0077] When the iteration process meets the preset stop condition (such as reaching the maximum number of iterations or detecting convergence), the CPU outputs the final signal estimation result. These results represent the best estimate of the symbols transmitted by the user equipment.

[0078] In the embodiments of the present application, by iteratively updating Gaussian parameters, the system can gradually improve the estimation of transmitted symbols, thereby improving the detection accuracy. The introduction of extrinsic information enables each iteration to be optimized using global information, further enhancing the detection effect. Through efficient signal processing methods, the system can better utilize spectrum resources and computing resources, improving spectrum efficiency and system capacity. The design of the distributed architecture enables the system to flexibly respond to different user requirements and network loads.

[0079] Optionally, Figure 4 is one of the schematic diagrams of the simulation results provided by the embodiments of the present application. As Figure 4 shown, EP-dAMP and LC-dAMP can maintain performance similar to that of distributed EP under different QAM modulations, and are superior to the centralized LMMSE detector.

[0080] Figure 5 is the second schematic diagram of the simulation results provided by the embodiments of the present application. As Figure 5 shown, the simulation results show that EP-dAMP and LC-dAMP can maintain performance similar to that of distributed EP under different CF-mMIMO antenna scenarios, and are superior to the centralized LMMSE detector.

[0081] Figure 6 is one of the analysis schematic diagrams provided by the embodiments of the present application. Figure 7 is the second analysis schematic diagram provided by the embodiments of the present application. As Figure 6 and Figure 7 shown, the complexity analysis results show that the inventive EP-dAMP and LC-dAMP detectors can save 54% - 94.9% of the complexity in distributed EP and maintain similar performance. Among them, the LC-dAMP detector can save 60% of the computational complexity in the AP of the EP-dAMP detector.

[0082] It can be seen that the two distributed detectors for CF-mMIMO of the present invention have low implementation complexity and excellent and robust performance.

[0083] Next, the signal processing device of the cell-free massive multiple-input multiple-output system provided by the present invention will be described. The signal processing device of the cell-free massive multiple-input multiple-output system described below can be correspondingly referred to the signal processing method of the cell-free massive multiple-input multiple-output system described above.

[0084] Figure 8 is the schematic diagram of the structure of the signal processing device of the cell-free massive multiple-input multiple-output system provided by the embodiments of the present application. As Figure 8 shown, it includes: An acquisition module 810, which is configured to acquire first Gaussian parameters passed from factor nodes to variable nodes; wherein, the factor nodes are determined according to received signals of access points in the cell-free massive multiple-input multiple-output system, and the variable nodes are determined according to transmission symbols of the cell-free massive multiple-input multiple-output system; the first Gaussian parameters are calculated by each of the access points based on initialized Gaussian parameters, received signals, and channel information; A calculation module 820 is configured to determine target extrinsic information according to the first Gaussian parameters, and calculate second Gaussian parameters passed from the variable nodes to the factor nodes according to the target extrinsic information; An update module 830 is configured to update the first Gaussian parameters according to the second Gaussian parameters, re-determine the extrinsic information, thereby updating the second Gaussian parameters until a preset iteration condition is met, and obtain target transmission symbols.

[0085] In an embodiment of the present application, by modeling transmission symbols and received signals as factor nodes and variable nodes, the signal detection problem can be transformed into a message passing problem on a factor graph, so as to more accurately estimate transmission symbols, and by passing Gaussian parameters between access points and a central processor, complex matrix inversion operations are avoided. Especially in a cell-free massive multiple-input multiple-output system, the complexity of matrix inversion operations is very high, while the passing of Gaussian parameters greatly reduces the computational complexity. At the same time, through the determination process of determining target extrinsic information according to the first Gaussian parameters, the statistical characteristics of channel information and received signals can be better utilized, thereby optimizing the system performance; finally, the transmission symbols are recovered through the second Gaussian parameters, and the cell-free massive multiple-input multiple-output system detection is effectively performed, so as to help the cell-free massive multiple-input multiple-output system better exert its advantages.

