Direct-path interference suppressed distributed passive radar signal-level robust localization method

Abstract

The application discloses a direct-path interference suppression distributed passive radar signal-level robust positioning method, comprising the following steps: S1, system parameter initialization; S2, establishing a double-channel observation model of the distributed passive radar; S3, constructing a high-dimensional multi-parameter target position optimization problem based on monitoring channel observation; S4, constructing an external radiation source signal estimator based on reference channel observation; S5, constructing a signal-level low-complexity high-precision robust target position estimator combined with monitoring channel and reference channel observation, and acquiring the maximum likelihood estimation of the target position through multi-stage grid search. The estimator designed by the application can realize signal-level low-complexity high-precision robust target positioning in a direct-path interference environment, and can be applied to target positioning of the distributed passive radar and the like.

Description

Technical Field

[0001] This invention belongs to the field of passive radar target localization, and specifically relates to a distributed passive radar signal-level low-complexity, high-precision, robust localization method with direct path interference suppression suitable for non-cooperative external radiation source scenarios. Background Technology

[0002] Passive radar is a new type of radar that does not radiate electromagnetic signals itself, but detects targets by receiving signals from external radiation sources reflected from the target. Unlike active radar, which actively radiates electromagnetic signals, passive radar has the advantages of low power consumption, low electromagnetic pollution, and strong stealth. Especially with the increasing use of signal sources such as digital audio / video broadcasting systems, global navigation satellite systems, and cellular base stations as non-cooperative external radiation sources, passive radar is widely used in military and civilian fields. Distributed passive radar consists of widely distributed multi-station passive radars, which can observe targets from multiple angles, suppress target RCS scintillation, and obtain spatial diversity gain. Compared to monostation passive radar, distributed passive radar can acquire more sampling points, accumulate a higher signal-to-noise ratio, and thus improve the detection performance of passive radar. As one of the typical radar detection tasks, target localization aims to accurately estimate the position parameters of the detected target, providing high-precision measurements for target tracking and situational awareness. Target localization has been extensively studied in different radar systems, but a good solution for localization algorithms suitable for distributed passive radar, especially a signal-level direct localization algorithm suitable for distributed passive radar affected by direct path interference, has yet to be found.

[0003] Specifically, target localization algorithms for distributed passive radar can be divided into two categories. The first category is the traditional two-step localization method. Each passive radar first estimates intermediate parameters about the target position, such as the echo angle of arrival and time difference of arrival. Then, the intermediate parameters from different passive radars are combined to establish a geometric equation about the target position, and the target position is calculated. Although this method has the advantages of low computational cost and low transmission bandwidth, the estimation error in the first step may propagate and accumulate in the position estimation in the second step, resulting in limited localization accuracy in low signal-to-noise ratio environments. In addition, data association in multi-target scenarios is also a challenge for two-step localization. The second category is the direct localization algorithm. With the improvement of computational efficiency and communication bandwidth, direct localization has received increasing attention. For distributed passive radar, direct localization refers to obtaining the maximum likelihood estimate of the target by directly maximizing the joint likelihood function constructed from the original received signals of multiple distributed passive radars. This algorithm avoids the estimation of intermediate parameters and data association, and can achieve better performance. Therefore, compared with two-step localization, direct localization has higher localization accuracy and robustness, especially in low signal-to-noise ratio and small sample point situations.

[0004] A major challenge in distributed passive radar target localization algorithms is the problem of target echo obscuring caused by interference from direct path signals (i.e., signals received by the passive radar directly from external radiation sources). Since direct path signals do not undergo target reflection or attenuation, they contain no target information and are typically tens or even hundreds of decibels higher than the target echo, thus obscuring the target echo; this is known as direct path interference. The impact of direct path interference on passive radar is significant; some literature demonstrates that it can reduce the detection probability of the generalized likelihood ratio passive tester and increase the parameter estimation error of the passive radar.

[0005] In addition to the surveillance channel used to receive target echoes, each passive radar is typically equipped with a reference channel pointing towards the external radiation source to receive the direct path signal from the external radiation source as a reference signal. The introduction of the reference channel is considered one of the effective ways for passive radar to solve the problems of unknown signals from non-cooperative radiation sources and direct path interference.

