Bayesian inference based generalized likelihood ratio detection method and apparatus

Abstract

The application relates to a generalized likelihood ratio test method and device based on Bayesian inference. The method comprises the following steps: acquiring reference channel and monitoring channel signals of a transmitting source and a receiving node, and constructing a binary hypothesis test signal model of target presence / absence; modeling unknown channel response coefficients as random variables with a known prior distribution, and modeling a transmitting signal symbol sequence as a deterministic variable, and constructing a joint probability distribution function of two types of hypotheses; obtaining an edge probability function through integral edge processing, maximizing the edge probability function to obtain a deterministic variable estimation value, constructing a Bayesian generalized likelihood ratio test statistic, and comparing the Bayesian generalized likelihood ratio test statistic with a preset threshold to determine whether the target exists. By using the method, channel characteristics and signal structure prior knowledge can be fully fused, detection robustness and sensitivity in a low signal-to-noise ratio and high dynamic environment can be improved, calculation complexity can be simplified, spatial diversity gain can be efficiently utilized, and quasi-optimal detection under the constraint of a constant false alarm probability can be realized.

Description

Technical Field

[0001] This application relates to the fields of wireless communication, radar signal processing and navigation signal processing, and in particular to a generalized likelihood ratio detection method and apparatus based on Bayesian inference. Background Technology

[0002] In navigation, wireless communication, and radar systems, such as satellite navigation augmentation systems, 5G / 6G multi-antenna networks, and distributed sensor networks, weak signal detection, abnormal signal identification, or interference detection are often required under conditions of strong noise, multipath propagation, and dynamic channels. Traditional detection methods, such as matched filtering, energy detection, or the classical generalized likelihood ratio test, typically assume that channel coefficients and signal parameters are deterministic variables, failing to fully utilize statistical prior knowledge of the channel and signal. This leads to performance degradation in low signal-to-noise ratio, high dynamic, or multi-user interference environments.

[0003] Bayesian inference methods, by introducing prior distributions of parameters, can effectively fuse historical information and observational data, thereby improving detection robustness. Especially in applications such as navigation signal authentication, blind detection of communication signals, and spectrum sensing, combining prior knowledge of channel fading models and signal modulation structures can significantly enhance detection sensitivity and reliability. Summary of the Invention

[0004] Therefore, it is necessary to provide a generalized likelihood ratio detection method and device based on Bayesian inference that can comprehensively integrate prior knowledge of channel characteristics and signal structure and improve detection performance to address the above-mentioned technical problems.

[0005] A generalized likelihood ratio detection method based on Bayesian inference, applied to a distributed multi-antenna system, the method comprising:

[0006] Acquire the reference channel signals and monitoring channel signals corresponding to multiple transmitters and multiple receivers in a distributed multi-antenna system;

[0007] Based on the reference channel signal and the monitoring channel signal, a binary hypothesis testing signal model is constructed for the unit to be detected; wherein, in the binary hypothesis testing signal model, the first hypothesis corresponds to the existence of the target, and the second hypothesis corresponds to the absence of the target;

[0008] Based on the binary hypothesis testing signal model, the unknown channel response coefficients are modeled as random variables following a known prior distribution, and the unknown transmitted signal symbol sequence is modeled as a deterministic variable, thus constructing a joint probability distribution function under the first and second hypotheses.

[0009] By performing integral marginalization on the random variables in the joint probability distribution function under the first and second hypotheses respectively, a first marginal probability function and a second marginal probability function related to the deterministic variable are obtained;

[0010] Maximize the first marginal probability function and the second marginal probability function respectively to obtain the estimated values ​​of the deterministic variables under the first assumption and the second assumption;

[0011] Based on the estimated value, construct the Bayesian generalized likelihood ratio test statistic.

[0012] The test statistic is compared with a preset detection threshold, and the presence of a target within the detection unit is determined based on the comparison result.

[0013] In one embodiment, the binary hypothesis test signal model is:

[0014] ;

[0015] in, Represents the reference channel signal vector. Represents the monitoring channel signal vector. This represents an unknown transmitted signal symbol sequence vector of length L. It is normalized Point discrete Fourier transform matrix, Indicate the first hypothesis, Indicates the second hypothesis; These are the unknown complex coefficients of the direct wave. Indicates the target reflectance coefficient. and These are independent Gaussian noise vectors in the reference channel and the monitoring channel, respectively. Indicates the serial number of the emission source. This indicates the sequence number of the receiving node.