[0086] Figure 9 Illustrates a schematic physical structure diagram of an electronic device, such as Figure 9As shown, the electronic device may include: a processor 910, a communications interface 920, a memory 930, and a communication bus 940. Among them, the processor 910, the communications interface 920, and the memory 930 complete communication with each other through the communication bus 940. The processor 910 may call logic instructions in the memory 930 to execute a signal processing method for a cell-free massive multiple-input multiple-output system. The method includes: obtaining first Gaussian parameters transmitted from factor nodes to variable nodes; wherein, the factor nodes are determined according to the received signals of access points in the cell-free massive multiple-input multiple-output system, and the variable nodes are determined according to the transmitted symbols of the cell-free massive multiple-input multiple-output system; the first Gaussian parameters are calculated by each of the access points based on initialization Gaussian parameters, received signals, and channel information; Determine target extrinsic information according to the first Gaussian parameters, and calculate second Gaussian parameters transmitted from the variable nodes to the factor nodes according to the target extrinsic information; Update the first Gaussian parameters according to the second Gaussian parameters, re-determine the extrinsic information, thereby updating the second Gaussian parameters until a preset iteration condition is met, and obtain target transmitted symbols.

[0087] In addition, when the logic instructions in the above-mentioned memory 930 are implemented in the form of software function units and sold or used as independent products, they may be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, may be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs that can store program codes.

[0088] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the signal processing method of the cell-free massive multiple-input multiple-output system provided by the above-mentioned various methods. The method includes: obtaining a first Gaussian parameter transmitted from a factor node to a variable node; wherein, the factor node is determined according to the received signal of an access point in the cell-free massive multiple-input multiple-output system, and the variable node is determined according to the transmitted symbol of the cell-free massive multiple-input multiple-output system; the first Gaussian parameter is calculated by each of the access points according to an initial Gaussian parameter, a received signal, and channel information. Determine a target extrinsic information according to the first Gaussian parameter, and calculate a second Gaussian parameter transmitted from the variable node to the factor node according to the target extrinsic information. Update the first Gaussian parameter according to the second Gaussian parameter, re-determine the extrinsic information, and thus update the second Gaussian parameter until a preset iteration condition is satisfied to obtain a target transmitted symbol.

[0089] In another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the signal processing method of the cell-free massive multiple-input multiple-output system provided by the above-mentioned various methods. The method includes: obtaining a first Gaussian parameter transmitted from a factor node to a variable node; wherein, the factor node is determined according to the received signal of an access point in the cell-free massive multiple-input multiple-output system, and the variable node is determined according to the transmitted symbol of the cell-free massive multiple-input multiple-output system; the first Gaussian parameter is calculated by each of the access points according to an initial Gaussian parameter, a received signal, and channel information. Determine a target extrinsic information according to the first Gaussian parameter, and calculate a second Gaussian parameter transmitted from the variable node to the factor node according to the target extrinsic information. Update the first Gaussian parameter according to the second Gaussian parameter, re-determine the extrinsic information, and thus update the second Gaussian parameter until a preset iteration condition is satisfied to obtain a target transmitted symbol.

[0090] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative labor.

[0091] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A signal processing method for a non-cellular large-scale multiple-input multiple-output system, characterized in that: include: Acquire a first Gaussian parameter transmitted from a factor node to a variable node; wherein the factor node is determined according to a received signal of an access point in the non-cellular large-scale multi-input multi-output system, and the variable node is determined according to a transmitted symbol of the non-cellular large-scale multi-input multi-output system; the first Gaussian parameter is calculated by each of the access points according to an initialization Gaussian parameter, a received signal and channel information; Determine the off-target information according to the first Gaussian parameter, and calculate the second Gaussian parameter transmitted from the variable node to the factor node according to the off-target information; According to the second Gaussian parameter, the first Gaussian parameter is updated, and the extrinsic information is re-determined, thereby updating the second Gaussian parameter until a preset iteration condition is met to obtain a target transmission symbol.

