Optimal reconnaissance node selection method for direct tracking of moving target

Abstract

The invention discloses an optimal reconnaissance node selection method for direct tracking of a moving target, and the method comprises the steps: carrying out the data collection of a target signal at each distributed reconnaissance node, configuring an omnidirectional antenna on each reconnaissance node, and building a signal receiving model based on the signal arrival time in a distributed reconnaissance scene; the method comprises the following steps: constructing a Cramer-Rao lower bound of a target tracking posterior aiming at a distributed interception system, and establishing a cost function of optimal node selection according to the Cramer-Rao lower bound; and carrying out convex relaxation on the cost function, and solving the cost function through a convex optimization tool so as to realize optimal node selection in the distributed target tracking scene. According to the method, the optimal reconnaissance node selection of target tracking in a distributed collaborative scene can be realized, and the target tracking precision is ensured while the system complexity is reduced.

Description

Technical Field

[0001] This invention belongs to the field of electronic reconnaissance technology, specifically relating to an optimal reconnaissance node selection method for direct tracking of moving targets. Background Technology

[0002] Distributed collaborative electronic reconnaissance systems utilize multi-node joint interception and fusion processing to analyze enemy target location information parameters implicit in signals, thereby achieving high-precision target positioning and real-time tracking. Compared to single-node reconnaissance methods, distributed collaborative reconnaissance systems offer advantages such as strong positioning capabilities, robustness, high fault tolerance, and high reconfigurability. Efficient and reliable positioning and tracking results facilitate accurate generation of battlefield intelligence, continuous awareness and rapid construction of battlefield situation, and gaining battlefield control, providing crucial support for ultimately dismantling the enemy's operational system.

[0003] In distributed collaborative reconnaissance networks, increasing the number of reconnaissance nodes can improve the system's battlefield information acquisition capabilities and overall stability, but it inevitably leads to increased communication load and energy consumption, and significantly increases system complexity. Furthermore, the relationship between the increase in the number of reconnaissance nodes and the final positioning and tracking error exhibits a submodal relationship; that is, as the node scale expands, the contribution of newly added nodes to improving positioning and tracking performance gradually weakens. Especially in moving target reconnaissance scenarios, due to the continuous changes in target trajectories, activating all nodes throughout the entire positioning and tracking cycle is often not the optimal choice. Some nodes far from the target have limited impact on improving positioning accuracy, but lead to a significant increase in system complexity and communication bandwidth. Therefore, in practical reconnaissance applications, a dynamic balance must be achieved between positioning accuracy and system resource constraints to effectively reduce system complexity and operating costs while ensuring the accuracy of battlefield intelligence. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, this invention provides an optimal reconnaissance node selection method for direct tracking of moving targets. Target signal data is collected at each distributed reconnaissance node, and each node is equipped with an omnidirectional antenna, thereby establishing a received signal model based on signal arrival time in a distributed reconnaissance scenario. For the distributed reconnaissance system, a posterior Cramer-Rao lower bound for target tracking is constructed, and a cost function for optimal node selection is established based on this. The cost function is then convexly relaxed and solved using a convex optimization tool, thereby achieving optimal node selection in a distributed target tracking scenario. This invention enables optimal reconnaissance node selection for target tracking in distributed cooperative scenarios, reducing system complexity while maintaining target tracking accuracy.

[0005] The technical solution adopted by this invention to solve its technical problem is as follows: Step 1: Each distributed reconnaissance node collects data on the target signal, thereby establishing a received signal model based on signal arrival time in a distributed reconnaissance scenario; Step 2: For the received signals of each reconnaissance node, establish the likelihood function about the target position, and derive the data Fisher information matrix from the measurement data of each station; Step 3: Combining the Fisher information matrix derived from the measurement data and the moving target observation model, derive the Bayesian Fisher information matrix for target tracking; Step 4: Based on the Bayesian Fisher information matrix, establish the cost function for node selection; Step 5: Perform convex relaxation on the node selection cost function, and use convex optimization tools to solve the cost function to obtain the optimal node selection result; Step 6: Repeat steps 1 to 5 throughout the entire moving target tracking process, selecting the optimal reconnaissance node in each positioning process to achieve precise scheduling of reconnaissance resources and global optimization of system performance.