[0016] In one embodiment, the channel response coefficients include: unknown complex coefficients of the direct wave. and target reflectivity ;

[0017] The unknown complex coefficients of the direct wave The prior distribution is:

[0018] ;

[0019] The target reflectivity The prior distribution is:

[0020] ;

[0021] in, This represents the variance parameter corresponding to the prior distribution of the unknown complex coefficients of the direct wave. This represents the variance parameter corresponding to the prior distribution of the target reflectance coefficient. Indicates the serial number of the emission source. Indicates the sequence number of the receiving node. It is a cyclically symmetric complex Gaussian distribution. Indicates a direct wave. Indicates the goal.

[0022] In one embodiment, constructing the joint probability distribution function under the first and second hypotheses includes:

[0023] The joint probability distribution function under the first hypothesis is constructed as follows:

[0024] ;

[0025] .

[0026] The joint probability distribution function under the second hypothesis is constructed as follows:

[0027] ;

[0028] .

[0029] In one embodiment, the first edge probability function and the second edge probability function are maximized respectively, specifically including:

[0030] In the first assumption Below, the first marginal probability function is:

[0031] ;

[0032] .

[0033] In the second hypothesis Below, the second marginal probability function is:

[0034] ;

[0035] .

[0036] In one embodiment, the first marginal probability function and the second marginal probability function are maximized respectively to obtain estimates of the deterministic variable under the first and second assumptions, including:

[0037] Maximizing the first marginal probability function yields the first objective function:

[0038] ;

[0039] in, Denotes the normalization factor, and by maximizing the first objective function, we obtain... , , , ,and Represents the largest eigenvalue. It is the total number of transmitters in a distributed multi-antenna system; This represents the total number of receiving nodes;

[0040] Maximizing the second marginal probability function yields the second objective function:

[0041] ;

[0042] Maximize the second objective function to obtain , .

[0043] In one embodiment, the Bayesian generalized likelihood ratio test statistic is constructed based on the estimated value as follows:

[0044] ;

[0045] in, This represents the test statistic. It is a threshold value that has been appropriately adjusted.

[0046] A generalized likelihood ratio detection device based on Bayesian inference, the device comprising:

[0047] The signal acquisition module is used to acquire the reference channel signals and monitoring channel signals corresponding to multiple transmitters and multiple receivers in a distributed multi-antenna system.

[0048] The signal modeling module is used to construct a binary hypothesis testing signal model for the unit to be detected based on the reference channel signal and the monitoring channel signal; wherein, in the binary hypothesis testing signal model, the first hypothesis corresponds to the existence of the target, and the second hypothesis corresponds to the absence of the target.

[0049] The probability construction module is used to model the unknown channel response coefficients as random variables that follow a known prior distribution and the unknown transmitted signal symbol sequence as deterministic variables based on the binary hypothesis testing signal model, and to construct a joint probability distribution function under the first and second hypotheses.

[0050] The marginalization processing module is used to perform integral marginalization processing on the random variables in the joint probability distribution function under the first hypothesis and the second hypothesis, respectively, to obtain the first marginal probability function and the second marginal probability function related to the deterministic variable;

[0051] The estimation module is used to maximize the first marginal probability function and the second marginal probability function respectively to obtain the estimated values ​​of the deterministic variables under the first assumption and the second assumption.

[0052] The statistic construction module is used to construct the Bayesian generalized likelihood ratio test statistic based on the estimated value.

[0053] The target determination module is used to compare the test statistic with a preset detection threshold and determine whether a target exists in the unit to be detected based on the comparison result.

[0054] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program performing the following steps:

[0055] Acquire the reference channel signals and monitoring channel signals corresponding to multiple transmitters and multiple receivers in a distributed multi-antenna system;

[0056] Based on the reference channel signal and the monitoring channel signal, a binary hypothesis testing signal model is constructed for the unit to be detected; wherein, in the binary hypothesis testing signal model, the first hypothesis corresponds to the existence of the target, and the second hypothesis corresponds to the absence of the target;

[0057] Based on the binary hypothesis testing signal model, the unknown channel response coefficients are modeled as random variables following a known prior distribution, and the unknown transmitted signal symbol sequence is modeled as a deterministic variable, thus constructing a joint probability distribution function under the first and second hypotheses.

[0058] By performing integral marginalization on the random variables in the joint probability distribution function under the first and second hypotheses respectively, a first marginal probability function and a second marginal probability function related to the deterministic variable are obtained;

[0059] Maximize the first marginal probability function and the second marginal probability function respectively to obtain the estimated values ​​of the deterministic variables under the first assumption and the second assumption;

[0060] Based on the estimated value, construct the Bayesian generalized likelihood ratio test statistic.