[0006] The paper "Joint Delay and Doppler Estimation for Passive Sensing With Direct-Path Interference" (IEEE Transactions on Signal Processing, 2016, vol. 64, no. 3, pp. 630-640) obtains maximum likelihood estimates of target delay and Doppler by jointly using observations from the surveillance and reference channels of a single passive radar, while suppressing direct path interference. The paper "Direct path interference suppression for short-range passive bistatic synthetic aperture radar imaging based on atomic norm minimisation and Vandermonde decomposition" (IET Radar, Sonar & Navigation, 2019, vol. 13, no. 7, pp. 1171-1179) utilizes passive radar for target imaging and proposes estimating the delay and amplitude of direct path interference by solving atomic norm and least-squares optimization problems, respectively. These works are primarily applicable to single passive radars and cannot be directly extended to distributed passive radars because distributed passive radars involve more complex system geometries and more unknown parameters. The paper "Multistatic detection for passive radar with direct-path interference, IEEE Transactions on Aerospace and Electronic Systems, 2017, vol. 53, no. 2, pp. 915-925" proposes target detection based on distributed passive radar. While this work is applicable to distributed passive radar, it primarily addresses target detection and cannot be directly extended to target localization. Patent application "A distributed passive radar system and its target localization method, CN108519586A" proposes first estimating the time difference of arrival of the distributed passive radar for the same target, and then using these parameters to locate the target. This localization method is essentially a two-step localization method, which has limited performance in low signal-to-noise ratio environments and is weaker than direct localization algorithms. In summary, existing passive radar correlation detection and parameter estimation algorithms are not fully applicable to the direct target localization problem of distributed passive radar. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide a robust distributed passive radar signal-level positioning method that can achieve low complexity and high accuracy in direct path interference environments.

[0008] The objective of this invention is achieved through the following technical solution: a robust distributed passive radar signal-level localization method with direct path interference suppression, comprising the following steps:

[0009] S1. System parameter initialization: Initialize the number N and location coordinates of external radiation sources. The superscript (·) T Represents transpose; Number M and location coordinates of distributed passive radar nodes. Target position coordinates p = [x, y] T For a bistatic pair consisting of the nth external radiation source and the mth passive radar, the signal amplitudes of the direct path interference and the target echo in its monitoring channel are respectively expressed as α. m,n and β m,n The signal amplitude of the direct path signal in its reference channel is represented by γ. m,n The passive radar has K sampling points and a sampling period of T. s The noise introduced by each passive radar surveillance channel and reference channel is independent of each other, with a mean of 0 and a variance of . Complex circular symmetric Gaussian white noise;

[0010] S2. Establish a dual-channel observation model for distributed passive radar;

[0011] S3. Construct a high-dimensional, multi-parameter target location optimization problem based on surveillance channel observations;

[0012] S4. Construct an external radiation source signal estimator based on reference channel observations;

[0013] S5. Construct a signal-level, low-complexity, high-precision, robust target position estimator based on joint monitoring and reference channel observations, and obtain the maximum likelihood estimate of the target position through multi-level grid search.

[0014] The advantages of this invention are as follows: By considering direct path interference in the surveillance channel that can overwhelm the target echo during the design phase, the resulting target position optimization problem includes not only a correlation-based position estimator but also a direct path interference canceller. The latter can suppress direct path interference in the likelihood function of the surveillance channel, thereby highlighting the target echo and obtaining a high-precision and robust target position estimate. The introduction of a reference channel helps to obtain the maximum likelihood estimate of the signal from an unknown external radiation source, thus transforming the high-dimensional optimization problem into a low-dimensional optimization problem, significantly reducing the computational complexity of the algorithm. Based on this, the estimator designed in this invention can achieve signal-level low-complexity, high-precision, and robust target localization in direct path interference environments, and can be applied to fields such as target localization in distributed passive radar. Attached Figure Description

[0015] Figure 1 This is a flowchart of the positioning algorithm based on distributed passive radar of the present invention;

[0016] Figure 2 A two-dimensional scene diagram for target localization by distributed passive radar;

[0017] Figure 3 This is a likelihood plane diagram of the signal-level position estimator of the present invention;

[0018] Figure 4 The positioning error of the signal-level position estimator of this invention varies with the intensity of direct path interference in the monitoring channel;

[0019] Figure 5 The positioning error of the signal-level position estimator of this invention varies with the target echo intensity in the surveillance channel;

[0020] Figure 6 The positioning error of the signal-level position estimator of this invention varies with the signal strength of the direct path of the reference channel. Detailed Implementation

[0021] To facilitate the description of the present invention, the following terms will first be explained:

[0022] Term 1: Passive radar: A radar that does not radiate electromagnetic signals itself, but detects targets by observing the signals of external radiation sources reflected by the target.

