Communication signal detection method for terahertz single-bit quantization of multiple analog-to-digital converter branches

Abstract

The invention provides a multi-analog-to-digital converter branch terahertz single-bit quantization communication signal detection method, which relates to the technical field of communication, and comprises the following steps: S1, constructing a terahertz single-bit quantization communication system with a plurality of analog-to-digital converter branches, each analog-to-digital converter branch comprises a phase rotation module and an analog-to-digital converter which are connected in sequence, the phase rotation modules of different analog-to-digital converter branches have different phase rotation angles; s2, representing a communication system by using a vector value factor graph; and S3, performing vector approximate message passing on the vector value factor graph, iteratively calibrating the estimation of the noise-free receiving signal, and finally realizing the detection of the sending symbol. According to the invention, reliable signal detection can be realized under the influence of single-bit quantization distortion, and the detection performance is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of communication technology, in particular to a multi-ADC branch terahertz single-bit quantization communication signal detection method. BACKGROUND

[0002] With the development of wireless communication, the number of users accessing and the data transmission required by wireless communication services are growing at an exponential rate. According to statistics, people's demand for wireless communication bandwidth will double every 18 months. However, in the current lower frequency range commonly used in mobile communication technology, it is still difficult to achieve a data transmission rate of more than several Gbps, even if a variety of complex modulation and multiplexing techniques are applied. Therefore, in order to meet the higher demand of mobile communication development for wireless communication speed, more spectrum resources in the higher frequency range must be developed and utilized.

[0003] Terahertz (THz) communication technology utilizes the rich bandwidth resources brought by high frequency bands, which can support information transmission requirements exceeding 10 12 bit per second level. However, one of the major bottlenecks of this technology is the complexity and power consumption of the receiver. The receiving end of the wireless communication system, especially the Analog-to-Digital Converter (ADC) module, the high-precision ADC under the THz frequency band will introduce unacceptable complexity and power consumption. In order to solve this problem, researchers are exploring the use of low-precision quantization ADC in THz communication systems. This requires the THz system to be able to achieve reliable signal detection under the influence of severe nonlinear quantization distortion. SUMMARY

[0004] The main purpose of the present application is to provide a multi-ADC branch terahertz single-bit quantization communication signal detection method, which can achieve reliable signal detection under the influence of single-bit quantization distortion and effectively improve the detection performance.

[0005] The present application is realized by the following technical solutions: The multi-ADC branch terahertz single-bit quantization communication signal detection method comprises the following steps: Step S1, a terahertz single-bit quantization communication system with multiple ADC branches is constructed, the ADC branch comprises a phase rotation module and an ADC connected in sequence, and the phase rotation modules of different ADC branches have different phase rotation angles, wherein the communication system is a multi-user multiple-input multiple-output system; Step S2, the communication system is represented by a vector-valued factor graph; Step S3: Perform vector approximation message passing on the vector value factor graph, iteratively calibrate the estimation of the noiseless received signal, and finally realize the detection of the transmitted symbol. During the message passing process, the single-bit sampling information of different analog-to-digital converter branches is aggregated.

[0006] Furthermore, in step S1, the received signal model in the communication system is represented as follows: The vector composed of the noiseless received signals at all receiving antennas is represented as ,in, The single-bit quantization process implemented in the analog-to-digital converter branch. This is a vector composed of the phase rotation angles of the phase rotation modules in each analog-to-digital converter branch. It is the identity matrix. The equivalent channel matrix for terahertz block fading baseband is derived from the center frequency of the communication system, the locational relationship between the base station and each user, and the surrounding scattering environment. Symbols are transmitted to all users, and these transmitted symbols satisfy a discrete uniform distribution on the modulation constellation diagram used by the communication system. It is a zero-mean Gaussian white noise matrix.

