A method and system for joint parameter estimation and signal reconstruction under distributed antenna system

Abstract

The application discloses a kind of method and system of joint parameter estimation and signal reconstruction under distributed antenna system, method includes: S1: constructing the generalized linear model of parameter estimation and signal reconstruction under distributed antenna system;S2: noise variance parameter and signal estimation value initialization, set iteration stop condition and obtain the input value required by preset algorithm;S3: respectively obtain the input parameter required by the noise variance estimator of each cluster antenna end and the approximate posterior probability of signal;And the estimated value of signal is solved;S4: the noise variance parameter corresponding to signal is estimated independently by minimizing variational bet free energy;S5: whether the preset iteration stop condition is reached, if yes, then flow to step S6;If not, then return to step S3 and carry out next round iteration;S6: iteration ends, and the estimated value of signal and the estimated value of corresponding noise variance of each cluster antenna end are output.The application realizes the joint estimation of noise variance and signal under different environments, and improves signal reconstruction performance.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of signal reconstruction, and more particularly to a method and system for joint parameter estimation and signal reconstruction in a distributed antenna system. BACKGROUND

[0002] A distributed antenna system is a network composed of spatially separated antenna nodes connected to a common source through a transmission medium, providing wireless services within a geographical area or structure. In the rapid development of 5G technology, the distributed antenna system, as a highly competitive technology, effectively improves system capacity and reliability, reduces interference cancellation complexity, and becomes an indispensable key technology without additional spectrum resources and transmission power.

[0003] The uplink massive MIMO communication with low-precision analog-to-digital converters (ADCs) in a distributed antenna system can be abstracted as a generalized linear model. When using an approximate Bayesian inference algorithm to reconstruct the signal in the model, the covariance matrix of the noise needs to be known to obtain a high accuracy. However, the currently good algorithm can only reconstruct the signal for the standard linear model with unknown noise covariance matrix, and if the requirement is not met, the estimated value of the signal will be far from the true value, which cannot meet the basic requirements of signal transmission.

[0004] In the prior art, an uplink channel estimation method for a massive MIMO system is disclosed, which includes the following steps: (1) using a Gaussian mixture model to model the probability model of the channel; (2) using optimal Bayesian parameter estimation for channel estimation; (3) using a hierarchical clustering algorithm to give an initial value of iteration; (4) using an approximate message passing algorithm to solve the edge probability density function in step two; (5) using an expectation-maximization algorithm to iteratively solve the parameters of the Gaussian mixture model. This method is mainly applied to the channel estimation of the uplink of a massive MIMO system, and it can use the approximate message passing algorithm and the expectation-maximization algorithm in the channel gain probability model to estimate the prior parameters and the channel. However, this method is not applicable to the generalized linear model under ADC, and it cannot estimate the parameters (noise covariance matrix) in the likelihood function. Therefore, in the implementation process, this method cannot effectively estimate the noise variance and the signal in the distributed antenna system. SUMMARY

[0005] The present application provides a method and system for joint parameter estimation and signal reconstruction in a distributed antenna system, which solves the problem of poor signal reconstruction performance due to inaccurate noise variance parameter estimation in the current distributed antenna system.

[0006] The primary object of the present application is to solve the above technical problems, and the technical scheme of the present application is as follows:

[0007] The first aspect of the present application provides a method for joint parameter estimation and signal reconstruction under a distributed antenna system, comprising the following steps:

[0008] S1: constructing a generalized linear model for parameter estimation and signal reconstruction under a distributed antenna system;

[0009] S2: initializing noise variance parameters and signal estimation values in the model, setting iteration stop conditions and obtaining input values required by a preset algorithm, and the preset algorithm is an expectation maximization algorithm and a distributed generalized expectation propagation algorithm;

[0010] S3: according to the required input parameters and the factor graph corresponding to the model, using the distributed generalized expectation propagation algorithm to obtain the approximate posterior probability of the signal and the input parameters required by the noise variance estimator of each cluster antenna end under the preset number of iterations; according to the approximate posterior probability of the signal, using a minimum mean square error estimator to obtain the estimation value of the signal;

[0011] S4: the noise variance estimator of each cluster antenna end estimates the noise variance parameters corresponding to the signal independently by minimizing the variational free energy as the objective function according to the expectation maximization algorithm;

[0012] S5: determining whether the preset iteration stop condition is reached, if yes, proceeding to step S6; if no, returning to step S3 for the next iteration;

[0013] S6: the iteration is ended, and the estimation value of the signal and the estimation value of the noise variance corresponding to each cluster antenna end are output.