[0023] Term 2, Distributed Passive Radar: Each passive radar node is far enough apart to observe the target from multiple angles and obtain spatial diversity gain.

[0024] Term 3, Reference Channel: A passive radar receiving channel that points directly to an external radiation source to receive direct path signals.

[0025] Term 4, Surveillance Channel: A passive radar receiving channel that refers to the surveillance area and is used to receive target echoes.

[0026] Term 5, Direct Path Interference: Signals from external radiation sources that are superimposed on the target echo in the surveillance channel.

[0027] Term 6, Cramero bound: Lower bound on the variance of any unbiased estimator.

[0028] This invention proposes a robust distributed passive radar signal-level localization method for suppressing direct path interference. First, a target echo model contaminated by noise and direct path interference is established in the surveillance channel, and a direct path signal model contaminated by noise is established in the reference channel. Second, a high-dimensional optimization problem concerning the target position and the unknown external radiation source signal is established by combining the surveillance channel observation model of the distributed passive radar. This problem includes a correlation-based target position estimator and a direct path interference canceller. Then, the maximum likelihood estimate of the external radiation source signal is obtained using the direct path signal observed from the reference channel. Finally, the unknown parameter dimension of the high-dimensional optimization problem of the surveillance channel is reduced by combining the reference channel observations, and a multi-level grid search is used to estimate the maximum likelihood of the target position. The technical solution of this invention is further illustrated below with reference to the accompanying drawings.

[0029] This invention is primarily verified using simulation experiments; all steps and conclusions have been verified as correct using Matlab R2020b. Figure 1 As shown, the distributed passive radar signal-level robust localization method for direct path interference suppression of the present invention includes the following steps:

[0030] S1. System parameter initialization: Initialize the number N and location coordinates of external radiation sources. The superscript (·) T Represents transpose; Number M and location coordinates of distributed passive radar nodes. Target position coordinates p = [x, y] T For a bistatic pair consisting of the nth external radiation source and the mth passive radar, the signal amplitudes of the direct path interference and the target echo in its monitoring channel are respectively expressed as α. m,n and β m,n The signal amplitude of the direct path signal in its reference channel is represented by γ. m,n The passive radar has K sampling points and a sampling period of T. s The noise introduced by each passive radar surveillance channel and reference channel is independent of each other, with a mean of 0 and a variance of . Complex circular symmetric Gaussian white noise.

[0031] This embodiment simulates a scenario where distributed passive radar locates a stationary target within a two-dimensional region, such as... Figure 2 As shown. Initialize system parameters: The number of external radiation sources is N=2, located at... The number of passive radars is M=4, located at... The target is located at p = [0, 12] T Km; the radiation source signal is a linear frequency modulated signal with a carrier frequency of 9.7 GHz and a bandwidth of 2 MHz; the number of sampling points K = 128, and the sampling period is 0.25 μs.

[0032] S2. Establish a dual-channel observation model for distributed passive radar. Traditional passive radar target localization algorithms typically only consider noise-contaminated target echoes in the surveillance channel, neglecting direct path interference in the same frequency band as the target echo. This makes the localization algorithm incapable of operating in environments with strong direct path interference. Therefore, this invention considers that in addition to noise-contaminated target echoes, the passive radar surveillance channel observation also includes direct path interference; the reference channel observation includes noise-contaminated direct path signals. The specific method for this step is as follows: The direct path signal received by the m-th passive radar reference channel from the n-th external radiation source is:

[0033]

[0034] Among them, the superscript (·) d Indicates the direct path, γ m,n It is the complex amplitude of the direct path signal, which takes into account antenna gain and channel propagation; The baseband signal of the nth external radiation source at time... Time sampling;

[0035] It is the path propagation delay from the nth external radiation source to the mth passive radar, and c0 is the electromagnetic wave propagation speed; It's kT s The additive noise at time step is modeled as having a variance of A complex circularly symmetric white Gaussian random variable;