[0007] Furthermore, in step S2, in the vector value factor graph, all users send symbols The vector consisting of the noise-free received signals at all receiving antennas Let and represent the two variable nodes of the vector-valued factor graph. Each factor obtained by decomposing the joint probability density function of the communication system according to different vector parameters using Bayes' theorem is represented as a factor node in the vector-valued factor graph. The factor nodes include the conditional probability distribution of the noise-free received signal relative to the transmitted symbol and the channel matrix. The conditional probability distribution of single-bit sampled information of the analog-to-digital converter branch relative to the noise-free received signal. and the prior distribution of transmitted symbols .

[0008] Furthermore, in step S2, in Corresponding factor nodes and When the corresponding variable nodes pass messages, a phase alignment operation is set up with the phase difference between different analog-to-digital converter branches as a parameter to aggregate single-bit sampling information from different analog-to-digital converter branches.

[0009] Furthermore, in step S3, the vector approximation message passing on the vector value factor graph specifically includes the following steps: Step S31, Initialization: Initialize the two variable nodes and the messages between factor nodes in the vector value factor graph; Step S32, Linearity Detection: Based on the initialization or the two variable nodes from the previous iteration, send the data to the factor node. The message, using the product formula of the Gaussian distribution of vector values, calculates... A linear posterior estimate and A linear posterior estimate ; Step S33: Calculate the likelihood message based on the message passing rules on the factor graph: The result obtained in step S32... A linear posterior estimate Divide by The prior distribution of the linear detection in this round is used to obtain the likelihood message of the transmitted symbol, and this likelihood message is used as the linear detection pair for the transmitted symbol. The basis for iterative correction of the posterior estimate; using the calculation rule of division of complex Gaussian distribution, the result obtained in step S32 is... A linear posterior estimate Divide by Prior distribution of linear detection in this round of iteration This yields the likelihood message used to calibrate the estimate of the noiseless received signal. The likely message This serves as the basis for iterative calibration of the estimation of the noiseless received signal using linear detection; Step S34, Update Posterior estimation: Using a scalar Gaussian denoiser, prior information about the transmitted symbol modulation constellation is incorporated to calculate... Posterior estimated mean and the corresponding posterior estimate of scalar variance and will Likelihood messages are removed according to the calculation rules of the complex Gaussian distribution. To obtain the variable node corresponding to the sent symbol The corresponding factor node message, i.e., the next round of iterative linear detection. The prior distribution; Step S35, Update Posterior estimation: Sampling information from each analog-to-digital converter branch is aggregated through phase alignment, and Bayes' theorem is applied to update the linear detection for the next iteration. The prior distribution; Step S36: Repeat steps S32 to S35 until the number of iterations reaches the set maximum, or the user sends a symbol. If the estimated variance is below a set threshold, then vector approximation message passing stops. posterior estimation For output, based on Hard decision on the Euclidean distance of each constellation point in the modulation constellation to output the signal detection result.

[0010] Further, in the step S32, the linear posterior estimate , is the mean of the linear posterior estimate, is the mean of the variance of the linear posterior estimate, wherein denotes a Gaussian distribution in the complex domain, denotes the mean of the X-prior for linear detection, denotes the variance of the X-prior for linear detection, denotes the mean of the noiseless received signal likelihood estimate for linear detection, denotes the variance of the noiseless received signal likelihood estimate, denotes the conjugate transpose of denotes the identity matrix, denotes the mean of the diagonal elements of a matrix; the linear posterior estimate , is the mean of the linear posterior estimate, is the mean of the variance of the linear posterior estimate, wherein denotes the number of single-antenna users of the transmitted signal, denotes the number of receive antennas of the base station; in the step S33, the likelihood message , .

[0011] Further, in the step S34, the , , , wherein denotes the mean, denotes the mean of the message to the node, i.e. the mean of the likelihood estimate of given by the linear detection, denotes the variance of the message to the node, i.e. the variance of the likelihood estimate of given by the linear detection, denotes the variance, denotes a proportionality, i.e. the formula on the right-hand side omits the constant factor of the exponential term of the distribution e.