[0061] The test statistic is compared with a preset detection threshold, and the presence of a target within the detection unit is determined based on the comparison result.

[0062] A computer-readable storage medium having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0063] Acquire the reference channel signals and monitoring channel signals corresponding to multiple transmitters and multiple receivers in a distributed multi-antenna system;

[0064] Based on the reference channel signal and the monitoring channel signal, a binary hypothesis testing signal model is constructed for the unit to be detected; wherein, in the binary hypothesis testing signal model, the first hypothesis corresponds to the existence of the target, and the second hypothesis corresponds to the absence of the target;

[0065] Based on the binary hypothesis testing signal model, the unknown channel response coefficients are modeled as random variables following a known prior distribution, and the unknown transmitted signal symbol sequence is modeled as a deterministic variable, thus constructing a joint probability distribution function under the first and second hypotheses.

[0066] By performing integral marginalization on the random variables in the joint probability distribution function under the first and second hypotheses respectively, a first marginal probability function and a second marginal probability function related to the deterministic variable are obtained;

[0067] Maximize the first marginal probability function and the second marginal probability function respectively to obtain the estimated values ​​of the deterministic variables under the first assumption and the second assumption;

[0068] Based on the estimated value, construct the Bayesian generalized likelihood ratio test statistic.

[0069] The test statistic is compared with a preset detection threshold, and the presence of a target within the detection unit is determined based on the comparison result.

[0070] The aforementioned generalized likelihood ratio detection method and device based on Bayesian inference, by introducing a Bayesian processing framework into the system, effectively solves the problem that traditional passive radar detection methods fail to fully utilize prior knowledge of parameters, thus bringing significant technical improvements and beneficial effects in many aspects. First, by modeling the unknown channel response coefficients as random variables following a known prior distribution, this method quantifies the statistical uncertainties present in the actual channel environment and incorporates them into the detection criteria, making the detector more adaptable and robust to channel changes. Second, by treating the unknown transmitted signal sequence as a deterministic variable for subsequent estimation, the structural characteristics of the signal, such as specific digital modulation constellations, are preserved, making it possible to extract weak target signals under low signal-to-noise ratio conditions. Subsequently, by performing integral marginalization processing on the random variables in the joint probability distribution, the direct influence of these uncertain parameters on the detection statistics is cleverly eliminated, transforming the complex joint estimation problem into an optimization problem of deterministic variables, greatly simplifying the detector structure and improving computational feasibility. Then, by maximizing the marginal probability function to estimate the transmitted signal, this method essentially finds the signal sequence most likely to generate observation data, a process that fully utilizes the signal energy and spatial diversity gain of all receiving channels. Based on this, the constructed Bayesian generalized likelihood ratio test statistic integrates the ratio of the best fit under both the presence and absence of the target, forming a highly discriminative detection index. Finally, by comparing it with a preset threshold, optimal or near-optimal detection decisions are achieved under a constant false alarm probability constraint. Overall, compared to traditional generalized likelihood ratio detectors, this method significantly improves detection performance by effectively utilizing prior knowledge of unknown channel coefficients and transmitted symbols. This advantage is particularly pronounced at higher reference signal-to-noise ratio levels, providing a high-performance and highly practical target detection solution for distributed multi-antenna systems. Attached Figure Description

[0071] Figure 1 This is an application scenario diagram of a generalized likelihood ratio detection method based on Bayesian inference in one embodiment;

[0072] Figure 2 This is a flowchart illustrating a generalized likelihood ratio detection method based on Bayesian inference in one embodiment.

[0073] Figure 3 This is a structural block diagram of a generalized likelihood ratio detection device based on Bayesian inference in one embodiment;

[0074] Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0075] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0076] The generalized likelihood ratio detection method based on Bayesian inference provided in this application can be applied to, for example... Figure 1 In the application environment shown, the distributed multi-antenna system consists of multiple dispersed transmitting sources (such as broadcast towers and communication base stations) 110 and multiple receiving nodes 120. Each receiving node 120 communicates with a central processing server 104 via a network or transmits data to a terminal 102 with processing capabilities. The terminal 102 and server 104 can collaboratively or independently execute the target detection method of this application. The terminal 102 can be, but is not limited to, various personal computers, laptops, workstations, etc., and the server 104 can be a standalone server or a server cluster composed of multiple servers. The reference channel signal and monitoring channel signal acquired by the receiving node 120 are transmitted to the processing unit, terminal 102, or server 104 for subsequent signal processing and target detection.