[0036] The signal received by the m-th passive radar monitoring channel from the n-th external radiation source is:

[0037]

[0038] Among them, the superscript (·) e Indicates the echo path, α m,n It is the complex amplitude of the direct path signal, which takes into account antenna gain and channel propagation; β m,n It is the complex amplitude of the target echo, which takes into account antenna gain, channel propagation, and target reflectivity;

[0039] It is the path propagation delay from the nth external radiation source through the target to the mth passive radar; It's kT sThe additive noise at time step is modeled as having a variance of A complex circularly symmetric white Gaussian random variable;

[0040] Assume that the reference channel noise and the surveillance channel noise are independent of each other and are spatially and temporally independent, that is:

[0041]

[0042] Among them, the superscript (·) d / e This indicates either a direct path or an echo path, meaning that either the direct path or the echo path satisfies the above formula; (symbol) (·) * δ(·) and δ(·) represent the expectation, conjugate, and impulse functions, respectively;

[0043] The observation vector consisting of K sampling points from the reference channel observation (1) and the surveillance channel observation (2) is represented as:

[0044]

[0045] in,

[0046]

[0047] Through discrete Fourier transform, the frequency domain expression of the time-domain observation vector in the above equation is:

[0048]

[0049] in, f Δ =1 / (T) s K); without loss of generality, assume ||s n || 2 =1; ||·|| represents the 2-norm; the parameter without a horizontal line above the symbol in formula (6) is the frequency domain expression of the corresponding parameter in formula (4).

[0050] Finally, the joint observations of the distributed passive radar reference channel and surveillance channel, using MN bistatic pairs, are expressed as follows:

[0051]

[0052]

[0053] In this embodiment, the chirp function of Matlab R2020b is used to generate a linear frequency modulated continuous wave; the Gaussian random noise of the monitoring channel and the reference channel is generated using the randn function, and its variance is... It remains unchanged.

[0054] S3. Construct a high-dimensional multi-parameter target position optimization problem based on surveillance channel observation; the specific method is as follows: establish a joint likelihood function based on the surveillance channel observation of distributed passive radar, and obtain the maximum likelihood estimate of the likelihood function with respect to the direct path interference amplitude and the target echo amplitude; then substitute the estimated amplitude into the likelihood function to construct a high-dimensional multi-parameter optimization problem with the target position and the external radiation signal waveform as unknown variables. This problem not only includes a target position estimator based on correlation, but also a direct path interference canceller; the specific calculation process is as follows: the joint likelihood function of the distributed passive radar surveillance channel observation (8) is expressed as:

[0055]

[0056] in, This is the observation of the m-th, n-th bistatic pair monitoring channel. The likelihood function is expressed as:

[0057]

[0058] In equation (10), the signal amplitude α m,n and β m,n Maximum likelihood estimation and Represented as:

[0059]

[0060] in symbol(·) H Indicates the conjugate transpose; the estimate obtained from equation (11) and Substitute α into equation (10) m,n and β m,n ,get:

[0061]

[0062] Inverse of a second-order matrix Represented as:

[0063]

[0064] Substituting equation (13) into equation (12), we get:

[0065]

[0066] in,

[0067]

[0068] Equation (14) Substituting into equation (9), we get:

[0069]

[0070] in, It is a constant independent of the unknown parameters; in the expansion (16) get:

[0071]

[0072] As can be seen from equation (17), the likelihood function based on surveillance channel observation consists of three parts: a constant term independent of the target position, a correlation-based position estimator, and a direct path interference canceller; when estimating the echo delay... Delay of direct route When they are equal, then we have In equation (17), the sum of the constant term, the position estimator, and the direct path interference canceller is equal to 0, meaning that the likelihood function value on the direct path is eliminated; this process does not affect the likelihood function value of the target echo.

[0073] The unknown parameters of the likelihood function (16) based on surveillance channel observations include the target position p and the external radiation source signal s. n Therefore, the multi-parameter optimization problem for target location estimation is expressed as:

[0074]

[0075] in, The optimization problem described above is a high-dimensional optimization problem of 2+2NK dimensions, which is computationally infeasible.