2. The signal processing method for a non-cellular large-scale multiple-input multiple-output system according to claim 1, characterized in that: According to the second Gaussian parameter, the first Gaussian parameter is updated, and the extrinsic information is re-determined, thereby updating the second Gaussian parameter until a preset iteration condition is met to obtain a target transmission symbol, including: For each of the second Gaussian parameters, sending the second Gaussian parameter back to each of the access points, so that each of the access points updates the first Gaussian parameter according to the received second Gaussian parameter; Re-determining the off-target information according to the updated first Gaussian parameters, and updating the second Gaussian parameters according to the re-determined off-target information; When a preset iteration stop condition is met, stopping updating the second Gaussian parameter; According to the first target Gaussian parameter, the posterior probability of the transmitted symbol is calculated to obtain the target transmitted symbol; wherein the first target Gaussian parameter is a first Gaussian parameter obtained when a preset iteration stop condition is met.

3. The signal processing method for a non-cellular large-scale multiple-input multiple-output system according to claim 2, characterized in that: Calculating a second Gaussian parameter transmitted from the variable node to the factor node according to the off-target information includes: Calculate the posterior probability of each of the transmitted symbols according to the mean and variance of the off-target information of each of the variable nodes corresponding to the transmitted symbols and the prior probability information of the transmitted symbols; According to the mean and variance of each of the posterior probabilities, a second Gaussian parameter from the variable node to the factor node is determined.

4. The signal processing method for a non-cellular large-scale multiple-input multiple-output system according to claim 1, characterized in that: The first Gaussian parameters include: a mean and a variance of a message transmitted from the factor node to the variable node.

5. The signal processing method for a non-cellular large-scale multiple-input multiple-output system according to claim 1, characterized in that: The off-target information includes: a weighted average of the means of messages transmitted from multiple factor nodes to variable nodes, and a weighted average of the variances of messages transmitted from multiple factor nodes to variable nodes.

6. The signal processing method for a non-cellular large-scale multiple-input multiple-output system according to claim 5, characterized in that: The calculation method of the second Gaussian parameter specifically includes: Based on the mean and variance of the probability distribution from the factor node to the variable node, the mean and variance of the approximate posterior probability distribution of each user are obtained; According to the variance of the approximate posterior probability and the variance of the message transmitted from the factor node to the variable node, updating the variance of the message transmitted from the variable node to the factor node; According to the mean of the approximate posterior probability, the variance of the approximate posterior probability, and the variance and mean of the message transmitted from the factor node to the variable node, the mean of the message transmitted from the variable node to the factor node is updated; The second Gaussian parameter is obtained according to the updated mean and variance of the message transmitted from the variable node to the factor node.

7. A non-cellular large-scale multiple-input multiple-output system signal processing device, characterized in that: include: An acquisition module is used to acquire a first Gaussian parameter transmitted from a factor node to a variable node; wherein the factor node is determined according to a received signal of an access point in the non-cellular large-scale multi-input multi-output system, and the variable node is determined according to a transmitted symbol of the non-cellular large-scale multi-input multi-output system; and the first Gaussian parameter is calculated by each of the access points according to an initialization Gaussian parameter, a received signal and channel information; a calculation module, configured to determine off-target information according to the first Gaussian parameter, and calculate a second Gaussian parameter transmitted from the variable node to the factor node according to the off-target information; The updating module is used to update the first Gaussian parameter according to the second Gaussian parameter, and re-determine the external information, thereby updating the second Gaussian parameter until a preset iteration condition is met to obtain a target transmission symbol.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the signal processing method for a non-cellular large-scale multiple-input multiple-output system as claimed in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the signal processing method for a non-cellular massive multiple-input multiple-output system as claimed in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the signal processing method for a non-cellular massive multiple-input multiple-output system as claimed in any one of claims 1 to 6 is implemented.

Images