[0077] In one embodiment, such as Figure 2 As shown, a generalized likelihood ratio detection method based on Bayesian inference is provided, which is then applied to... Figure 1 Taking the server in the example, the following steps are included:

[0078] Step 202: Obtain the reference channel signals and monitoring channel signals corresponding to multiple transmitters and multiple receivers in the distributed multi-antenna system.

[0079] Specifically, for each pair of transmitters M (M=1,2,…,m) and receivers N (N=1,2,…,n) in a distributed multi-antenna system, the reference channel signal vector, after preprocessing such as direct wave interference cancellation and time / Doppler compensation, is obtained. and monitoring channel signal vector These signals are typically processed in the frequency domain or baseband, and the signals from all the transmitter-receiver node pairs are combined to form a complete observation dataset.

[0080] Step 204: Based on the reference channel signal and the monitoring channel signal, construct a binary hypothesis test signal model for the unit to be detected; wherein, in the binary hypothesis test signal model, the first hypothesis corresponds to the existence of the target, and the second hypothesis corresponds to the absence of the target.

[0081] Specifically, for a range-Doppler cell to be detected, a binary hypothesis testing mathematical model is constructed based on the acquired reference channel signal and monitoring channel signal; this model contains two mutually exclusive hypotheses: the first hypothesis... This indicates that the unit has a target; second assumption This indicates that the unit does not have a target; the specific form of the model is as follows:

[0082] ;

[0083] in, Represents the reference channel signal vector. Represents the monitoring channel signal vector. This represents an unknown transmitted signal symbol sequence vector of length L. It is normalized The point discrete Fourier transform matrix, its first... The elements are And satisfy ; Indicate the first hypothesis, Indicates the second hypothesis; These are the unknown complex coefficients of the direct wave. Indicates the target reflectance coefficient. and These are independent Gaussian noise vectors in the reference channel and the monitoring channel, respectively. Indicates the serial number of the emission source. This indicates the sequence number of the receiving node.

[0084] Step 206: Based on the binary hypothesis testing signal model, the unknown channel response coefficients are modeled as random variables following a known prior distribution, and the unknown transmitted signal symbol sequence is modeled as a deterministic variable, thus constructing a joint probability distribution function under the first and second hypotheses.

[0085] Specifically, based on the assumption of a Gaussian distribution for constructing a binary hypothesis testing signal model and noise, we can write: and Assume the likelihood function of the observed data with respect to all unknown parameters; let and These represent the sets of all reference signals and monitoring signals, respectively; vectors and Collect the unknown coefficients separately; set Represents data symbols; based on the Gaussian assumption of noise, in the assumption Below, deterministic parameters , and The likelihood function is:

[0086] ;

[0087] in It can be decomposed into:

[0088] ;

[0089] in:

[0090] ;

[0091] .

[0092] Similarly, in The likelihood function under the assumption is:

[0093] ;

[0094] This function can also be decomposed, making The likelihood function under the assumptions can be rewritten as:

[0095] .

[0096] Step 208: Perform integral marginalization on the random variables in the joint probability distribution function under the first and second hypotheses respectively to obtain the first marginal probability function and the second marginal probability function related to the deterministic variable.

[0097] Specifically, in order to eliminate the uncertainty of random variables, namely the channel response coefficients, and focus on estimating deterministic variables... The random variables in the constructed joint probability distribution function are marginalized by integration; this is a typical Bayesian averaging process.

[0098] Likelihood function Through the Marginalization yields, i.e.:

[0099] ;

[0100] Among them, the reference channel weighting coefficient .

[0101] Similarly, the present invention can be achieved by... Marginalization yields the function The result is:

[0102] ;

[0103] Among them, the weighting coefficient of the monitoring channel .

[0104] Step 210: Maximize the first marginal probability function and the second marginal probability function respectively to obtain the estimated values ​​of the deterministic variables under the first and second hypotheses.

[0105] Specifically, maximizing the first and second marginal probability functions is equivalent to solving two optimization problems, with the goal of finding the transmitted signal sequence most likely to generate the observed data under their respective assumptions. The estimated value.

[0106] Maximizing the first marginal probability function is equivalent to maximizing its logarithm, that is, maximizing the first objective function is:

[0107] ;

[0108] The maximum value of this function is ,in , , ,and This represents the largest eigenvalue.

[0109] Maximizing the second objective function is:

[0110] ;

[0111] Its maximum value is ,in .