[0014] Further, in the factor graph, the transition probability corresponds to the factor node, if it is in the iteration stage of the distributed generalized expectation propagation algorithm, when the number of iterations is less than or equal to the preset number of iterations, the value of in the transition probability is equal to the estimation value of the noise variance in the last time The approximate posterior probability corresponding to the factor node is represented as If it is in the noise variance estimation stage, the transition probability in the transition probability exists as an unknown parameter, and a new noise variance estimation value is generated after step S4, wherein represents the received signal of the i-th cluster antenna, and is a random variable.

[0015] Furthermore, by relaxing and approximating the constraints of the E-step and M-step of the expectation-maximization algorithm, the solution obtained in the E-step can be linked to the fixed point generated by the distributed generalized expectation propagation algorithm in S3 during factor graph iteration. The result calculated in step S3 is used as the solution of the E-step and the input of the M-step. The M-step automatically updates the noise variance parameter by minimizing the variational Bethesda free energy, i.e., step S4, as follows:

[0016]

[0017] This represents the new noise variance estimate; and This indicates that the output approximate posterior probability is generated in step S3. The mean and variance; Represents the transition probability The scalar form of the m-th position in the scalar; Represented as a random variable; and They are vectors and The m-th scalar value.

[0018] Furthermore, in step S2, the noise variance parameter and signal estimate are initialized according to the estimation accuracy and computational complexity required by the actual distributed antenna system.

[0019] Furthermore, in step S3, the predetermined number of iterations of the distributed generalized expectation propagation algorithm is adjusted to suit the estimation accuracy and computational complexity required by the actual distributed antenna system.

[0020] Furthermore, the first Uplink massive MIMO communication using cluster antennas can be abstracted into a generalized linear model, expressed as:

[0021]

[0022] in, For the signal to be estimated, Q c (·) is a low-precision quantizer for complex numbers, and L is the total number of antenna clusters in the distributed antenna system; and They are the first The received signal and channel matrix of the cluster antenna. Then it is the first Additive white Gaussian noise corresponding to cluster antennas;

[0023] The uplink massive MIMO communication of a distributed antenna system can be abstracted into a generalized linear model, expressed as:

[0024] y = Qc (Ax+w)

[0025] wherein and

[0026] Further, each cluster of antennas in the distributed antenna system has different environmental noise variance due to different distribution locations, and thus the noise variance estimators of each cluster of antennas need to independently perform parameter estimation, and the probability density function of the random variable is wherein and I are unknown noise variance parameters and unit matrix, respectively, is the transition probability caused by the quantizer, and the expression is:

[0027]

[0028]

[0029] wherein

[0030] Further, when the variational free energy is minimized by a second-order optimization method, if the estimated noise variance deviates seriously from the normal value range of the noise variance in the actual environment, the noise variance is forced to be , i.e., the noise variance estimation value of the last time is retained.

[0031] Further, in step S3, if an individual antenna end computing unit cannot transmit computing data to the central processing unit due to failure, the performance of the system will not be obviously affected, and good robustness is achieved.