[0076] S4. Construct an external radiation source signal estimator based on reference channel observations; the specific method is as follows: establish a joint likelihood function based on the reference channel observations of distributed passive radar, and obtain the maximum likelihood estimate of the direct path signal amplitude; then substitute the estimated amplitude into the likelihood function and simplify it to the form of the Rayleigh quotient. The maximum likelihood estimate of the unknown external radiation source signal is obtained through the maximum Rayleigh quotient. The specific calculation process is as follows: the joint likelihood function of the distributed passive radar reference channel observations (7) is expressed as:

[0077]

[0078] in, This is the m-th, n-th bistatic pair reference channel observation. The likelihood function is expressed as:

[0079]

[0080] Signal amplitude γ m,nThe maximum likelihood estimate is expressed as:

[0081]

[0082] Substituting equation (21) into equations (19) and (20), we get:

[0083]

[0084] Where E d It is a constant independent of the unknown parameters;

[0085]

[0086] The unknown parameters of the joint likelihood function based on reference channel observations contain only the external radiation source signal s. n Therefore, the estimate of the external radiation source signal is expressed as:

[0087]

[0088] This is a typical maximum Rayleigh quotient problem; the maximum likelihood estimate of the signal from an external radiation source is a matrix. The eigenvector corresponding to the largest eigenvalue.

[0089] S5. Construct a low-complexity, high-precision, robust target position estimator based on the signal level of joint monitoring channel and reference channel observations, and obtain the maximum likelihood estimate of the target position through multi-level grid search; the specific method is as follows: the external radiation source signal estimate obtained based on the reference channel observation in equation (24) is used. Substituting this into the high-dimensional target position estimator obtained from surveillance channel observations in equation (18), we have:

[0090]

[0091] Compared to the position estimator based solely on surveillance channel observations in Equation (18), the introduction of reference channel observations reduces the unknown parameters of the position estimator based on dual-channel observations in Equation (25) from 2+2NK dimensions to 2 dimensions, significantly reducing the computational complexity of the estimator.

[0092] Based on the maximum likelihood estimator of Equation (25), a 2D coarse grid search is first performed on the monitored area to find the grid point with the maximum likelihood function. This grid point is then used as the center point of the next-level grid search to perform a finer grid search on the surrounding area to find the grid point with the maximum likelihood function. This process is repeated until the grid spacing is small enough to meet the requirements of high-precision positioning. The final maximum likelihood point is the maximum likelihood estimate of the target location.

[0093] The root mean square error (RMSE) of the maximum likelihood estimate of the target location is calculated using the following formula:

[0094]

[0095] Where L is the Monte Carlo number, and in this embodiment, L = 500; For the target location estimation in the l-th Monte Carlo simulation;

[0096] In the surveillance channel, the intensity ratio of direct path interference to target echo and the intensity ratio of target echo to noise are defined as follows:

[0097]

[0098]

[0099] In the reference channel, the ratio of the direct path signal intensity to the noise intensity is defined as:

[0100]

[0101] The estimator designed in this embodiment is applicable to non-cooperative radiation source scenarios with unknown signals and considers direct path interference; therefore, it is called NDL-DPI (non-cooperative direct localization with direct-path interference). To demonstrate the localization performance of the invented estimator NDL-DPI, three comparative estimators are presented: estimator CDL-DPI (cooperative direct localization with direct-path interference), i.e., cooperative radiation source scenarios with known radiation source signals and considering direct path interference; estimator NDL-NDPI (non-cooperative direct localization with no direct-path interference), i.e., non-cooperative radiation source scenarios with unknown radiation source signals and not considering direct path interference; and estimator CDL-NDPI (cooperative direct localization with no direct-path interference), i.e., cooperative radiation source scenarios with known radiation source signals and not considering direct path interference.

[0102] C1. By traversing the 2D grid plane of the monitored area, the likelihood function planes of the estimators CDL-NDPI and NDL-DPI at ISR = 10dB were obtained, as follows: Figure 3As shown in the figure, comparing the two figures reveals that the maximum likelihood estimation point of the estimator NDL-DPI in the strong direct path interference environment is near the real target, while the maximum likelihood estimation point of the estimator CDL-NDPI is far from the real target location. The invented estimator can effectively suppress direct path interference and estimate the target location.