[0006] Preferably, step 1 specifically comprises: Step 1-1: Suppose that there are distributed in a two-dimensional space The reconnaissance receiving node, of which the first l The spatial location of each receiving node is Each receiving node is equipped with an omnidirectional antenna; At what moment, the spatial position of the moving radiation source target in space is: ;therefore, At that moment, the Received signals at each reconnaissance node Represented as: (1) in, for The target signal is transmitted to the first time. l Channel coefficients at individual stations express The target signal complex envelope at time t, express Time of the first l Noise on each reconnaissance node; express The time signal is transmitted from the target radiation source to the first The transmission delay of each reconnaissance node is specifically expressed as follows: (2) in, Indicates signal transmission speed. This indicates that the vectors in the vector are taken as the second norm.

[0007] Steps 1-2: Continuously received signals conduct Point sampling is performed, and a Discrete Fourier Transform is applied to the sampled discrete signal. n The discrete signal model at each frequency point is represented as: (3) in, The first discrete Fourier transform n One frequency point, It is a natural constant. The imaginary unit, , , Receiving signals respectively Target signal envelope ,noise The frequency domain discrete form; Steps 1-3: Receive the signal conduct Point accumulation, expressed in vector form, yields: (4) in, (5) (6) (7) (8) in, This indicates taking the transpose of the vector within the parentheses. for Time of the first The received signal vector at each receiving station The time delay feature matrix, for The envelope vector of the signal at any given time. for Time of the first Noise vectors at each receiving station.

[0008] Preferably, step 2 specifically comprises: Step 2-1: Define the vector representation of unknown parameters in the distributed reconnaissance system as follows: , express The set of signals received by all reconnaissance nodes at any given time, and the likelihood function of the unknown term. Represented as: (9) in, Indicates the cumulative multiplication symbol. For noise variance, This indicates that an exponentiation operation is performed on the element within the parentheses; definition for The likelihood function with respect to the unknown term at time t, for the likelihood function Taking the natural logarithm, we have: (10) in, , All are constants; Maximize the log-likelihood function , equivalent to minimizing ; Step 2-2: In an unknown scenario, assume the received signal It follows a complex Gaussian distribution. Time of the first The average value of the received signal at each reconnaissance node is expressed as: ,have Its variance is , express A 3D identity matrix; therefore, Time of the first Fisher information matrix of signals received by each reconnaissance node Represented as: (11) in, The sign of the partial derivative; Steps 2-3: Based on the received signal distribution, the Fisher information matrix regarding the target location is further represented as: (12) in, This indicates taking the conjugate transpose of the elements within the parentheses. This indicates that the real part of the element within the parentheses is taken; Furthermore, (13) because (14) (15) Define a diagonal matrix based on angular frequencies: (16) have (17) Therefore, signal mean The partial derivative with respect to the current target position is expressed as: (18) in, (19) (20) Therefore, the Fisher information matrix regarding the target location is represented as: (twenty one) Fisher information matrix of joint data regarding target location for all reconnaissance nodes. Represented as: (twenty two).

[0009] Preferably, step 3 specifically comprises: Step 3-1: At any given time, the target's state information can be represented as: (twenty three) in, For the goal Location information at any given time For the goal Speed ​​information at any given moment; Step 3-2: The state transition model for the target motion is as follows: (twenty four) in, For the purpose of movement The process noise at time t has a mean of 0 and a variance of . Gaussian distribution; The state transition matrix is ​​expressed as: (25) in, The interval between two consecutive observations; Step 3-3: In the scenario of moving target tracking, Bayesian Fisher Information Matrix at a given time It consists of two parts, including the Fisher information matrix determined by measurement data. and the prior information matrix determined by the target motion state model It consists of two parts: (26) Moment-by-moment Bayesian Fisher information matrix The iterative expression is written as: (27) in, (28) (29) (30) in, Represented as: (31) in, express A matrix of all zeros; The iterative expression for the Bayesian Fisher information matrix is ​​further expressed as: (32) Among them, the prior information matrix determined by the target motion state model Represented as: (33).