[0012] Further, in the step S35, the applied Bayes formula is , denotes the posterior distribution of the noiseless received signal with respect to the single-bit sample information,​ This represents the likelihood distribution of the noiseless received signal relative to the single-bit sampled information. This represents the prior distribution of the noiseless received signal. This represents the single-bit sampled information of each analog-to-digital converter branch on all receiving antennas. This represents the joint probability distribution of single-bit sampled information from each analog-to-digital converter branch.

[0013] Furthermore, in step S35, the aggregation of sampling information from each analog-to-digital converter branch through phase alignment includes the following steps: Step S351: First calculate the posterior distribution corresponding to the first analog-to-digital converter branch. If the phase rotation angle of the first analog-to-digital converter branch is Then the mean of the posterior distribution is and variance Using the formulas respectively and Calculate, where, , When the quantization sampling result is less than 0, , When the quantization sampling result is greater than 0, , , This represents the single-bit sampled information of the first analog-to-digital converter branch on all receiving antennas. This represents the variance of the zero-mean Gaussian white noise at the receiving end of the communication system. The standard Gaussian probability density function is represented in The value at that location, This indicates that the standard Gaussian cumulative distribution function is in The value at; Step S352, Introduce the first One analog-to-digital converter branch, considering phase rotation When obtaining subsequent sampling information, it is necessary to update the posterior distribution. The mean is first multiplied by the phase alignment coefficient. , then calculate and Then multiply by the coefficient. To eliminate the influence of the phase alignment coefficient, i.e., to obtain up to the [number]th ... The posterior distribution of each analog-to-digital converter branch The mean and variance, This represents the single-bit sampled information of the k-th analog-to-digital converter branch on all receiving antennas; Step S353: Repeat step S352 until the sampling information of all analog-to-digital converter branches is taken into account, thus obtaining the posterior distribution of the noiseless received signal in the current iteration. the mean value of and variance .

[0014] From the above description of the present application, compared with the prior art, the present application has the following beneficial effects: The present application firstly constructs a terahertz single-bit quantization communication system with multiple analog-to-digital converter branches, then uses a vector-valued factor graph to represent the communication system and aggregate single-bit samples of different analog-to-digital converter branches, and finally implements vector approximate message passing on the vector-valued factor graph, iteratively calibrates the estimation of the noiseless received signal, to realize reliable detection of the transmitted symbol. The setting of multiple analog-to-digital converter branches and the phase rotation module of different analog-to-digital converter branches have different phase offsets of the information used for signal detection, which improves the amount of information used for signal detection of the communication system, and with the help of the factor graph and vector approximate message passing, the convergence of the detector can be guaranteed through the iterative process of step-by-step convergence, thereby overcoming quantization distortion, and the combination of the two can realize reliable signal detection under the influence of single-bit quantization distortion and effectively improve the detection performance. BRIEF DESCRIPTION OF DRAWINGS

[0015] The present application will be further described in detail below in combination with the drawings and specific embodiments.

[0016] Figure 1 is a flowchart of the present application.

[0017] Figure 2 is a structure block diagram of the communication system of the present application.

[0018] Figure 3 is a schematic diagram of the vector-valued factor graph model corresponding to the communication system of the present application.

[0019] Figure 4 is a flowchart of the vector approximate message passing implemented on the vector-valued factor graph of the present application.

[0020] Figure 5 is a flowchart of the phase alignment of the sampling information of each analog-to-digital converter branch of the present application.

[0021] Figure 6 is a detection performance comparison simulation diagram of the present application and the comparative scheme.