[0032] The second aspect of the present application provides a joint parameter estimation and signal reconstruction system under a distributed antenna system, which comprises a memory and a processor, wherein the memory comprises a joint parameter estimation and signal reconstruction method program under a distributed antenna system, and the joint parameter estimation and signal reconstruction method program under a distributed antenna system is executed by the processor to realize the following steps:

[0033] S1: constructing a generalized linear model of parameter estimation and signal reconstruction under a distributed antenna system;

[0034] S2: initializing noise variance parameters and signal estimation values in the model, setting an iteration stop condition and obtaining input values required by a preset algorithm, and the preset algorithm is an expectation maximization algorithm and a distributed generalized expectation propagation algorithm;

[0035] ​​S3: according to the required input parameters and the factor graph corresponding to the model, the approximate posterior probability of the signal and the input parameters required by the noise variance estimator of each cluster antenna end are obtained by using the distributed generalized expectation propagation algorithm under the preset number of iterations; according to the approximate posterior probability of the signal, the estimated value of the signal is obtained by using the minimum mean square error estimator;

[0036] S4: the noise variance estimator of each cluster antenna end estimates the noise variance parameter corresponding to the signal independently by taking the variational free energy as the objective function and minimizing the variational free energy according to the expectation maximum algorithm;

[0037] S5: whether the preset iteration stopping condition is reached is judged, if yes, the flow is transferred to step S6; if no, the next round of iteration is returned to step S3;

[0038] S6: the iteration is ended, and the estimated value of the signal and the estimated value of the noise variance corresponding to each cluster antenna end are output.

[0039] Compared with the prior art, the beneficial effects of the technical scheme of the present application are:

[0040] The present application can realize the joint estimation of noise variance and signal under different environments by combining the distributed generalized expectation propagation algorithm and the maximum expectation algorithm for continuous iteration calculation, and has robustness under the condition that the channel matrix has strong correlation, and improves the signal reconstruction performance. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 The flow chart of the joint parameter estimation and signal reconstruction method of the distributed antenna system of the present application.

[0042] Figure 2 The generalized linear model schematic diagram of the distributed antenna system of the embodiment of the present application.

[0043] Figure 3 The comparison schematic diagram of the bit error rate and iteration relationship between the EM-DeGEC algorithm and the existing DeGEC algorithm under the heteroscedastic noise quantization model of the embodiment of the present application.

[0044] Figure 4 The comparison schematic diagram of the signal reconstruction and parameter estimation accuracy of the EM-DeGEC algorithm and the existing DeGEC algorithm under the channel matrix with strong correlation in different signal-to-noise ratios of the embodiment of the present application.

[0045] Figure 5 The comparison schematic diagram of the signal reconstruction accuracy of the EM-DeGEC algorithm and the existing DeGEC algorithm under different correlation channel matrices of the embodiment of the present application. DETAILED DESCRIPTION

[0046] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0047] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0048] Example 1

[0049] like Figure 1 As shown, the first aspect of the present invention provides a method for joint parameter estimation and signal reconstruction in a distributed antenna system, comprising the following steps:

[0050] S1: Construct a generalized linear model for parameter estimation and signal reconstruction in a distributed antenna system;

[0051] S2: Initialize the noise variance parameter and signal estimate in the model, set the iteration stopping condition, and obtain the input values ​​required by the preset algorithm. The preset algorithm is the expectation-maximization algorithm and the distributed generalized expectation propagation algorithm.

[0052] It should be noted that, in practical implementation, in order to reduce the number of algorithm iterations and better estimate the noise variance parameter, generally... Initialize to 0.

[0053] S3: Based on the required input parameters and the factor graph corresponding to the model, the distributed generalized expectation propagation algorithm is used to obtain the approximate posterior probability of the signal and the input parameters required by the noise variance estimator at each cluster antenna end under the preset number of iterations; based on the approximate posterior probability of the signal, the minimum mean square error estimator is used to obtain the estimated value of the signal.

[0054] It should be noted that, in practical implementation, if the noise variance estimator is to update its parameters in real time to achieve better estimation accuracy, the preset number of iterations K should be reduced. MA It is generally set to 1; however, to reduce the overall computational complexity of the algorithm and shorten the computation time, the preset number of iterations K should be appropriately increased. MAx It is generally set to 3. Based on the approximate posterior probability of the signal, the estimated value of the signal is obtained using the minimum mean square error estimator;

[0055] S4: The noise variance estimator at each cluster antenna end estimates the noise variance parameter corresponding to the signal independently by minimizing the variational BET free energy, based on the expectation-maximization algorithm and with the variational BET free energy as the objective function.