[0010] Preferably, step 4 specifically comprises: Step 4-1: The posterior Cramer-Rao lower bound of the target parameter estimation satisfies: (34) in, This indicates that the mean of the elements is calculated. express target status at any time The estimated value; By minimizing the Bayesian Fisher information matrix The inverse of the trajectory is used to achieve the optimal selection of reconnaissance nodes for moving target tracking; Step 4-2: Definition At that moment, the The selection variables for each reconnaissance node are: , Indicates the first One reconnaissance node was not selected and did not participate in the target trajectory tracking data processing. Indicates the first Several reconnaissance nodes have been selected to participate in the subsequent target trajectory output process; the node selection vector for all receiving stations is defined as follows. ,have: (35) Assume the number of nodes selected at each tracking time is Then we have: (36) in, express A 1-dimensional vector; In the node selection scenario, the Bayesian Fisher information matrix is ​​further represented as a function of the variables. The functions are: (37) Therefore, the cost function for node selection is expressed as: (38) in, This indicates taking the trace of the matrix within the parentheses. Indicates to Optimize to minimize the value of the element within the parentheses.

[0011] Preferably, step 5 specifically comprises: Boolean constraints involved in the optimization problem Relaxing continuous variable constraints, we have (39) Introducing symmetric positive definite matrices ,satisfy: (40) in, Indicates the positive definite sign; Therefore, Therefore, the original problem is equivalent to: (41) Constraints Using Schul complement theorem, we can write the following linear matrix equation: (42) The dimension is The identity matrix; At this point, the optimization problem is transformed into: (43) First, the optimal result output by the solver is... Sort by size in descending order, and take the first few. Set the largest value to 1, and set the rest to 1. Each value is 0, thus obtaining the Boolean value for optimal node selection. This allows us to determine the optimal node distribution for the current system moment.

[0012] Preferably, the With a mean of 0 and a variance of Additive complex Gaussian white noise.

[0013] An electronic device includes: a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to enable the electronic device to perform the above-described optimal reconnaissance node selection method.

[0014] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the above-described optimal reconnaissance node selection method.

[0015] A chip includes a processor for retrieving and running a computer program from a memory, causing a device equipped with the chip to perform the above-described optimal reconnaissance node selection method.

[0016] A computer program product includes a computer storage medium storing a computer program, the computer program including instructions executable by at least one processor, which, when executed by the at least one processor, implement the above-described optimal reconnaissance node selection method.

[0017] The beneficial effects of this invention are as follows: This invention employs a distributed collaborative reconnaissance architecture, enhancing the robustness, reconfigurability, and fault tolerance of the reconnaissance system, and improving the accuracy of battlefield intelligence acquisition. Furthermore, in moving target tracking scenarios, this invention establishes a tracking performance evaluation standard through a Bayesian information matrix, thereby deriving a posterior Cramer-Rao lower bound for target tracking. This integrates prior information about target state transitions and information from measurement data to comprehensively evaluate the target tracking performance of the reconnaissance system. Finally, this invention establishes a cost function for reconnaissance node selection by minimizing the posterior Cramer-Rao lower bound, and transforms the non-convex optimization problem into a convex optimization problem through convex relaxation, achieving optimal reconnaissance node selection for target tracking in distributed collaborative scenarios, reducing system complexity while ensuring target tracking accuracy. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention.

[0019] Figure 2 This is a schematic diagram of a distributed collaborative reconnaissance scenario involved in the method of the present invention.

[0020] Figure 3 This is the optimal node selection scheme generated by the method of the present invention during the target tracking process.

