Networking radar multi-target tracking resource control method and device

Abstract

The invention discloses a networking radar multi-target tracking resource control method, which comprises the following steps of: constructing a networking radar signal model, and establishing a target motion model and a measurement model; constructing a target tracking strategy, and determining performance representation of target tracking dynamic parameter estimation; establishing a resource control model by taking radiation power, duty ratio, residence time and beam allocation as optimization variables, taking pulse string length intercepted by electronic reconnaissance equipment as a target function and taking target tracking precision as a constraint condition; and solving the resource control model.

Description

Technical Field

[0001] The present invention relates to the field of radar signal processing technology, and in particular to a method and device for controlling networked radar tracking multi-target resources. Background Art

[0002] The role of radar anti-interference is to weaken the impact of electronic interference on radar performance and ensure that the radar can effectively detect, identify and track targets. According to the purpose of use, it can be divided into active anti-interference and passive anti-interference. Passive anti-interference methods such as sidelobe cancellation, sidelobe blanking, subspace projection and blind source separation are all to identify and filter out interference signals after being interfered. Therefore, some useful signals may be lost while filtering out interference. Active anti-interference aims to actively increase the difficulty of interference equipment to release interference. For example, waveform diversity technology can increase the difficulty of interference equipment to track and copy radar signals. Therefore, active anti-interference technology has received widespread attention.

[0003] From the perspective of the radar confrontation process, after intercepting the radar signal, the radiation source is detected through parameter measurement and pulse sorting, and the threat level of the radiation source is determined before the interference decision can be made. Therefore, the use of low probability of intercept (LPI) technology to increase the difficulty of radar signal reconnaissance and realize radar RF stealth is the beginning of active anti-interference. At present, there are many quantitative indicators of radar RF stealth performance, among which the interception probability model based on window function has been widely used. This model regards the interception of radar signals as a problem of coincidence probability of multiple windows in space, time, frequency, and energy domains, that is, it calculates the probability that the radar signal falls into various reconnaissance windows of the interception receiver at the same time and can be detected. Therefore, the interception probability can be reduced by changing the distribution of radar signals in the space, time, and frequency energy domains through radiation control. For example, the probability of detecting radar from the side lobe can be reduced by controlling the distribution of signal energy in space through beamforming technology. Frequency-controlled array radar can control the distribution of signal energy in the distance dimension, which can be used as another implementation method of LPI radar. In addition, controlling the distribution of signal energy in the time domain, such as adjusting the radar revisit interval and dwell time, can also change the radar's LPI performance. However, due to limited detection resources, it is difficult for a single radar to strike a balance between detection performance and RF stealth.

[0004] A radar networking system is a radar system that connects multiple radars distributed in space to perform coordinated detection and information fusion of targets. Compared with a single radar system, a radar networking system has richer detection resources and can be flexibly configured under unified control to maximize the detection efficiency of the system. Based on this feature, many scholars have studied the LPI technology of radar networking systems, so that the radar networking system as a whole has a smaller radiation risk when completing detection tasks.

[0005] Although there are many studies on radar resource allocation strategies under LPI guidance, these studies actually focus on the interception of radar pulses, and rarely involve subsequent pulse sorting and radiation source identification, which is also one of the basic functions of the interception receiver and one of the research objects of LPI technology. As we all know, the basic principle of electronic reconnaissance equipment to intercept radiation sources is that the radar radiation source continuously transmits signals according to a certain rule when working. If this regularity can be analyzed from the (sequentially) intercepted pulse sequence, the existence of the radiation source can be determined. To intercept the pulse of a specified radiation source, in addition to requiring that the pulse falls into the reconnaissance window and can be detected, it must not be lost due to the simultaneous arrival of multiple pulses. Therefore, in order to improve the LPI performance of the radar system, efforts should be made to shorten the length of the pulse string intercepted by the interception receiver, not only to reduce the interception probability of the pulse signal, but also to increase the pulse loss probability, and ultimately achieve the goal of reducing the pulse sorting probability. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a resource control method for networked radar tracking multiple targets, which further reduces the probability of pulse sorting while increasing the probability of pulse loss by optimizing radiation power, duty cycle, dwell time and beam allocation.