[0056] S5: judging whether a preset iteration stopping condition is reached, if yes, proceeding to step S6; if no, returning to step S3 for next iteration;

[0057] S6: iteration is ended, and an estimated value of the signal and an estimated value of the noise variance corresponding to each cluster antenna end are outputted.

[0058] It should be noted that in the specific implementation process, the EM-DeGEC algorithm formed by combining the distributed generalized expectation propagation algorithm DeGEC and the expectation maximum algorithm EM can realize the joint estimation of the parameters and the signal, and the algorithm can still work normally and has robustness even under the condition that the channel matrix has strong correlation.

[0059] Further, in the factor graph, the transition probability corresponding to the factor node, if in the iteration stage (step S3) of the distributed generalized expectation propagation algorithm, when the iteration number is less than or equal to the preset iteration number, the transition probability , the value of equals the estimated value of the noise variance in the last time The approximate posterior probability corresponding to the factor node is expressed as If in the noise variance estimation stage, the transition probability , the value of exists as an unknown parameter, and a new noise variance estimation value is generated through step S4, wherein represents the received signal of the jth cluster antenna, and is a random variable.

[0060] Further, by relaxing and approximating the constraint terms of the E step and the M step of the expectation maximum algorithm, the solution obtained by the E step can be linked to the fixed point generated by the iteration of the factor graph in the distributed generalized expectation propagation algorithm in step S3, and the result calculated in step S3 is used as the solution of the E step and the input of the M step. The M step realizes the automatic update of the noise variance parameter by minimizing the variational free energy, that is, step S4, which is expressed as follows:

[0061]

[0062]

[0063] represents a new noise variance estimation value; and represent the mean and variance of the output approximate posterior probability in step S3; represents the transition probability the mth element of the vector is expressed as a random variable and are the mth element of the vector and are the mth element of the vector

[0064] Further, in step S2, the noise variance parameter and the signal estimation value are initialized according to the estimation accuracy and the calculation complexity required by the actual distributed antenna system.

[0065] Further, in step S3, the predetermined number of iterations of the adjusted distributed generalized expectation propagation algorithm is adapted to the estimation accuracy and the calculation complexity required by the actual distributed antenna system.

[0066] In the specific implementation process, the initialization parameters and the setting of the predetermined number of iterations can be flexibly adjusted according to the requirements of the distributed antenna system, so as to realize the relative balance between the performance and the calculation complexity of the system.

[0067] Further, in the actual distributed antenna system, as shown in Figure 2 , a central processing unit is connected with a plurality of antenna-end computing units, and the uplink massive MIMO communication of the first cluster of antennas can be abstracted as a generalized linear model, and the expression is:

[0068]

[0069] wherein, is the signal to be estimated, Q c (·) is a low-precision quantizer of a complex number, and L is the total number of clusters of antennas in the distributed antenna system; and are the received signal and the channel matrix of the first cluster of antennas, respectively, is the additive white Gaussian noise corresponding to the first cluster of antennas;

[0070] The uplink massive MIMO communication of the distributed antenna system is abstracted as a generalized linear model, and the expression is:

[0071] y=Q c (Ax+w)

[0072] wherein and

[0073] Further, due to the different distribution positions of the clusters of antennas in the distributed antenna system, the corresponding environmental noise variances are also different, and therefore the noise variance estimators of the cluster antenna-end computing units need to independently perform parameter estimation, and the random variable The probability density function is in I and I are the unknown noise variance parameter and identity matrix, respectively. It is the transition probability caused by the quantizer, expressed as:

[0074]

[0075] in

[0076] Furthermore, by minimizing the variational Bethesda free energy using a second-order optimization method, the estimation speed of the noise variance parameter by the computational units at each cluster antenna end is greatly improved, processing the variational Bethesda free energy... When dealing with an optimization problem with an objective function, if the estimated noise variance... If the value deviates significantly from the normal range of noise variance in the actual environment, then it is forced to... That is, retain the previous noise variance estimate.