[0021] Figure 4 This is a comparison of the tracking error of the present invention method and the random node selection method for moving targets. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] The purpose of this invention is to provide an optimal reconnaissance node selection method for direct tracking of moving targets, and to construct a dynamic optimization framework for node selection to solve problems such as high node communication load, high computational load of central nodes, and complex system integration in current distributed cooperative electronic reconnaissance systems. This invention adopts a distributed cooperative receiver positioning system, and establishes an optimal reconnaissance node selection cost function by minimizing the posterior Cramer-Rao lower bound for target tracking. Then, a convex optimization tool is used to solve the cost function to obtain the optimal node selection result, thereby reducing system complexity while improving target tracking accuracy in distributed reconnaissance.

[0024] The basic idea behind this invention is as follows: each distributed reconnaissance node collects data on the target signal, and each reconnaissance node is equipped with an omnidirectional antenna, thereby establishing a received signal model based on the signal arrival time in a distributed reconnaissance scenario; for the distributed reconnaissance system, a posterior Cramero lower bound for target tracking is constructed, and a cost function for optimal node selection is established based on this; the cost function is convexly relaxed, and the cost function is solved using a convex optimization tool, thereby achieving optimal node selection in a distributed target tracking scenario.

[0025] To achieve the aforementioned objectives, the present invention employs the following technical solution: an optimal reconnaissance node selection method for direct tracking of moving targets, comprising the following steps: Step 1: Each distributed reconnaissance node collects data on the target signal, thereby establishing a received signal model based on signal arrival time in a distributed reconnaissance scenario.

[0026] Assume that there are distributed in a two-dimensional space The reconnaissance receiving node, of which the first l The spatial location of each receiving node is Each receiving node is equipped with an omnidirectional antenna; At what moment, the spatial position of the moving radiation source target in space is: .therefore, At that moment, the Received signals at each reconnaissance node Represented as: (1) in, for The target signal is transmitted to the first time. l Channel coefficients at individual stations express The target signal complex envelope at time. express Time of the first l The noise at each reconnaissance node is set to have a mean of 0 and a variance of in this invention. Additive complex Gaussian white noise. express The time signal is transmitted from the target radiation source to the first The transmission delay of each reconnaissance node is specifically expressed as follows: (2) in, Indicates signal transmission speed. This indicates that the vectors in the vector are taken as the second norm.

[0027] For continuous received signals conduct Point sampling is performed, and a Discrete Fourier Transform is applied to the sampled discrete signal. n The discrete signal model at each frequency point can be represented as: (3) in, The first discrete Fourier transform n One frequency point, It is a natural constant. The imaginary unit, , , Receiving signals respectively Target signal envelope ,noise The frequency domain discrete form.

[0028] Receive signal conduct Point accumulation, expressed in vector form, yields: (4) in, (5) (6) (7) (8) in, This indicates that the vector inside the brackets is transposed. for Time of the first The received signal vector at each receiving station The time delay feature matrix, for The envelope vector of the signal at any given time. for Time of the first Noise vectors at each receiving station.

[0029] Step 2: For the received signals of each reconnaissance node, establish the likelihood function about the target location, and derive the data Fisher information matrix from the measurement data of each station.

[0030] Define the unknown parameter vector representation in a distributed reconnaissance system as: , express The set of signals received by all reconnaissance nodes at any given time, and the likelihood function of the unknown term. It can be represented as; (9) in, Indicates the cumulative multiplication symbol. For noise variance, This indicates that exponentiation is performed on the elements within the parentheses. (Definition) for The likelihood function with respect to the unknown term at time t, for the likelihood function Taking the natural logarithm, we have; (10) in, , All are constants. Maximize the log-likelihood function. , equivalent to minimizing .