[0112] Step 212: Construct the Bayesian generalized likelihood ratio test statistic based on the estimated value.

[0113] Specifically, by combining the prior distribution of unknown channel coefficients, i.e. and A closed-loop Bayesian generalized likelihood ratio test (GLRT) detector is proposed, in which the unknown emission symbol Treated as a deterministic unknown, according to the Neyman-Pearson criterion, the Bayesian GLRT detector expression is:

[0114] ;

[0115] in To determine the detection threshold; the obtained optimal estimate will be... and Substituting the maximum value into the Bayesian GLRT detector expression and through appropriate derivation, we obtain the Bayesian generalized likelihood ratio test detector as follows:

[0116] ;

[0117] in It is a threshold value that has been appropriately adjusted to achieve the set false alarm probability.

[0118] Step 214: Compare the test statistic with the preset detection threshold, and determine whether there is a target in the unit to be detected based on the comparison result.

[0119] Specifically, based on the estimated value, the test statistic T calculated by the Bayesian generalized likelihood ratio is compared with a preset detection threshold γ. This threshold γ is usually preset by theoretical calculation or Monte Carlo simulation based on the acceptable false alarm probability of the system. If T ≥ γ, the target is determined to exist in the unit to be detected (accept H1); otherwise, the target is determined not to exist (accept H0), and the final binary detection decision result is output.

[0120] The generalized likelihood ratio detection method based on Bayesian inference, by introducing a systematic Bayesian processing framework, effectively addresses the problem of traditional passive radar detection methods failing to fully utilize prior knowledge of parameters, thus bringing significant technical improvements and beneficial effects in multiple aspects. First, by modeling the unknown channel response coefficients as random variables following a known prior distribution, this method quantifies the statistical uncertainties present in the actual channel environment and incorporates them into the detection criteria, making the detector more adaptable and robust to channel changes. Second, by treating the unknown transmitted signal sequence as a deterministic variable for subsequent estimation, the structural characteristics of the signal, such as specific digital modulation constellations, are preserved, making it possible to extract weak target signals under low signal-to-noise ratio conditions. Subsequently, by performing integral marginalization processing on the random variables in the joint probability distribution, the direct influence of these uncertain parameters on the detection statistics is cleverly eliminated, transforming the complex joint estimation problem into an optimization problem of deterministic variables, greatly simplifying the detector structure and improving computational feasibility. Then, by maximizing the marginal probability function to estimate the transmitted signal, this method essentially finds the signal sequence most likely to generate observation data, a process that fully utilizes the signal energy and spatial diversity gain of all receiving channels. Based on this, the constructed Bayesian generalized likelihood ratio test statistic integrates the ratio of the best fit under both the presence and absence of the target, forming a highly discriminative detection index. Finally, by comparing it with a preset threshold, optimal or near-optimal detection decisions are achieved under a constant false alarm probability constraint. This provides a high-performance and highly practical target detection solution for distributed multi-antenna systems.

[0121] In one embodiment, the binary hypothesis test signal model is as follows:

[0122] ;

[0123] in, Represents the reference channel signal vector. Represents the monitoring channel signal vector. This represents an unknown transmitted signal symbol sequence vector of length L. It is normalized Point discrete Fourier transform matrix, Indicate the first hypothesis, Indicates the second hypothesis; These are the unknown complex coefficients of the direct wave. Indicates the target reflectance coefficient. and These are independent Gaussian noise vectors in the reference channel and the monitoring channel, respectively. Indicates the serial number of the emission source. This indicates the sequence number of the receiving node.

[0124] In practical implementation, consider a distributed passive MIMO radar consisting of M transmitters and N receivers. It is normalized The point discrete Fourier transform matrix, where the first... The elements are And satisfy It maps the transmitted symbol sequence s in the time or code domain to the domain of interest for radar signal processing, such as the frequency domain or the time-delay-Doppler domain; for a given range-Doppler cell, it compensates for the corresponding time delay and Doppler frequency shift, i.e. and The target echo (if present) can be aligned to this cell; the model assumes that direct path interference in the reference channel has been completely eliminated by preprocessing techniques, such as extended cancellation algorithms, so the reference signal... It mainly consists of noise and weak residual interference; among which and These are independent Gaussian noises in the reference channel and the monitoring channel, respectively. This model provides an accurate mathematical description for subsequent statistical-based detection, enabling joint detection using information from multiple independent propagation paths (spatial diversity), and enhancing robustness in the event of signal fading or interference in a single path.