[0077] Furthermore, in step S3, if an individual antenna-end computing unit is unable to transmit computing data to the central processing unit due to a fault, it will not have a significant impact on the system performance and has good robustness.

[0078] In practice, the optimization problem handling method in step S4 can be adjusted to achieve a balance between computation time and hardware cost. Using a second-order optimization method can speed up the estimation of noise variance, but it will increase computational complexity and place certain demands on the hardware performance of the estimator.

[0079] Example 2

[0080] A second aspect of the present invention provides a joint parameter estimation and signal reconstruction system for a distributed antenna system. The system includes a memory and a processor. The memory includes a method program for joint parameter estimation and signal reconstruction in a distributed antenna system. When executed by the processor, the method program performs the following steps:

[0081] S1: Construct a generalized linear model for parameter estimation and signal reconstruction in a distributed antenna system;

[0082] S2: Initialize the noise variance parameter and signal estimate in the model, set the iteration stopping condition, and obtain the input values ​​required by the preset algorithm. The preset algorithm is the expectation-maximization algorithm and the distributed generalized expectation propagation algorithm.

[0083] S3: According to the required input parameters and the factor graph corresponding to the model, the approximate posterior probability of the signal and the input parameters required by the noise variance estimator of each cluster antenna end are obtained by using the distributed generalized expectation propagation algorithm under the preset number of iterations; according to the approximate posterior probability of the signal, the estimated value of the signal is obtained by using the minimum mean square error estimator;

[0084] S4: The noise variance estimator of each cluster antenna end estimates the noise variance parameter corresponding to the signal by minimizing the variational free energy as the objective function according to the expectation maximization algorithm;

[0085] S5: Determine whether the preset iteration stopping condition is reached, if yes, proceed to step S6; if no, return to step S3 for the next iteration;

[0086] S6: The iteration is ended, and the estimated value of the signal and the estimated value of the noise variance corresponding to each cluster antenna end are output.

[0087] Embodiment 3

[0088] This embodiment specifically explains the pseudo code of the EM-DeGEC algorithm.

[0089] The pseudo code of EM-DeGEC is as follows:

[0090]

[0091] The pseudo code of DeGEC is as follows:

[0092]

[0093]

[0094] wherein, and Var[·] represent the calculation of expectation and variance respectively, and ⊙ represents point multiplication calculation, represents point division calculation, d(·) represents the diagonal elements of the matrix (·) and forms a vector, and Diag(·) represents the diagonalization of the vector (·). The b x defined in the DeGEC algorithm is and are the approximate posterior probabilities of x and

[0095] The initial value of the parameter in the EM-DeGEC algorithm depends on the performance requirements of the given distributed antenna system. If the system hardware performance is good and the operation of higher complexity can be accepted, the preset iteration number K MAX of the DeGEC algorithm can be reduced, and a second-order optimization method is used to solve the optimization problem in the fifth row of EM-DeGEC; otherwise, K MAX can be increased, and a first-order optimization method is used.

[0096] The iteration process of the EM-DeGEC algorithm is first performed from the DeGEC algorithm part, at this time the noise variance is the value taken out from the noise variance estimator last time, if it is the first iteration of the EM-DeGEC algorithm, then is the value given by the initialization. After the DeGEC algorithm is executed, the output and are taken as the input of the noise variance estimator, and the new noise variance estimation value is obtained by minimizing the variational Bethe free energy, so as to complete a round of iteration.