[0031] In an unknown scenario, assuming the received signal It follows a complex Gaussian distribution. Time of the first The average value of the received signal at each reconnaissance node is expressed as: ,have Its variance is , express A dimensional identity matrix. Therefore, Time of the first Fisher information matrix of signals received by each reconnaissance node Represented as; (11) in, Let be the partial derivative sign. Based on the received signal distribution, the Fisher information matrix regarding the target location can be further expressed as: (12) in, This indicates taking the conjugate transpose of the elements within the parentheses. This indicates that the real part of the element within the parentheses is taken. Furthermore, (13) because (14) (15) Defined by the diagonal matrix of angular frequencies (16) have (17) Therefore, signal mean The partial derivative with respect to the current target position can be expressed as: (18) in, (19) (20) Therefore, the Fisher information matrix regarding the target location can be represented as: (twenty one) Fisher information matrix of joint data regarding target location for all reconnaissance nodes. It can be represented as: (twenty two) Step 3: Combining the Fisher information matrix derived from the measurement data and the moving target observation model, derive the Bayesian Fisher information matrix for target tracking.

[0032] At any given time, the target's state information can be represented as: (twenty three) in, For the goal Location information at any given time For the goal Speed ​​information at any given moment.

[0033] The state transition model for the target motion is as follows: (twenty four) in, For the purpose of movement The process noise at time t has a mean of 0 and a variance of . The Gaussian distribution. The state transition matrix can be represented as: (25) The second is the interval between two consecutive observations.

[0034] In moving target tracking scenarios, the Bayesian Fisher information matrix consists of two parts, including the Fisher information matrix determined by measurement data. and the prior information matrix determined by the target motion state model It consists of two parts. (26) At time t, the iterative expression for the Bayesian Fisher information matrix can be written as: (27) in, (28) (29) (30) in, Represented as: (31) in, express A matrix of all zeros. The iterative expression for the Bayesian Fisher information matrix can be further expressed as: (32) Among them, the prior information matrix determined by the target motion state model It can be represented as: (33) Step 4: Based on the Bayesian Fisher information matrix, establish the cost function for node selection.

[0035] The posterior Cramerlow lower bound of the target parameter estimation satisfies: (34) in, This indicates that the mean of the elements is calculated. express target status at any time The estimated value. Therefore, it can be obtained by minimizing the Bayesian Fisher information matrix. The optimal reconnaissance node selection is achieved by reversing the trajectory of the target.

[0036] definition At that moment, the The selection variables for each reconnaissance node are: , Indicates the first One reconnaissance node was not selected and did not participate in the target trajectory tracking data processing. Indicates the first Several reconnaissance nodes have been selected to participate in the subsequent target trajectory output process. The node selection vector for all receiving stations is defined as follows: ,have: (35) Assume the number of nodes selected at each tracking time is Then we have: (36) in, express A one-row vector. In node selection scenarios, the Bayesian Fisher information matrix can be further represented as a vector about the variables. The functions are: (37) Therefore, the cost function for node selection can be expressed as: (38) in, This indicates taking the trace of the matrix within the parentheses.

[0037] Step 5: Perform convex relaxation on the node selection cost function, and use convex optimization tools to solve the cost function to obtain the optimal node selection result.

[0038] Boolean constraints involved in the optimization problem For relaxation of continuous variable constraints, we have: (39) Introducing symmetric positive definite matrices ,satisfy: (40) in, This represents the positive definite sign. Therefore, we can conclude that... Therefore, the original problem is equivalent to: (41) Constraints Using Schul complement theorem, we can write the following linear matrix equation: (42) The dimension is The identity matrix is ​​. At this point, the optimization problem is transformed into: (43) This problem has a convex cost function and constraint equations, which can be solved using the convex optimization toolbox. Since the original discrete variables... Since convex relaxations are continuous variables, the optimal result output by the solver needs to be optimized. Discretization is performed to obtain the optimal node selection result. This invention first analyzes the optimal result output by the solver. Sort by size in descending order, and take the first few. Set the largest value to 1, and set the rest to 1. Each value is 0, thus obtaining the Boolean value for optimal node selection. This allows us to determine the optimal node distribution for the current system moment.