[0022] Figure 7 is a detection performance comparison simulation diagram of the present application under different parameter configurations. DETAILED DESCRIPTION

[0023] As shown in Figure 1 , the communication signal detection method of the terahertz single-bit quantization of the multiple analog-to-digital converter branches comprises the following steps: Step S1: Construct a terahertz single-bit quantization communication system with multiple analog-to-digital converter branches. Each analog-to-digital converter branch includes a phase rotation module and an analog-to-digital converter connected in sequence. The phase rotation modules of different analog-to-digital converter branches have different phase rotation angles. The communication system is a multi-user multiple-input multiple-output system. like Figure 2 As shown, the communication system includes N t A single-antenna user and a base station with multiple receiving antennas are configured, each antenna connected to a corresponding down-conversion low-pass filter module, each down-conversion low-pass filter module connected to a corresponding analog-to-digital converter (ADC) branch, and each ADC branch connected to a vector message passing detector. The down-conversion low-pass filter module is existing technology. In the communication system, the received signal model is represented as... The vector composed of the noiseless received signals at all receiving antennas is represented as ,in, The single-bit quantization process implemented in the analog-to-digital converter branch. The phase rotation angle of the phase rotation module in each analog-to-digital converter branch. The vector formed Given an identity matrix, according to the central limit theorem, the phase of the received signal in a practical system generally exhibits a uniform distribution. Two analog-to-digital converter (ADC) branches can be used, with phase rotation angles of 0° and π / 4 respectively. Alternatively, multiple ADC branches can be used. When there are K ADC branches, the... The phase rotation angle of each analog-to-digital converter branch is ; The equivalent channel matrix for terahertz (THz) block fading baseband is derived from the center frequency of the communication system, the location relationships between the base station and each user, and the surrounding scattering environment, and is represented as follows: , The exponential term represents the path loss corresponding to different signal propagation paths. The phase shift caused by the delay of different transmission paths to the transmitted symbols. The complex gain is the result of the combined effects of the angle of arrival, departure angle, and antenna gain for different paths. Symbols are transmitted to all users, and these transmitted symbols are in the modulation constellation diagram used by the communication system. The above satisfies a discrete uniform distribution. The matrix is ​​a zero-mean Gaussian white noise matrix with variance . .

[0024] Step S2: Represent the communication system using a vector value factor graph; The constructed multi-user multiple-input multiple-output system has a joint probability distribution. The joint probability distribution can be decomposed using Bayes' theorem as follows: , and the vector-valued message passing algorithm is implemented on the vector-valued factor graph. Figure 3 The vector-valued factor graph model is shown in the figure, and the vector-valued message passing algorithm is implemented on the vector-valued factor graph. The two identical variable nodes are represented as , and the equality of the two values is ensured by the Dirac delta distribution . The joint probability density function of the communication system is decomposed according to different vector parameters using the Bayesian formula, and each factor obtained is represented as a factor node of the vector-valued factor graph. The factor node includes the conditional probability distribution of the noiseless received signal with respect to the user transmitted symbol and the channel matrix , the conditional probability distribution of the single-bit sampling information of the analog-to-digital converter branch with respect to the noiseless received signal , and the prior distribution of the user transmitted symbol . Wherein, represents the single-bit sampling information of each analog-to-digital converter branch on all receiving antennas, is the single-bit sampling information of the th analog-to-digital converter branch on all receiving antennas.

[0025] In , a phase alignment operation with as a parameter is set when message passing is performed between the corresponding factor node and the variable node corresponding to , so as to aggregate the single-bit sampling information of different analog-to-digital converter branches in the message passing process.

[0026] Step S3, vector approximation message passing is implemented on the vector-valued factor graph, and the estimation of the noiseless received signal is iteratively calibrated to finally realize the detection of the transmitted symbol. The single-bit sampling information of different analog-to-digital converter branches is aggregated in the message passing process. Wherein, the vector approximation message passing implemented on the vector-valued factor graph specifically includes the steps as shown in Figure 4 . Step S31, initialization: the messages between the two variable nodes and the factor nodes of the vector-valued factor graph are initialized. Specifically, in the vector-valued factor graph, the distribution of the messages between the nodes is characterized by a mean vector and a scalar variance value. In this embodiment, the probability distribution of the variable node corresponding to the transmitted symbol on the factor graph is initialized using a mean vector of all 0 and a variance of 2, the probability distribution of the variable node corresponding to the noiseless received signal on the factor graph is initialized using a mean vector of all 0 and a variance of 2, and the messages between the vector-valued nodes are initialized using a probability distribution with a mean vector of all 0 and a variance of 1.