[0007] To achieve the above-mentioned purpose, the present invention provides, on the one hand, a resource control method for networked radar tracking multiple targets, comprising the following steps: constructing a networked radar signal model, establishing a target motion model and a measurement model; constructing a target tracking strategy, and determining the performance characterization of target tracking dynamic parameter estimation; using radiation power, duty cycle, dwell time and beam allocation as optimization variables, the length of the pulse string intercepted by the electronic reconnaissance equipment as the objective function, and using target tracking accuracy as a constraint to establish a resource control model; and solving the resource control model.

[0008] On the other hand, the present invention also provides a networked radar tracking multi-target resource control device, including the following modules: a signal model construction module, used to construct a networked radar signal model, establish a target motion model and a measurement model; a resource control model construction module, used to establish a resource control model with radiation power, duty cycle, dwell time and beam allocation as optimization variables, the length of the pulse string intercepted by the electronic reconnaissance equipment as the objective function, and the target tracking accuracy as a constraint condition; a target tracking module, used to construct a target tracking strategy and determine the performance characterization of target tracking dynamic parameter estimation; a solution module, used to solve the resource control model. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A flow chart of a resource control method for tracking multiple targets using a networked radar according to the present invention;

[0010] Figure 2 is a flowchart of step S20;

[0011] Figure 3 A schematic diagram of the scenario shows radar distribution and target tracks;

[0012] Figures 4a to 4c The following are schematic diagrams of the changes in the radiation parameters of the radar node: radiation power, duty cycle, and dwell time;

[0013] Figure 5 This is a schematic diagram of beam allocation;

[0014] Figures 6a to 6d PCRLB under four different resource control methods;

[0015] Figure 7 is the change of RMSE of each target;

[0016] Figures 8a to 8d PRI spectra of four different resource control methods at the 10th frame;

[0017] Figures 9a to 9d The following is a graph showing the sorting probability of each radar node under four different resource control methods;

[0018] Fig.10 are the pulse train lengths under four different resource control methods;

[0019] Fig.11 The changes of the number of pulses at the 2nd, 20th and 40th frames;

[0020] Fig.12 It is a structural schematic diagram of the networked radar tracking multi-target resource control device of the present invention.

[0021] Wherein, the reference numerals are:

[0022] 1-Networking radar tracking multi-target resource control method;

[0023] 2-Networking radar tracking multi-target resource control device;

[0024] 20-Signal model building module;

[0025] 21- target tracking module;

[0026] 22-Resource control model building module;

[0027] 23-Solving module;

[0028] S10~S40, S200, S201-steps. DETAILED DESCRIPTION

[0029] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments to further understand the purpose, solution and effect of the present invention, but it is not intended to limit the scope of protection of the claims attached to the present invention.

[0030] References to "an embodiment", "another embodiment", "this embodiment", etc. in the specification refer to the embodiment being described which may include specific features, structures or characteristics, but not every embodiment must include these specific features, structures or characteristics. In addition, such expressions do not refer to the same embodiment. Furthermore, when describing specific features, structures or characteristics in conjunction with an embodiment, whether or not there is an explicit description, it is indicated that incorporating such features, structures or characteristics into other embodiments is within the knowledge of those skilled in the art.

[0031] Certain words are used in the specification and subsequent claims to refer to specific components or parts. People of ordinary skill in the art should understand that technical users or manufacturers can refer to the same component or part with different nouns or terms. This specification and claims do not use differences in names as a way to distinguish components or parts, but use differences in the functions of components or parts as the criteria for distinction. "Including" and "including" mentioned throughout the specification and claims are open-ended terms and should be interpreted as "including but not limited to". In addition, the word "connected" here includes any direct and indirect electrical connection means. Indirect electrical connection means include connection through other devices.

[0032] The embodiment of the present invention aims at the scenario of multi-radar cooperative tracking of multiple targets, and proposes a resource control solution for countermeasure signal sorting from the perspective of active anti-interference. The details are as follows:

[0033] 1) A signal model for MIMO radar network tracking multiple targets was established. Each radar transmits a wide beam to cover the entire surveillance area and receives signals from different targets through multiple narrow beams. Subsequently, each radar estimates the angle and Doppler frequency from the observed data using the maximum likelihood estimation method, and then performs data fusion at the information fusion center and uses an extended Kalman filter.