[0103] C2. Change the monitoring channel ISR and repeat the operation of this invention to obtain... Figure 4 .from Figure 4 It can be seen that as the direct path interference in the monitoring channel increases, the estimator NDL / CDL-NDPI quickly deviates from the Cramer-Rao boundary and fails to locate the target. In contrast, the invented estimator NDL-DPI can maintain convergence to the Cramer-Rao boundary even at higher ISRs, demonstrating good tolerance to direct path interference.

[0104] C3. Change the monitoring channel SNR and repeat the operation of this invention to obtain... Figure 5 .from Figure 5 It can be seen that as the signal-to-noise ratio (SNR) of the monitoring channel increases, the invented estimator NDL-DPI can converge to the Cramer-Rao bound first, and is therefore effective. However, the estimator NDL / CDL-NDPI cannot converge to the Cramer-Rao bound.

[0105] C4. Change the reference channel DNR and repeat the operation of this invention to obtain... Figure 6 .from Figure 6 It can be seen that as the reference channel signal-to-noise ratio (DNR) increases, the invented estimator NDL-DPI can gradually converge to the Cramer-Rao boundary. When the DNR is greater than 15dB, the positioning error of the NDL-DPI of the present invention is smaller than that of the estimator NDL / CDL-NDPI, and it gradually converges to the Cramer-Rao boundary.

[0106] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A robust distributed passive radar signal-level localization method with direct path interference suppression, characterized in that, Includes the following steps: S1. System parameter initialization: Initialize the number of external radiation sources. and location coordinates superscript Represents transposition; Number of distributed passive radar nodes and location coordinates Target location coordinates For the first External radiation sources and the first The signal amplitudes of direct path interference and target echo in the monitoring channel of a bistatic pair composed of passive radars are respectively expressed as: and The signal amplitude of the direct path signal in its reference channel is expressed as: ; The number of sampling points for passive radar is The sampling period is The noise introduced by each passive radar surveillance channel and reference channel is independent of each other, with a mean of 0 and a variance of . Complex circular symmetric Gaussian white noise; S2. Establish a dual-channel observation model for distributed passive radar; including: establishing a target echo model in the monitoring channel contaminated by noise and direct path interference, and a direct path signal model in the reference channel contaminated by noise. S3. Construct a high-dimensional multi-parameter target position optimization problem based on surveillance channel observations; including: establishing a joint likelihood function based on surveillance channel observations from distributed passive radars, and obtaining the maximum likelihood estimate of the likelihood function with respect to the direct path interference amplitude and the target echo amplitude; then substituting the estimated amplitude into the likelihood function to construct a high-dimensional multi-parameter optimization problem with the target position and the external radiation signal waveform as unknown variables. This problem includes not only a target position estimator based on correlation, but also a direct path interference canceller. S4. Construct an external radiation source signal estimator based on reference channel observations; including: establishing a joint likelihood function based on reference channel observations of distributed multi-passive radar, and obtaining the maximum likelihood estimate of the direct path signal amplitude; then substituting the estimated amplitude into the likelihood function and simplifying it to the form of the Rayleigh quotient; the maximum likelihood estimate of the unknown external radiation source signal is obtained through the maximum Rayleigh quotient. S5. Construct a signal-level, low-complexity, high-precision, robust target position estimator based on joint surveillance and reference channel observations, and obtain the maximum likelihood estimate of the target position through multi-level grid search; including: substituting the external radiation source signal estimate obtained based on reference channel observations into the high-dimensional target position estimator obtained based on surveillance channel observations; based on the maximum likelihood estimator, first performing a 2D coarse grid search on the surveillance area to find the grid point with the maximum likelihood function; using this grid point as the center point of the next level grid search, performing a more refined grid search on the surrounding area to find the grid point with the maximum likelihood function; repeating the above process until the grid spacing meets the high-precision positioning requirements; the final maximum likelihood point is the maximum likelihood estimate of the target position.