[0027] Step S32, Linearity Detection: Based on the initialization or the two variable nodes from the previous iteration, send the data to the factor node. The message, using the product formula of the Gaussian distribution of vector values, calculates... A linear posterior estimate and A linear posterior estimate ; Specifically, linear posterior estimation middle, Let be the mean of this linear posterior estimate. Let be the mean of the variance of this linear posterior estimate, where Describes the Gaussian distribution in the complex field. Indicates to The mean of the messages from the nodes, which is the mean of the X prior used for linear detection. This represents the variance of the message, specifically the variance of the X prior used for linear detection. Indicates to The node's message is the mean of the likelihood estimate of the noiseless received signal used for linear detection. Indicates to The variance of a node's messages represents the variance of the likelihood estimate of the noiseless received signal. express The conjugate transpose of . Represents the identity matrix. This indicates that the mean of the diagonal elements of the matrix is ​​calculated. Linear posterior estimation middle, Let be the mean of this linear posterior estimate. Let be the mean of the variance of this linear posterior estimate, where This indicates the number of users per antenna transmitting signals. This indicates the number of receiving antennas at the base station.

[0028] Step S33: Calculate the likelihood message based on the message passing rules on the factor graph: The result obtained in step S32... A linear posterior estimate Divide by The prior distribution of the linear detection in this round is used to obtain the likelihood message of the transmitted symbol. The likely message As a linear detection pair for transmitted symbols The basis for iterative correction of the posterior estimate; using the calculation rule of division of complex Gaussian distribution, the result obtained in step S32 is... A linear posterior estimate Divide by Prior distribution of linear detection in this round of iteration , to obtain a likelihood message for calibrating the estimate of the noiseless received signal , the likelihood message is used as the basis for iteratively calibrating the estimate of the noiseless received signal by linear detection; Specifically, the likelihood message is given by , ; Step S34, updating the posterior estimate: using the scalar Gaussian denoiser, the prior information of the transmit symbol modulation constellation is introduced, and the posterior estimate mean and the corresponding posterior estimate scalar variance are calculated according to the likelihood estimation result given by the linear detection, and the likelihood message is removed according to the calculation rule of the complex Gaussian distribution, to obtain the message from the variable node corresponding to the transmit symbol to the factor node corresponding to , i.e., the prior distribution of for the next round of iterative linear detection; ; Specifically, is given by , , wherein denotes the mean of the likelihood distribution in the parentheses with respect to the prior distribution of the transmit symbol, denotes the mean of the message to the node, i.e., the mean of the likelihood estimation of given by the linear detection, denotes the variance of the message to the node, i.e., the variance of the likelihood estimation of given by the linear detection, denotes the variance of the likelihood distribution in the parentheses with respect to the prior distribution of the transmit symbol, denotes the proportional relationship, i.e., the formula on the right side of the symbol omits the constant coefficient except the distribution e exponential term.

[0029] Step S35, updating the posterior estimate: by phase alignment, the sampling information of each analog-to-digital converter branch is aggregated, and the Bayesian formula is applied to update the prior distribution of for the next round of iterative linear detection; The Bayesian formula applied is , denotes the posterior distribution of the noiseless received signal with respect to the single-bit sampling information, denotes the likelihood distribution of the noiseless received signal with respect to the single-bit sampling information, denotes the prior distribution of the noiseless received signal, represents the single-bit sampling information of all the branches of the first ADC, represents the joint probability distribution of the single-bit sampling information of all the branches of the ADC.