[35] A global estimation is performed to minimize information loss. In order to qualitatively describe the relationship between target tracking accuracy and radar parameters, the mathematical expression of the Posterior Cramér-Rao Lower Bound (PCRLB) is derived, and the radar radiation parameters that jointly affect tracking accuracy and stealth performance are determined.

[0034] 2) A resource control scheme combining single pulse anti-interception and pulse train anti-sorting is proposed for radar networking systems. Existing RF stealth solutions usually only focus on the low pulse interception probability. On this basis, the embodiment of the present invention takes into account the subsequent pulse sorting process. The technical solution includes: shortening the dwell time to reduce the pulse train length, optimizing the radiation power and duty cycle to increase the pulse loss, thereby further reducing the pulse sorting probability. A resource control model is established with radiation power, duty cycle, dwell time and beam allocation as optimization variables, the pulse train length intercepted by the electronic reconnaissance equipment as the objective function, and the target tracking accuracy as the constraint condition, and the alternating minimization method is used to solve the model.

[0035] 3) Pulse sorting probability is used to characterize the LPI performance of the networked radar. The existing technology focuses on resource optimization for LPI, and there is little intuitive display of LPI performance improvement. In the simulation, the embodiment use The stealth performance of the networked radar is described by using multiple radar sorting probabilities. The results of the embodiment intuitively show that the resource optimization strategy proposed by the present invention can achieve a balance between radar detection performance and radio frequency stealth capability.

[0036] In the embodiments of the present invention, vectors are represented by lowercase bold fonts, and matrices are represented by uppercase bold fonts. (·) T and(·) -1 denote transpose and inverse respectively. represents the Kronecker product. It means that x follows a Gaussian distribution with mean m and covariance ∑. represents mathematical expectation. Represents a collection of N×M dimensional matrices. Indicates the extraction of the real part. [·] i，j represents the element in the i-th row and j-th column of the matrix, [·] i Represents the i-th element of a vector.

[0037] Figure 1 A flowchart of a networked radar tracking multi-target resource control method 1 provided in an embodiment of the present invention includes:

[0038] S10: Build a network radar signal model, and establish a target motion model and a measurement model;

[0039] S20: construct a target tracking strategy and determine the performance characterization of target tracking dynamic parameter estimation;

[0040] S20: A resource control model is established with radiation power, duty cycle, dwell time, and beam allocation as optimization variables, the length of the pulse train intercepted by the electronic reconnaissance equipment as the objective function, and target tracking accuracy as the constraint condition;

[0041] S30: Solve the resource control model.

[0042] In step S10,

[0043] Assume that a radar networking system contains N co-located MIMO radar nodes, monitoring Q targets. Each co-located MIMO radar node works in the "wide transmit and narrow receive" mode, that is, each array element at the transmitting end transmits orthogonal signals to form a wide beam covering the entire monitoring area, and the receiving end uses digital beamforming technology to form multiple narrow pulses to track different targets respectively.

[0044] The specific steps to establish the target motion model are as follows:

[0045] Assume that Q targets are moving in a uniform linear motion in the XOY plane, and the motion state of the qth target at time k is Then the motion state equation of the qth target can be expressed as:

[0046]

[0047] Among them, F is the target state transfer matrix, which is expressed as:

[0048]

[0049] ΔT is the sampling interval, is zero-mean white Gaussian process noise, and the covariance matrix is:

[0050]

[0051] k q is the noise intensity during the target q-state transition process.

[0052] The specific steps to establish the measurement model are as follows:

[0053] Take the nth radar node as an example, assuming it is located at (x n ,y n ), including M tn The transmitting array element M rn receiving array elements, transmitting at the kth tracking time A pulse signal, and residence time The relationship is is the pulse interval of radar node n at time k. Definition It represents the (measurement) pairing relationship of radar n to Q targets at time k, It means that radar node n measures target q at time k, Indicates that radar node n does not measure target q at time k. The first signal received by radar node n at time k about target q The pulse signal can be expressed as:

[0054]

[0055] in, represents the propagation loss and the distance of target q from radar node n at time k Related, satisfying is the transmit power of radar node n at time k, is the radar cross section (RCS) of target q relative to radar node n at time n, is the azimuth of target q relative to radar node n at time k, is the propagation delay of target q relative to radar node n at time k, is the Doppler frequency of target q relative to radar node n at time k, is the transmitted signal vector of radar node n at time k, which are orthogonal to each other and satisfy:

[0056]

[0057] in, is the signal pulse width of radar node n at time k, α rn and α tn are the receiving vector and transmitting vector of radar node n, respectively, satisfying:

[0058]

[0059] and

[0060]

[0061] d rn and d tn are the receiving array element spacing and transmitting array element spacing of radar node n, respectively, n is the emission wavelength of radar node n. Note that equation (4) ignores the intra-pulse Doppler. After receiving the signal, radar node n uses Perform matched filtering on the received signal of each array element to obtain the M signal of target q observed by radar node n at time k. tn M rn ×1 observation vector:

[0062]

[0063] in right The pulse signal is processed in the same way, and then Stacked into one The observation vector of dimension is:

[0064]

[0065] Here, is the Doppler frequency vector of target q relative to radar node n at time k, defined as:

[0066]

[0067] If the existence of additive Gaussian noise is considered, its covariance matrix is make but:

[0068]

[0069] From formula (9), we can know that from the observation vector Directly estimate and definition but and The maximum likelihood estimate of and It can be obtained from the following formula:

[0070]

[0071] If the observation error is considered, assuming it is zero-mean Gaussian noise, the covariance is The observation model (i.e., measurement model) can be expressed as:

[0072]

[0073] Among them, about Nonlinear function Defined as:

[0074]

[0075] Under high signal-to-noise ratio conditions, it can be replaced by the CRLB matrix.

[0076] like Figure 2 In step S20, constructing a target tracking strategy further includes:

[0077] S200: forming a global measurement model by splicing measurement models of multiple radar nodes;

[0078] S201: Use a nonlinear filter to estimate the target state.

[0079] Specifically, for the target q, define is the set of radars tracking target q at time k. When multiple radars simultaneously track target q, that is, In order to improve the tracking accuracy and reduce information loss, each radar transmits the observation data directly to the information fusion center for global fusion. The global measurement model of multiple radar nodes on the target q at time k is It is composed of the local measurement models of multiple radar nodes participating in tracking the target at that moment, as shown in the following formula (15):

[0080]

[0081] in, Representing a collection The nth element of Representing a collection Correspondingly, the global measurement model can be expressed as formula (16):

[0082]

[0083] in,

[0084]

[0085] For the nonlinear measurement models shown in equations (16) and (17), a nonlinear filter can be used to estimate the target state. EKF is an extension of the Kalman filter, which handles nonlinear systems by using a Taylor series expansion in the state transfer and measurement model. Compared with the particle filter, the EKF based on the analytical solution usually has a lower computational complexity. From the perspective of saving computing resources, this embodiment selects EKF to estimate the target state, and the target state estimate value at each moment is and its covariance matrix It can be calculated according to the following formula (19):

[0086]

[0087] in, is the Kalman gain, calculated according to the following formula (20):

[0088]

[0089] and is a function About vector variables The Jacobian matrix of is defined as formula (21):

[0090]

[0091] in, Function again About vector variables The Jacobian matrix of is defined as formula (22):

[0092]

[0093] The elements of the first row are:

[0094]

[0095] The elements of the second row are:

[0096]

[0097] In step S20, PCRLB is used as the performance characterization of target tracking dynamic parameter estimation, specifically:

[0098] Parameters such as radar power, duty cycle and number of pulse trains affect the detection and sorting of radar by reconnaissance equipment. In order to ensure the effectiveness of radar detection, it is also necessary to analyze the relationship between these parameters and radar effectiveness (target tracking accuracy in tracking scenarios). Therefore, it is necessary to first establish a mathematical model to quantitatively characterize the target tracking accuracy. In the parameter estimation problem, CRLB provides the minimum variance bound for parameter estimation under a given data model and estimation method. The posterior CRLB describes the minimum variance bound for parameter estimation given prior information and observed data. This embodiment uses the posterior Cramer-Rao lower bound PCRLB (Posterior Cramer-Rao Lower Bound) as a performance characterization in the estimation problem of dynamic parameters such as target tracking (parameters such as radar power, duty cycle and number of pulse trains), because PCRLB not only considers the information of the observed data, but also considers the influence of prior knowledge on parameter estimation. PCRLB is defined as follows:

[0099]

[0100] in represents all (current) measurements of the target q at time k, is the Bayesian information matrix about the target state, calculated as follows:

[0101]

[0102] in It means to find the first-order partial derivative of vector θ. Because:

[0103]

[0104] so Can be further written

[0105]

[0106] in

[0107]

[0108] and

[0109]

[0110] represent the prior Fisher information matrix and the Fisher information matrix from the observed data, respectively. can be calculated recursively as:

[0111]

[0112] in

[0113]

[0114] Considering the linear target motion model shown in equation (1), equation (32) can be further written as:

[0115]

[0116] so,

[0117]

[0118] for When the observations (measurements) of target q by each radar node are independent of each other,

[0119]

[0120] because Directly with Relationship rather than So applying the chain rule will Rewritten as:

[0121]

[0122] in,

[0123]

[0124] Its elements can be calculated as follows

[0125]

[0126] in,

[0127]

[0128] By calculation, we can get:

[0129]

[0130]

[0131] The above approximation uses and at this time, It can be simplified as:

[0132]

[0133] Defined as:

[0134]

[0135] From formula (35), we know that The mathematical expectation needs to be obtained, which can be achieved using the Monte Carlo method. However, in order to meet the real-time requirements of target tracking and reflect the pre-emptive nature of resource allocation, the zero process noise prediction value can be used. replace In summary, It can be expressed as:

[0136]

[0137] In step S30, first, two assumptions are made about the radar signal intercepted by the electronic reconnaissance equipment. The first assumption is that the radar pulse signal must fall into the reconnaissance window of the reconnaissance equipment; the second assumption is that when a pulse enters the interception receiver while the previous pulse is still being processed (the processing time is equal to the pulse width), the pulse is declared lost. Specifically:

[0138] Electronic reconnaissance equipment passively receives non-cooperative radar signals. Due to the limitations of cost-effectiveness, the size, weight and power consumption of the loading platform, as well as the need to deal with unknown threats or detect unknown radar signals, it is impossible to use multiple dedicated receivers that receive specific radar signals to jointly cover the required signal types, frequency domains and airspace, etc.; due to the response time requirements, it is also impossible to use a search receiver to receive different airspace, frequency domain, time domain and different types of radar signals in time sharing. Therefore, most radar reconnaissance equipment has to use a single-channel universal receiver to receive various types of radar signals in the frequency domain, air domain and time domain. Therefore, the first assumption is that the radar pulse signal must fall into the reconnaissance window of the reconnaissance equipment. At this time, only the detection probability of a single pulse needs to be considered:

[0139]

[0140] in, SNR I It is the signal-to-noise ratio of a single pulse at the output of the intercept receiver.r To intercept the radar radiation power received by the receiver, then:

[0141]

[0142] Among them, P t is the radiated power of the radar node, G t , G r and G P are the transmit gain of the radar node in the direction of the intercept receiver, the receive gain of the intercept receiver in the direction of the radar node, the processing gain of the intercept receiver, and t is the wavelength, F p is the propagation factor, R is the distance from the radar node to the intercept receiver, and L p is the polarization loss. SNR I It can be calculated as:

[0143]

[0144] Among them, k B is the Boltzmann constant, which is 1.38×10 -23 J / K, T0 is the noise temperature, B is the intercept receiver bandwidth, N F is the intercept receiver noise factor. It can be seen that reducing the radar radiated power can reduce the probability of the radar pulse being detected by the intercept receiver.

[0145] When there are multiple radiation sources, assuming there are N, and they are turned on and working at the same time, the wide-open reconnaissance equipment receives multiple pulse trains, which are intertwined and easily cause pulse loss. For example, if the leading edge arrival time of two pulses is less than the width of the first pulse, the two pulses will overlap, and the pulse overlap causes the parameter measurement to deviate, resulting in pulse loss (which pulse is lost or both pulses are lost is related to the degree of aliasing and amplitude difference of the two pulses). For another example, for the instantaneous frequency measurement receiver IFM, there is a certain recovery period after completing the parameter measurement of a pulse. The new pulse arriving during this recovery period will not be processed, which is equivalent to being discarded. In order to quantitatively describe the probability of pulse loss, the present invention simplifies the model and makes a second assumption: when a pulse enters the interception receiver, the previous pulse is still being processed (the processing time is equal to the pulse width), then the pulse is declared lost. Therefore, the disappearing single-channel random service model in queuing theory can be used to describe the interception process of the reconnaissance equipment on the pulse. According to statistical knowledge, the pulse stream formed by the pulse signals emitted simultaneously (asynchronously) by multiple radars obeys the Poisson distribution, and the pulse width obeys the negative exponential distribution. Therefore, the M|M|1 queueing model can be used to calculate the pulse loss probability of the wide-open receiver:

[0146]

[0147] in, is the service intensity of the queuing system, λ is the pulse flow density, is the average processing time of a single pulse. In the case of simultaneous radiation from multiple radars, It can also be regarded as the total duty cycle D of the pulse flow c :

[0148]

[0149] Among them, D cn is the duty cycle of the nth radar, T dn and T rn are the pulse width and pulse repetition interval of the nth radar respectively. It can be seen that the pulse loss probability when the wide-open receiver receives multi-radiator signals is related to the total duty cycle of the pulse stream. By adjusting the duty cycle of each radar, the pulse loss probability can be increased and the probability of radar reconnaissance equipment intercepting radar signals can be reduced.

[0150] In step S30, a resource control model is established.

[0151] Specifically, define They represent the radiated power, duty cycle and dwell time control vector of the radar network (radar node) at time k, The matrix represents the radar node beam allocation, and the following formula (55) is used as the measurement index of target tracking accuracy, namely the resource control model:

[0152]

[0153] From formula (48), we know that and and U k From equations (49), (51) and (52), we know that the probability of the radar reconnaissance equipment detecting each radar node at time k is The radiation power of each radar node From formula (53), we know that the probability of the radar reconnaissance equipment losing a pulse at time k is Total duty cycle of pulse flow It is also known that the length of the pulse sequence intercepted by the radar detection equipment is The dwell time of each radar node Duty cycle of each radar node and pulse width T dn related

[0154] Therefore, in order to achieve active anti-interference, the radar networking system should reasonably adjust the radiation parameters of each radar node at each tracking moment ( and ) and the pairing relationship with the target, and minimize the interception probability of the radar radiation source by the reconnaissance equipment under the premise of ensuring that the target tracking accuracy meets the mission requirements. Therefore, the following optimization problem can be established:

[0155]

[0156] Among them, α1 and α2 are weight coefficients, Constrain the tracking error of each target not to exceed the upper limit and Constrain the radiation power, duty cycle and dwell time of each radar node to be within a certain range; It means that each radar node can form at most μ n receive beams; Each target is required to be tracked by at least one radar node. and They are and A monotonically increasing function of Therefore, the resource control model (55) can be rewritten as:

[0157]

[0158] In step S40, the resource control model equation (57) is solved by using an alternating minimization method.

[0159] Specifically, because the 0-1 variable The existence of , the above optimization problem is a non-convex nonlinear mixed integer programming problem. One exhaustive search method is based on and These two constraints enumerate all possible pairings between radar beams and targets, and select the one that minimizes the objective function value. and However, this exhaustive search method takes too long and is not conducive to real-time processing. Relax to a continuous value on the interval [0, 1], and then alternately solve U k and other variables.

[0160] First initialize and Because the objective function In fact, with U k It is irrelevant, so solve U k The following models can be used:

[0161]

[0162] That is, in a given and In the case of k Minimize the tracking error of each target. The optimization model can be solved using the interior point method. Assume that the relaxation solution is It is also necessary to construct a feasible solution U that meets the original problem k The specific method is: let U k is equal to an all-zero matrix. First, transform the matrix Arrange the elements of each column in ascending order to find the position of the maximum value. k The value of 1 is assigned to the location of this target, which ensures that each target is tracked by radar; secondly, the matrix Sort each row of in ascending order and select the largest μ n elements (if an element has been selected in this row, the number of corresponding elements is subtracted), in U k Assign 1 to the corresponding position of , which can ensure that the receiving beam of each radar is tracking the target. Obtain a feasible solution U k After that, the following optimization problem can be solved by the interior point method to obtain and U k A feasible solution for :

[0163]

[0164] This cycle repeats alternately. When the objective function value is When the difference is less than the preset value, the final and U k .