[0039] Throughout the moving target tracking process, steps 1 to 5 of this invention are repeated, and the optimal reconnaissance node is selected in each positioning process to achieve precise scheduling of reconnaissance resources and global optimization of system performance.

Claims

1. An optimal scout node selection method for moving target direct tracking, characterized in that, The method comprises the following steps: Step 1: Each distributed reconnaissance node collects data of a target signal, thereby establishing a received signal model based on signal arrival time in a distributed reconnaissance scene; Step 2: A likelihood function about the target position is established for the received signal of each reconnaissance node, and a data Fisher information matrix is derived from the measurement data of each station; Step 3: A Bayesian Fisher information matrix of target tracking is derived by combining the Fisher information matrix derived from the measurement data and a motion target observation model; Step 4: A cost function of node selection is established according to the Bayesian Fisher information matrix; Step 5: The node selection cost function is convexly relaxed, and a convex optimization tool is used to solve the cost function to obtain the result of optimal node selection; Step 6: In the whole motion target tracking process, steps 1 to 5 are repeated, and the optimal reconnaissance node is selected in each positioning process, so as to realize accurate scheduling of reconnaissance resources and global optimization of system performance.

2. The optimal scout node selection method for moving target direct tracking according to claim 1, wherein, The step 1 is specifically: Step 1-1: Assume that there are surveillance receiving nodes distributed in a two-dimensional space, wherein the spatial positions of the l receiving nodes are , and an omnidirectional antenna is installed on each surveillance receiving node; at time , the spatial position of a moving radiation source target in the space is ; therefore, at time , the received signal at the surveillance receiving node is represented as: （1） in, for The target signal is transmitted to the first time. l Channel coefficients at individual stations express The target signal complex envelope at time t, express Time of the first l Noise on each reconnaissance node; express The time signal is transmitted from the target radiation source to the first The transmission delay of each reconnaissance node is specifically expressed as follows: （2） wherein denotes the signal transmission speed, denotes the two-norm of the vector Steps 1-2: Continuous reception of signals Proceeding Point sampling and Discrete Fourier Transform of the sampled discrete signal, the model of the discrete signal at the first n frequency point is expressed as: （3） wherein is the number of frequency bins of the discrete Fourier transform, n is the number of frequency bins, is a natural constant, is the imaginary unit, , , are the frequency domain discrete forms of the received signal , the target signal envelope , the noise , respectively; Steps 1-3: The received signal is performed point accumulation, and written in vector form, there is: （4） Wherein, （5） （6） （7） （8） in, This indicates taking the transpose of the vector within the parentheses. for Time of the first The received signal vector at each receiving station The time delay feature matrix, for The envelope vector of the signal at any given time. for Time of the first Noise vectors at each receiving station.