[0030] As shown in Figure 5 , aggregating the sampling information of all the branches of the ADC through phase alignment comprises the following steps: Step S351, first calculate the mean value and the variance of the posterior distribution corresponding to the first branch of the ADC, wherein the mean value and the variance are calculated by the following formulas respectively: , wherein , In the present application, the quantized sampling result is a single-bit quantization, so when the quantized sampling result is less than 0, , when the quantized sampling result is greater than 0, , b=2, represents the single-bit sampling information of all the branches of the first ADC, represents the variance of the 0-mean Gaussian white noise at the receiving end of the communication system, represents the value of the standard Gaussian probability density function at , represents the value of the standard Gaussian cumulative distribution function at ; Step S352, introduce the second branch of the ADC, and when considering the sampling information obtained after phase rotation , the mean value of the updated posterior distribution is multiplied by the phase alignment coefficient first, then and are calculated, and then the coefficient is multiplied to eliminate the influence of the phase alignment coefficient, that is, the mean value and the variance of the posterior distribution of the second branch of the ADC are obtained. Step S353, repeat step S352 until the sampling information of all the branches of the ADC is considered, that is, the mean value and the variance of the posterior distribution of the current iteration of the noise-free received signal are obtained.

[0031] ​​​Step S36, repeating steps S32 to S35, when the number of iterations reaches a set maximum number, or the variance of the user's estimate of the transmitted symbol has been below a set threshold, then stop vector approximation message passing to output the posterior estimate of the transmitted symbol and make a hard decision based on the Euclidean distance between the posterior estimate and each constellation point in the modulation constellation to output a signal detection result.

[0032] In Figure 6 , when the base station antenna number is 32 and there are 8 single antenna users, the symbol error rate level of the application is significantly better than the comparative scheme (LMMSE) in all signal-to-noise ratio ranges. For a QPSK modulated transmitted signal, LMMSE cannot achieve high reliability signal detection under single-bit quantization, while the application can achieve a symbol error rate level of at a signal-to-noise ratio of 11dB.

[0033] In Figure 7 , when the base station antenna number is 32 and there are 8 single antenna users, and two analog-to-digital converter (ADC) branches are set, due to the central limit theorem, the phase of the received signal is uniformly distributed, so when the phase rotation value setting between the branches is , the application has the optimal detection performance. If we further increase the quantization branch, we can obtain better detection performance under the appropriate phase rotation setting, i.e., the phase rotation angle between each branch is . For example, when there are 3 ADC branches and the phase rotation angle is , a 2dB performance improvement can be obtained at the symbol error rate level of .

[0034] The application will be further described below through specific embodiments.

[0035] In the application, the terms "first", "second", "third", etc. are only used to distinguish similar objects, and do not necessarily describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. In the description, the orientations or positions indicated by "up", "down", "left", "right", "front" and "back" are based on the orientations or positions shown in the drawings, and are only for the convenience of describing the application, and do not indicate or imply that the device must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the scope of protection of the application. For those skilled in the art, the specific meaning of the above terms in this application can be understood according to the specific circumstances.

[0036] ​In addition, in the description of the present application, "multiple" refers to two or more, unless otherwise specified. The association relationship of the associated objects described by "and / or" indicates that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist together, and B exists alone. The character " / " generally represents an "or" relationship between the associated objects before and after it.

[0037] The above is only a specific embodiment of the present application, but the design concept of the present application is not limited thereto, and any non-essential modification of the present application using this concept shall be deemed to infringe the protection scope of the present application.

Claims

1. A method for detecting communication signals using terahertz single-bit quantization in a multi-analog-to-digital converter branch, characterized in that: Includes the following steps: Step S1: Construct a terahertz single-bit quantization communication system with multiple analog-to-digital converter branches. Each analog-to-digital converter branch includes a phase rotation module and an analog-to-digital converter connected in sequence. The phase rotation modules of different analog-to-digital converter branches have different phase rotation angles. The communication system is a multi-user multiple-input multiple-output system. Step S2: Represent the communication system using a vector value factor graph; Step S3: Perform vector approximation message passing on the vector value factor graph, iteratively calibrate the estimation of the noiseless received signal, and finally realize the detection of the transmitted symbol. During the message passing process, the single-bit sampling information of different analog-to-digital converter branches is aggregated.