[0125] In one embodiment, the channel response coefficients include: unknown complex coefficients of the direct wave. and target reflectivity Unknown complex coefficients of direct waves The prior distribution is:

[0126] ;

[0127] Target reflectance The prior distribution is:

[0128] ;

[0129] in, This represents the variance parameter corresponding to the prior distribution of the unknown complex coefficients of the direct wave. This represents the variance parameter corresponding to the prior distribution of the target reflectance coefficient. Indicates the serial number of the emission source. Indicates the sequence number of the receiving node. It is a cyclically symmetric complex Gaussian distribution. Indicates a direct wave. Indicates the goal.

[0130] In practical implementation, in real wireless propagation environments, channel coefficients are usually not deterministic; the unknown complex coefficients of direct waves... This reflects the complex gain of the direct path from the transmitter M to the receiver N. Its fluctuations are caused by factors such as shadow fading and multipath effects along the propagation path, making its modeling as a complex Gaussian random variable common and reasonable; target reflection coefficient. This includes information such as the fluctuations in the target's radar cross-section (RCS), two-way path loss, and phase changes, and is often modeled as a complex Gaussian random variable. When the prior information is accurate, it can guide the detector to utilize the observation data more effectively; even if the prior information is not very accurate, the Bayesian framework can correct the posterior inference through the observation data, making the detector more robust and efficient in the face of channel uncertainties than traditional methods that completely ignore the prior.

[0131] In one embodiment, constructing the joint probability distribution function under the first and second hypotheses includes constructing the joint probability distribution function under the first hypothesis as follows:

[0132] ;

[0133] .

[0134] The joint probability distribution function under the second hypothesis is constructed as follows:

[0135] ;

[0136] .

[0137] In one embodiment, the first marginal probability function and the second marginal probability function are maximized respectively, specifically including: under the first assumption Below, the first marginal probability function is:

[0138] ;

[0139] .

[0140] Second hypothesis Below, the second marginal probability function is:

[0141] ;

[0142] .

[0143] In one embodiment, the first marginal probability function and the second marginal probability function are maximized respectively to obtain the estimated values ​​of the deterministic variables under the first and second assumptions. This includes maximizing the first marginal probability function to obtain the first objective function:

[0144] ;

[0145] in, Denotes the normalization factor, and maximizing the first objective function yields... , , , ,and Represents the largest eigenvalue. It is the total number of transmitters in a distributed multi-antenna system; This represents the total number of receiving nodes.

[0146] Maximizing the second marginal probability function yields the second objective function:

[0147] ;

[0148] Maximize the second objective function to obtain , .

[0149] In one embodiment, the Bayesian generalized likelihood ratio test statistic is constructed based on the estimated value as follows:

[0150] ;

[0151] in, This represents the test statistic. It is a threshold value that has been appropriately adjusted.

[0152] It should be understood that, although Figure 2 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 2At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but may be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0153] In one embodiment, such as Figure 3 As shown, a generalized likelihood ratio detection device based on Bayesian inference is provided, comprising: a signal acquisition module 302, a signal modeling module 304, a probability construction module 306, an edge detection module 308, an estimation module 310, a statistics construction module 312, and a target determination module 314, wherein:

[0154] The signal acquisition module 302 is used to acquire the reference channel signal and monitoring channel signal corresponding to multiple transmitters and multiple receivers in a distributed multi-antenna system.

[0155] The signal modeling module 304 is used to construct a binary hypothesis testing signal model for the unit to be detected based on the reference channel signal and the monitoring channel signal; wherein, in the binary hypothesis testing signal model, the first hypothesis corresponds to the existence of the target, and the second hypothesis corresponds to the absence of the target.

[0156] The probability construction module 306 is used to model the unknown channel response coefficients as random variables that follow a known prior distribution and the unknown transmitted signal symbol sequence as deterministic variables based on the binary hypothesis testing signal model, and to construct the joint probability distribution function under the first and second hypotheses.

[0157] The marginalization processing module 308 is used to perform integral marginalization processing on the random variables in the joint probability distribution function under the first hypothesis and the second hypothesis, respectively, to obtain the first marginal probability function and the second marginal probability function related to the deterministic variable;

[0158] The estimation module 310 is used to maximize the first marginal probability function and the second marginal probability function respectively to obtain the estimated values ​​of the deterministic variables under the first and second assumptions.

[0159] The statistic construction module 312 is used to construct the Bayesian generalized likelihood ratio test statistic based on the estimated value.