3. The optimal scout node selection method for moving target direct tracking according to claim 2, wherein, The step 2 is specifically: Step 2-1: Define the unknown parameter vector in the distributed reconnaissance system as , represents the set of all reconnaissance nodes receiving signals at time t, the likelihood function about the unknown term is represented as: （9） wherein represents a cumulative multiplication symbol, is the noise variance, represents an exponential operation on the elements within the parentheses; Definitions To the likelihood function for the unknown terms at time t, L(t) = P(Y(t) | X(t), θ), where Y(t) is the observed data at time t, X(t) is the unknown terms at time t, and θ is the model parameters. Taking the natural logarithm, we have （10） wherein , are constants; maximizing the log-likelihood function , which is equivalent to minimizing ; Step 2-2: Let the received signal obeys a complex Gaussian distribution, The mean of the received signal at the kth surveillance node is denoted as , The variance of the received signal is denoted as , denotes the kth surveillance node receives the signal The Fisher information matrix of the received signal at the kth surveillance node is denoted as: （11） wherein is the partial derivative symbol; Step 2-3: According to the received signal distribution, the Fisher information matrix about the target position is further expressed as: （12） wherein denotes taking the conjugate transpose of the elements within the parentheses, denotes taking the real part of the elements within the parentheses; Further, （13） Since （14） （15） The diagonal matrix is defined by the angular frequency: （16） There is （17） Thus, the signal mean The partial derivative with respect to the current target position is given by （18） Wherein, （19） （20） Therefore, the data Fisher information matrix about the target position is expressed as: （21） Fisher information matrix for the joint data on the target position for all scout nodes is represented as: （22）。 4. The optimal scout node selection method for moving target direct tracking according to claim 3, wherein, The step 3 is specifically: Step 3-1: At the moment, the state information of the target can be expressed as: （23） wherein, is the position information at the time instant is the velocity information at the time instant is the position information at the time instant is the velocity information at the time instant Step 3-2: The state transition model of target motion is: （24） wherein, is the process noise of the motion target at time instant, obeying a Gaussian distribution with mean 0 and variance is the state transition matrix, expressed as:​​ （25） wherein is the interval between two consecutive observations; Step 3-3: In the moving target tracking scenario, Bayesian Fisher information matrix at time instant consists of two parts, including the Fisher information matrix determined by the measurement data and the prior information matrix determined by the target motion state model consists of two parts: （26） Fisher information matrix The moment-based representation is written as: （27） Wherein, （28） （29） （30） wherein is represented by: （31） wherein represents identity 0 matrix; The iterative expression of the Bayesian Fisher information matrix is further expressed as: （32） wherein the a priori information matrix determined by the target motion state model is represented as: （33）。 5. The optimal scout node selection method for moving target direct tracking according to claim 4, wherein, The step 4 is specifically: Step 4-1: The posterior Cramer-Rao lower bound of target parameter estimation satisfies: （34） wherein, denotes the mean value of the elements, denotes the estimated value of the target state at the time instant ; Optimal scout node selection for moving target tracking is achieved by minimizing the trace of the inverse of the Bayesian Fisher information matrix ​ Step 4-2: Definition At that moment, the The selection variables for each reconnaissance node are: , Indicates the first One reconnaissance node was not selected and did not participate in the target trajectory tracking data processing. Indicates the first Several reconnaissance nodes have been selected to participate in the subsequent target trajectory output process; the node selection vector for all receiving stations is defined as follows. ,have: （35） Assume the number of selected nodes at each tracking time is then there is: （36） wherein represents a column vector of ones; Under the node selection scenario, the Bayesian Fisher information matrix is further expressed as a function of the variable and has: （37） Therefore, the cost function about node selection is expressed as: （38） wherein denotes the trace of the matrix in the brackets, denotes optimizing such that the elements in the brackets take on a minimum value.

6. The optimal scout node selection method for moving target direct tracking according to claim 5, wherein, The step 5 is specifically: Boolean constraints involved in the optimization problem are relaxed as continuous variable constraints, have a relaxation of the optimization problem （39） Introducing a symmetric positive definite matrix , satisfying: （40） wherein represents a positive definite sign; Thus, the original problem is equivalent to: ; thus, the original problem is equivalent to: （41） impose By the Schur complement theorem, the linear matrix equation can be written as （42） denotes the identity matrix of dimension ; At this time, the optimization problem is converted to: （43） First, the optimal result output by the solver is arranged in descending order according to size, and the first large value is taken as 1, and the remaining small values are taken as 0, so as to obtain the Boolean quantity of optimal node selection, thereby determining the optimal node distribution of the system at the current time.

7. The optimal scout node selection method for moving target tracking according to claim 6, wherein, The is an additive complex Gaussian white noise with mean 0 and variance ​ 8. An electronic device, comprising: Including: A processor and a memory; The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the method in any one of claims 1 to 7.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the method in any one of claims 1 to 7.

10. A chip, characterized by Including: A processor is used to call and run a computer program from a memory, so that the device installed with the chip executes the method in any one of claims 1 to 7.

Images