2. The communication signal detection method using terahertz single-bit quantization in a multi-analog-to-digital converter branch according to claim 1, characterized in that: In step S1, the received signal model in the communication system is represented as follows: The vector composed of the noiseless received signals at all receiving antennas is represented as ,in, The single-bit quantization process implemented in the analog-to-digital converter branch. This is a vector composed of the phase rotation angles of the phase rotation modules in each analog-to-digital converter branch. It is the identity matrix. The equivalent channel matrix for terahertz block fading baseband is derived from the center frequency of the communication system, the locational relationship between the base station and each user, and the surrounding scattering environment. Symbols are transmitted to all users, and these transmitted symbols satisfy a discrete uniform distribution on the modulation constellation diagram used by the communication system. It is a zero-mean Gaussian white noise matrix.

3. The communication signal detection method using terahertz single-bit quantization in a multi-analog-to-digital converter branch according to claim 2, characterized in that: In step S2, in the vector value factor graph, all users send symbols. The vector consisting of the noise-free received signals at all receiving antennas Let and represent the two variable nodes of the vector-valued factor graph. Each factor obtained by decomposing the joint probability density function of the communication system according to different vector parameters using Bayes' theorem is represented as a factor node in the vector-valued factor graph. The factor nodes include the conditional probability distribution of the noise-free received signal relative to the transmitted symbol and the channel matrix. The conditional probability distribution of single-bit sampled information of the analog-to-digital converter branch relative to the noise-free received signal. and the prior distribution of transmitted symbols .

4. The communication signal detection method using terahertz single-bit quantization in a multi-analog-to-digital converter branch according to claim 3, characterized in that: In step S3, Corresponding factor nodes and When the corresponding variable nodes pass messages, a phase alignment operation is set up with the phase difference between different analog-to-digital converter branches as a parameter to aggregate single-bit sampling information from different analog-to-digital converter branches.

5. The communication signal detection method using terahertz single-bit quantization in a multi-analog-to-digital converter branch according to claim 4, characterized in that: In step S3, the vector approximation message passing on the vector value factor graph specifically includes the following steps: Step S31, Initialization: Initialize the two variable nodes and the messages between factor nodes in the vector value factor graph; Step S32, Linearity Detection: Based on the initialization or the two variable nodes from the previous iteration, send the data to the factor node. The message, using the product formula of the Gaussian distribution of vector values, calculates... A linear posterior estimate and A linear posterior estimate ; Step S33: Calculate the likelihood message based on the message passing rules on the factor graph: The result obtained in step S32... A linear posterior estimate Divide by The prior distribution of the linear detection in this round is used to obtain the likelihood message of the transmitted symbol, and this likelihood message is used as the linear detection pair for the transmitted symbol. The basis for iterative correction of the posterior estimate; using the calculation rule of division of complex Gaussian distribution, the result obtained in step S32 is... A linear posterior estimate Divide by Prior distribution of linear detection in this round of iteration This yields the likelihood message used to calibrate the estimate of the noiseless received signal. The likely message This serves as the basis for iterative calibration of the estimation of the noiseless received signal using linear detection; Step S34, Update Posterior estimation: Using a scalar Gaussian denoiser, prior information about the transmitted symbol modulation constellation is incorporated to calculate... Posterior estimated mean and the corresponding posterior estimate of scalar variance and will Likelihood messages are removed according to the calculation rules of the complex Gaussian distribution. To obtain the variable node corresponding to the sent symbol The corresponding factor node message, i.e., the next round of iterative linear detection. The prior distribution; Step S35, Update Posterior estimation: Sampling information from each analog-to-digital converter branch is aggregated through phase alignment, and Bayes' theorem is applied to update the linear detection for the next iteration. The prior distribution; Step S36: Repeat steps S32 to S35 until the number of iterations reaches the set maximum, or the user sends a symbol. If the estimated variance is below a set threshold, then vector approximation message passing stops. posterior estimation For output, based on Hard decision is made using the Euclidean distances between the points in the modulation constellation and the signal detection results.