[0160] The target determination module 314 is used to compare the test statistic with the preset detection threshold and determine whether there is a target in the unit to be detected based on the comparison result.

[0161] In one embodiment, the signal modeling module 304 is further used for binary hypothesis testing of the signal model as follows:

[0162] ;

[0163] in, Represents the reference channel signal vector. Represents the monitoring channel signal vector. This represents an unknown transmitted signal symbol sequence vector of length L. It is normalized Point discrete Fourier transform matrix, Indicate the first hypothesis, Indicates the second hypothesis; These are the unknown complex coefficients of the direct wave. Indicates the target reflectance coefficient. and These are independent Gaussian noise vectors in the reference channel and the monitoring channel, respectively. Indicates the serial number of the emission source. This indicates the sequence number of the receiving node.

[0164] In one embodiment, the signal modeling module 304 is further configured to include the channel response coefficients as: unknown complex coefficients of the direct wave. and target reflectivity Unknown complex coefficients of direct waves The prior distribution is:

[0165] ;

[0166] Target reflectance The prior distribution is:

[0167] ;

[0168] in, The variance of the corresponding distribution is known a priori or can be estimated.

[0169] In one embodiment, the probability construction module 306 is further configured to construct a joint probability distribution function under the first hypothesis and the second hypothesis, including constructing the joint probability distribution function under the first hypothesis as follows:

[0170] ;

[0171] .

[0172] The joint probability distribution function under the second hypothesis is constructed as follows:

[0173] .

[0174] In one embodiment, the edge processing module 308 is further configured to maximize the first edge probability function and the second edge probability function respectively, specifically including: under the first assumption Below, the first marginal probability function is:

[0175] ;

[0176] .

[0177] Second hypothesis Below, the second marginal probability function is:

[0178] ;

[0179] .

[0180] In one embodiment, the estimation module 310 is further configured to maximize the first marginal probability function and the second marginal probability function respectively to obtain estimated values ​​of the deterministic variables under the first and second assumptions, including maximizing the first marginal probability function to obtain the first objective function as follows:

[0181] ;

[0182] in, Denotes the normalization factor, and maximizing the first objective function yields... , , , ,and This represents the largest eigenvalue.

[0183] Maximizing the second marginal probability function yields the second objective function:

[0184] ;

[0185] Maximize the second objective function to obtain , .

[0186] In one embodiment, the statistic construction module 312 is further configured to construct the Bayesian generalized likelihood ratio test statistic based on the estimated value:

[0187] ;

[0188] in, This represents the test statistic. It is a threshold value that has been appropriately adjusted.

[0189] Specific limitations regarding the Bayesian inference-based generalized likelihood ratio detection device can be found in the limitations of the Bayesian inference-based generalized likelihood ratio detection method described above, and will not be repeated here. Each module in the aforementioned Bayesian inference-based generalized likelihood ratio detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0190] In one embodiment, a computer device is provided, which may be a terminal, such as a local processing unit of a receiving station, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface is used to communicate with an external central server or other receiving node via a network connection. When the computer program is executed by the processor, it implements the steps in the above method embodiments. The display screen can be a liquid crystal display or an e-ink display, used to display detection results, system status, and performance curves, such as the relationship between the detection probability Pd and the signal-to-noise ratio SNR. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device casing, or an external keyboard, touchpad, or mouse, used for parameter settings, such as the false alarm probability Pfa, prior variance, and command input.

[0191] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0192] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the method described above.

[0193] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0194] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0195] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0196] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A generalized likelihood ratio detection method based on Bayesian inference, applied to distributed multi-antenna navigation, communication, or radar systems, characterized in that... The method includes: Acquire the reference channel signals and monitoring channel signals corresponding to multiple transmitters and multiple receivers in a distributed multi-antenna system; Based on the reference channel signal and the monitoring channel signal, a binary hypothesis testing signal model is constructed for the unit to be detected; wherein, in the binary hypothesis testing signal model, the first hypothesis corresponds to the existence of the target, and the second hypothesis corresponds to the absence of the target; Based on the binary hypothesis testing signal model, the unknown channel response coefficients are modeled as random variables following a known prior distribution, and the unknown transmitted signal symbol sequence is modeled as a deterministic variable, thus constructing a joint probability distribution function under the first and second hypotheses. By performing integral marginalization on the random variables in the joint probability distribution function under the first and second hypotheses respectively, a first marginal probability function and a second marginal probability function related to the deterministic variable are obtained; Maximize the first marginal probability function and the second marginal probability function respectively to obtain the estimated values ​​of the deterministic variables under the first assumption and the second assumption; Based on the estimated value, construct the Bayesian generalized likelihood ratio test statistic. The test statistic is compared with a preset detection threshold, and the presence of a target in the unit to be detected is determined based on the comparison result. The channel response coefficients include: unknown complex coefficients of the direct wave. and target reflectivity ; The unknown complex coefficients of the direct wave The prior distribution is: The target reflectivity The prior distribution is: in, This represents the variance parameter corresponding to the prior distribution of the unknown complex coefficients of the direct wave. This represents the variance parameter corresponding to the prior distribution of the target reflectance coefficient. Indicates the serial number of the emission source. Indicates the sequence number of the receiving node. It is a cyclically symmetric complex Gaussian distribution. Indicates a direct wave. Indicates the goal.