[0097] Figure 2 is the factor graph of the EM-DeGEC algorithm of the embodiment. In the embodiment, there is a distributed antenna system which places 8 antenna end computing units in areas with different noise variances, and each antenna end computing unit has 250 antennas. Under distributed MIMO communication, the mobile terminal device transmits a power normalized 32QAM modulation signal, and the distributed antenna system uses the EM-DeGEC algorithm to reconstruct the received observation signal. As shown in Figure 3 , under the condition that the channel matrix has strong correlation, the EM-DeGEC algorithm can still accurately estimate the noise variance parameter and the transmission signal, and the accuracy of the signal estimation value can reach the running level of the DeGEC algorithm (noise variance known). As shown in Figure 4 , when the signal-to-noise ratio is in the normal variation range, the EM-DeGEC algorithm can still work well, and its performance is comparable to that of the DeGEC algorithm (noise variance known). As shown in Figure 5 , the stronger the correlation of the channel matrix, the worse the performance of the EM-DeGEC algorithm, but it can still reach the level of the DeGEC algorithm (noise variance known), and compared with the EM-GAMP algorithm, it cannot work normally. According to the test results, it can be known that the application of the EM-DeGEC algorithm in the distributed antenna system can effectively perform parameter estimation and signal reconstruction.

[0098] Obviously, the above embodiments of the present application are only examples for clearly illustrating the present application, and are not intended to limit the embodiments of the present application. For those skilled in the art, other different forms of changes or modifications can be made on the basis of the above description. Here, it is not necessary and impossible to enumerate all the embodiments. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the claims of the present application.

Claims

1. A method for joint parameter estimation and signal reconstruction in a distributed antenna system, characterized in that, Includes the following steps: S1: Construct a generalized linear model for parameter estimation and signal reconstruction in a distributed antenna system; S2: Initialize the noise variance parameter and signal estimate in the model, set the iteration stopping condition and obtain the input value required by the preset algorithm. The preset algorithm is an algorithm formed by combining the improved expectation-maximization algorithm and the distributed generalized expectation propagation algorithm. S3: Based on the required input parameters and the factor graph corresponding to the model, the distributed generalized expectation propagation algorithm is used to obtain the approximate posterior probability of the signal and the input parameters required by the noise variance estimator at each cluster antenna end under the preset number of iterations; based on the approximate posterior probability of the signal, the minimum mean square error estimator is used to obtain the estimated value of the signal. S4: The noise variance estimator at each cluster antenna end estimates the noise variance parameter corresponding to the signal independently by minimizing the variational BET free energy, based on the expectation-maximization algorithm and with the variational BET free energy as the objective function. S5: Determine whether the preset iteration stop condition has been met. If yes, proceed to step S6; otherwise, return to step S3 for the next iteration. S6: The iteration ends, and the estimated values ​​of the output signal and the corresponding noise variance of each cluster antenna are obtained.

2. The method for joint parameter estimation and signal reconstruction in a distributed antenna system according to claim 1, characterized in that, In the factor graph, the transition probability p(y) l ∣z l ;σ l If the factor node corresponding to y is in the iterative stage of the distributed generalized expectation propagation algorithm, and the number of iterations is less than or equal to the preset number of iterations, then the transition probability p(y) is... l ∣z l ;σ l σ in ) l The value of is equal to the previous estimate of the noise variance. The approximate posterior probability corresponding to this factor node is expressed as: If we are in the noise variance estimation stage, then the transition probability p(y) l ∣z l ;σ l σ in ) l As an unknown parameter, after step S4, a new noise variance estimate is generated, where y l Let z represent the received signal of the l-th antenna cluster, and z l =A l x is a random variable.

3. The method for joint parameter estimation and signal reconstruction in a distributed antenna system according to claim 1, characterized in that, By relaxing and approximating the constraints of the E-step and M-step of the expectation-maximization algorithm, the solution obtained in the E-step can be linked to the fixed point generated by the distributed generalized expectation propagation algorithm in S3 during factor graph iteration. The result calculated in step S3 is used as the solution of the E-step and the input of the M-step. The M-step automatically updates the noise variance parameter by minimizing the variational Bethesda free energy, i.e., step S4, as follows: This represents the new noise variance estimate; and This indicates that the output approximate posterior probability is generated in step S3. The mean and variance of p(y). lm |Z lm ;σ l ) represents the transition probability p(y) l ∣z l ;σ l The scalar form of the m-th bit in ); z lm Represented as a random variable; and They are vectors and The m-th scalar value.