6. The communication signal detection method for terahertz single-bit quantization of a multi-analog-to-digital converter branch according to claim 5, characterized in that: In step S32, the linear posterior estimation middle, Let be the mean of this linear posterior estimate. Let be the mean of the variance of this linear posterior estimate, where Describes the Gaussian distribution in the complex field. Let X represent the mean of the prior X used for linear detection. Let X represent the variance of the prior X used for linear detection. This represents the mean of the likelihood estimate of the noiseless received signal used for linear detection. This represents the variance of the likelihood estimate for the noiseless received signal. express The conjugate transpose of . Represents the identity matrix. This represents the mean of the diagonal elements of the matrix; the linear posterior estimate middle, Let be the mean of this linear posterior estimate. Let be the mean of the variance of this linear posterior estimate, where This indicates the number of users per antenna transmitting signals. Indicates the number of receiving antennas of the base station; in step S33, the likelihood message middle, , .

7. The communication signal detection method for terahertz single-bit quantization of a multi-analog-to-digital converter branch according to claim 6, characterized in that: In step S34, the middle, , , ,in, This indicates calculating the mean. Indicates to The mean of the messages from the nodes, i.e., the pair given by linear detection. The mean of the likelihood estimate, Indicates to The variance of the node's messages, i.e., the variance given by linear detection. The variance of the likelihood estimate, This indicates the calculation of variance. This indicates a proportional relationship, meaning that the constant coefficients other than the exponent term e are omitted from the expression on the right side of the symbol.

8. The communication signal detection method using terahertz single-bit quantization in a multi-analog-to-digital converter branch according to claim 7, characterized in that: In step S35, the Bayesian formula applied is: , This represents the posterior distribution of the noiseless received signal relative to the single-bit sampled information. This represents the likelihood distribution of the noiseless received signal relative to the single-bit sampled information. This represents the prior distribution of the noiseless received signal. This represents the single-bit sampled information of each analog-to-digital converter branch on all receiving antennas. This represents the joint probability distribution of single-bit sampled information from each analog-to-digital converter branch.

9. The communication signal detection method using terahertz single-bit quantization in a multi-analog-to-digital converter branch according to claim 8, characterized in that: In step S35, the step of aggregating the sampling information of each analog-to-digital converter branch through phase alignment includes the following steps: Step S351: First calculate the posterior distribution corresponding to the first analog-to-digital converter branch. If the phase rotation angle of the first analog-to-digital converter branch is Then the mean of the posterior distribution is and variance Using the formulas respectively and Calculate, where, , When the quantization sampling result is less than 0, , When the quantization sampling result is greater than 0, , , This represents the sampling information of the first analog-to-digital converter branch on all receiving antennas. This represents the variance of the zero-mean Gaussian white noise at the receiving end of the communication system. The standard Gaussian probability density function is represented in The value at that location, This indicates that the standard Gaussian cumulative distribution function is in The value at; Step S352, Introduce the first One analog-to-digital converter branch, considering phase rotation When obtaining subsequent sampling information, it is necessary to update the posterior distribution. The mean is first multiplied by the phase alignment coefficient. , then calculate and Then multiply by the coefficient. To eliminate the influence of the phase alignment coefficient, i.e., to obtain up to the [number]th ... The posterior distribution of each analog-to-digital converter branch The mean and variance, This represents the single-bit sampled information of the k-th analog-to-digital converter branch on all receiving antennas; Step S353: Repeat step S352 until the sampling information of all analog-to-digital converter branches is taken into account, thus obtaining the posterior distribution of the noiseless received signal in the current iteration. mean and variance .

Images