2. The method according to claim 1, characterized in that, The binary hypothesis test signal model is as follows: in, Represents the reference channel signal vector. Represents the monitoring channel signal vector. This represents an unknown transmitted signal symbol sequence vector of length L. It is normalized Point discrete Fourier transform matrix, Indicate the first hypothesis, Indicates the second hypothesis; These are the unknown complex coefficients of the direct wave. Indicates the target reflectance coefficient. and These are independent Gaussian noise vectors in the reference channel and the monitoring channel, respectively. Indicates the serial number of the emission source. This indicates the sequence number of the receiving node.

3. The method according to claim 1, characterized in that, Constructing the joint probability distribution function under the first and second hypotheses includes: The joint probability distribution function under the first hypothesis is constructed as follows: The joint probability distribution function under the second hypothesis is constructed as follows: 。 4. The method according to claim 3, characterized in that, Maximizing the first edge probability function and the second edge probability function respectively includes: In the first assumption Below, the first marginal probability function is: In the second hypothesis Below, the second marginal probability function is: This represents the normalization factor of the reference channel. This represents the normalization factor for the monitoring channel.

5. The method according to claim 4, characterized in that, Maximizing the first and second marginal probability functions respectively yields estimates of the deterministic variables under the first and second assumptions, including: Maximizing the first marginal probability function yields the first objective function: Where, maximizing the first objective function yields... , , , ,and Represents the largest eigenvalue. It is the total number of transmitters in a distributed multi-antenna system; This represents the total number of receiving nodes; Maximizing the second marginal probability function yields the second objective function: Maximize the second objective function to obtain , .

6. The method according to claim 5, characterized in that, Based on the estimated value, the Bayesian generalized likelihood ratio test statistic is constructed as follows: in, This represents the test statistic. This is a preset detection threshold.

7. A generalized likelihood ratio detection device based on Bayesian inference, characterized in that, The device includes: The signal acquisition module is used to acquire the reference channel signals and monitoring channel signals corresponding to multiple transmitters and multiple receivers in a multi-antenna system. The signal modeling module is used to construct a binary hypothesis testing signal model for the unit to be detected based on the reference channel signal and the monitoring channel signal; wherein, in the binary hypothesis testing signal model, the first hypothesis corresponds to the existence of the target, and the second hypothesis corresponds to the absence of the target. The probability construction module is used to model the unknown channel response coefficients as random variables that follow a known prior distribution and the unknown transmitted signal symbol sequence as deterministic variables based on the binary hypothesis testing signal model, and to construct a joint probability distribution function under the first and second hypotheses. The marginalization processing module is used to perform integral marginalization processing on the random variables in the joint probability distribution function under the first hypothesis and the second hypothesis, respectively, to obtain the first marginal probability function and the second marginal probability function related to the deterministic variable; The estimation module is used to maximize the first marginal probability function and the second marginal probability function respectively to obtain the estimated values ​​of the deterministic variables under the first assumption and the second assumption. The statistic construction module is used to construct the Bayesian generalized likelihood ratio test statistic based on the estimated value. The target determination module is used to compare the test statistic with a preset detection threshold and determine whether a target exists in the unit to be detected based on the comparison result. The channel response coefficients include: unknown complex coefficients of the direct wave. and target reflectivity ; The unknown complex coefficients of the direct wave The prior distribution is: The target reflectivity The prior distribution is: in, This represents the variance parameter corresponding to the prior distribution of the unknown complex coefficients of the direct wave. This represents the variance parameter corresponding to the prior distribution of the target reflectance coefficient. Indicates the serial number of the emission source. Indicates the sequence number of the receiving node. It is a cyclically symmetric complex Gaussian distribution. Indicates a direct wave. Indicates the goal.

Images