2. The distributed passive radar signal-level robust localization method for direct path interference suppression according to claim 1, characterized in that, The specific method for step S2 is as follows: The passive radar reference channel received data from the first... The direct path signal of the external radiation source is: （1）； Among them, superscript Indicates the direct path. It is the complex amplitude of the direct path signal; It is the first The baseband signal of an external radiation source at time Time sampling; From the first External radiation source to the first The path propagation delay of a passive radar. It is the speed of electromagnetic wave propagation; yes The additive noise at time step is modeled as having a variance of A complex circularly symmetric white Gaussian random variable; No. The passive radar surveillance channel received data from the first... The signal from the external radiation source is: （2）； Among them, superscript Indicates the echo path, It is the complex amplitude of the direct path signal; It is the complex amplitude of the target echo; From the first An external radiation source passes through the target to the first The path propagation delay of a passive radar; yes The additive noise at time step is modeled as having a variance of A complex circularly symmetric white Gaussian random variable; Assume that the reference channel noise and the surveillance channel noise are independent of each other and are spatially and temporally independent, that is: （3）； Among them, superscript Indicates direct path or echo path; symbol , and These represent the expectation, conjugate, and impulse functions, respectively. Reference channel observation (1) and surveillance channel observation (2) The observation vector consisting of sampling points is represented as: （4）； in, （5）； Through discrete Fourier transform, the frequency domain expression of the time-domain observation vector in the above equation is: （6）； in, , ; Assumption ; Represents the 2-norm; Ultimately, the alliance The joint observations of the bistatic pairs, distributed passive radar reference channel, and surveillance channel are respectively represented as: （7）； （8）。 3. The distributed passive radar signal-level robust localization method for direct path interference suppression according to claim 2, characterized in that, The specific method for step S3 is as follows: The joint likelihood function of the distributed passive radar surveillance channel observation (8) is expressed as: （9）； in, It is the first Two-base pair surveillance channel observations The likelihood function is expressed as: （10）； Signal amplitude in equation (10) and Maximum likelihood estimation and Represented as: （11）； in ,symbol Indicates the conjugate transpose; the result obtained by estimating equation (11) and Substitute into equation (10) and ,get: （12）； Inverse of a second-order matrix Represented as: （13）； Substituting equation (13) into equation (12), we get: （14）； in, ; （15）； Equation (14) Substituting into equation (9), we get: （16）； in, It is a constant independent of the unknown parameters; in the expansion (16) ,get: （17）； As can be seen from equation (17), the likelihood function based on surveillance channel observation consists of three parts: a constant term independent of the target position, a correlation-based position estimator, and a direct path interference canceller; when estimating the echo delay... Delay with direct route When they are equal, then we have In equation (17), the sum of the constant term, the position estimator, and the direct path interference canceller is equal to 0, meaning that the likelihood function value on the direct path is eliminated. The unknown parameters of the likelihood function (16) based on surveillance channel observations include the target location. and external radiation source signal Therefore, the multi-parameter optimization problem for target location estimation is expressed as: （18）； in, The above optimization problem is a High-dimensional optimization problems.

4. The distributed passive radar signal-level robust localization method for direct path interference suppression according to claim 3, characterized in that, The specific method for step S4 is as follows: The joint likelihood function of the distributed passive radar reference channel observation (7) is expressed as: （19）； in, It is the first Observations of a bistatic pair reference channel The likelihood function is expressed as: （20）； signal amplitude The maximum likelihood estimate is expressed as: （21）； Substituting equation (21) into equations (19) and (20), we get: （22）； in It is a constant independent of the unknown parameters; （23）； The unknown parameters of the joint likelihood function based on reference channel observations contain only signals from the external radiation source. Therefore, the estimate of the external radiation source signal is expressed as: （24）； This is a typical maximum Rayleigh quotient problem; the maximum likelihood estimate of the signal from an external radiation source is a matrix. The eigenvector corresponding to the largest eigenvalue.

5. The distributed passive radar signal-level robust localization method for direct path interference suppression according to claim 4, characterized in that, The specific method for step S5 is as follows: estimate the external radiation source signal obtained based on the reference channel observation in equation (24). Substituting this into the high-dimensional target position estimator obtained from surveillance channel observations in equation (18), we have: （25）； Compared to the position estimator in equation (18) which is based solely on surveillance channel observations, the introduction of reference channel observations reduces the unknown parameters of the position estimator in equation (25) which is based on dual-channel observations. Dimensions dropped to dimension; Based on the maximum likelihood estimator of equation (25), a 2D coarse grid search is first performed on the monitored area to find the grid point with the maximum likelihood function. This grid point is then used as the center point of the next-level grid search to perform a finer grid search on the surrounding area to find the grid point with the maximum likelihood function. This process is repeated until the grid spacing meets the requirements for high-precision positioning. The final maximum likelihood point is the maximum likelihood estimate of the target location. .

Images