4. The method for joint parameter estimation and signal reconstruction in a distributed antenna system according to claim 1, characterized in that, In step S2, the noise variance parameter and signal estimate are initialized according to the estimation accuracy and computational complexity required by the actual distributed antenna system.

5. The method for joint parameter estimation and signal reconstruction in a distributed antenna system according to claim 1, characterized in that, In step S3, the predetermined number of iterations of the distributed generalized expectation propagation algorithm is adjusted to suit the estimation accuracy and computational complexity required by the actual distributed antenna system.

6. The method for joint parameter estimation and signal reconstruction in a distributed antenna system according to claim 1, characterized in that, The uplink massive MIMO communication of the l-th cluster of antennas can be abstracted into a generalized linear model, expressed as: y l =Q c (A l x+w l ),l=1,2,…,L z l =A l x in, For the signal to be estimated, Q c (·) is a low-precision quantizer for complex numbers, and L is the total number of antenna clusters in the distributed antenna system; and These are the received signal and channel matrix of the l-th antenna cluster, respectively, w l This is the additive white Gaussian noise corresponding to the l-th cluster of antennas; The uplink massive MIMO communication of a distributed antenna system can be abstracted into a generalized linear model, expressed as: y=Q c (Ax+w) in and 7. The method for joint parameter estimation and signal reconstruction in a distributed antenna system according to claim 6, characterized in that, The environmental noise variance σ varies among the antenna clusters in a distributed antenna system due to their different locations. l They are also different, therefore the noise variance estimator of each cluster antenna terminal computing unit needs to independently perform parameter estimation, and the random variable w l The probability density function is Where σ l and I are the unknown noise variance parameter and identity matrix, respectively, p(y l ∣z l ;σ l The transition probability is caused by the quantizer, and its expression is: in 8. The method for joint parameter estimation and signal reconstruction in a distributed antenna system according to claim 1, characterized in that, The variational Betheska free energy is minimized using a second-order optimization method, and the variational Betheska free energy is then processed. When dealing with an optimization problem with an objective function, if the estimated noise variance... If the value deviates significantly from the normal range of noise variance in the actual environment, then it is forced to... That is, retain the previous noise variance estimate.

9. The method for joint parameter estimation and signal reconstruction in a distributed antenna system according to claim 1, characterized in that, In step S3, if an individual antenna-end computing unit is unable to transmit computing data to the central processing unit due to a fault, it will not have a significant impact on the system performance and has good robustness.

10. A joint parameter estimation and signal reconstruction system for a distributed antenna system, characterized in that, The system includes a memory and a processor. The memory contains a method program for joint parameter estimation and signal reconstruction in a distributed antenna system. When the processor executes the method program for joint parameter estimation and signal reconstruction in a distributed antenna system, it performs the following steps: S1: Construct a generalized linear model for parameter estimation and signal reconstruction in a distributed antenna system; S2: Initialize the noise variance parameter and signal estimate in the model, set the iteration stopping condition, and obtain the input values ​​required by the preset algorithm. The preset algorithm is the expectation-maximization algorithm and the distributed generalized expectation propagation algorithm. S3: Based on the required input parameters and the factor graph corresponding to the model, the distributed generalized expectation propagation algorithm is used to obtain the approximate posterior probability of the signal and the input parameters required by the noise variance estimator at each cluster antenna end under the preset number of iterations; based on the approximate posterior probability of the signal, the minimum mean square error estimator is used to obtain the estimated value of the signal. S4: The noise variance estimator at each cluster antenna end estimates the noise variance parameter corresponding to the signal independently by minimizing the variational BET free energy, based on the expectation-maximization algorithm and with the variational BET free energy as the objective function. S5: Determine whether the preset iteration stop condition has been met. If yes, proceed to step S6; otherwise, return to step S3 for the next iteration. S6: The iteration ends, and the estimated values ​​of the output signal and the corresponding noise variance of each cluster antenna are obtained.

Images