[0165] The following simulation experiment will verify the effectiveness of a networked radar tracking multi-target resource control method provided by an embodiment of the present invention in reducing the radar sorting probability.

[0166] Assume that there are 6 co-located MIMO radar nodes monitoring 4 moving targets on a one-dimensional plane. The simulation parameters of each radar node are shown in Table 1, where the radiation power P t , duty cycle D c and residence time T dw It is a controllable variable, its value is distributed within a certain range, and the pulse width T d and the number of simultaneous beams μ are invariant parameters. The simulation parameters of each target are shown in Table 2. The sampling interval of the radar node ΔT = 1s, and the tracking process lasts for 50 frames. The wavelength λ of the radar node n =0.3m, the noise intensity k during the state transition q =1, observation noise intensity σ n=1. There are two points to be explained. First, the preset upper limit of the tracking error is not constant, but decreases linearly with the tracking process to achieve a gradual improvement in tracking accuracy. The initial values ​​are 65m, 175m, 105m and 85m respectively; second, the initial Fisher Information Matrix (FIM) of all targets is calculated by Q q Instead, that is: for all targets,

[0167] Table 1: Simulation parameters of radar node

[0168] radar Location <![CDATA[P t (in)]]> <![CDATA[D c (%)]]> <![CDATA[T dw (ms)]]> <![CDATA[T d (μs)]]> μ 1 [0,15] [1477,7341] [1.7,9] [48,260] 19 3 2 [0,30] [1641,7267] [0.8,14] [35,390] 20 3 3 [15,30] [1056,7934] [1.2,13] [20,290] 16 3 4 [30,15] [1126,4754] [1.7,5] [24,130] 11 3 5 [30,0] [1151,7007] [0.6,6] [36,440] 12 3 6 [15,0] [1348,4059] [0.6,18] [39,390] 20 2

[0169] Table 2: Target simulation parameters

[0170] Target Location (km) Speed ​​(m / s) <![CDATA[RCS(m 2 )]]> 1 [11,17] [-78,-31] 4 2 [20,24] [10,38] 4 3 [7,12] [-57,-87] 2 4 [19,10] [-70,8] 4

[0171] Figure 3 The radar distribution and target trajectory are shown. The dotted line represents the target trajectory estimated under a Monte Carlo test, and the solid line represents the actual trajectory. It can be seen that the method of the present invention can ensure that the networked radar tracks the target in real time and accurately.

[0172] Figures 4a to 4c The optimized radiated power, duty cycle, and dwell time during target tracking are shown in turn. Note that these values ​​have been normalized. Taking radiated power as an example, assuming that the optimized power is After normalization, it becomes

[0173]

[0174] Therefore, when When the value of is 1, it indicates the maximum radiation power, while the value of 0 indicates the minimum radiation power. It can be seen that when the targets move, their distance from each radar will change, so the radar radiation parameters need to be adjusted dynamically to maintain tracking accuracy. Especially in the early stage of tracking (the first five frames), the values ​​of all radiation parameters are large, and they become very small from the sixth frame, and the change trends of the two parameters, radiation power and dwell time, have more overlap. This is because in the early stage of tracking, due to the large tracking error, the system needs to increase the radiation power and the number of pulses to improve the tracking accuracy, and the LPI requirement increases the duty cycle of each radar to increase the pulse loss and shorten the length of the pulse string intercepted by the interception receiver. When the system enters the stable tracking stage, the LPI requirement will greatly reduce the radiation power and the number of pulses, so there is no need to modulate the duty cycle to increase the pulse loss, so the duty cycle value becomes very small after the sixth frame.

[0175] Figure 5The beam allocation during the whole process is shown. It is obvious that each radar dynamically adjusts the tracking target at each moment, rather than being fixed to a specific target. In addition, in most cases, each radar allocates all available beams to the target, and there are only a few cases where the target is tracked with fewer beams than available beams.

[0176] In order to quantitatively describe the target tracking accuracy under the proposed resource control strategy, Figure 6 depicts the Curve. Note that for comparison, Figure 6 also plots the effects of the other three resource control methods on target tracking accuracy. The first control method is the minimum tracking accuracy mode, that is, the stealth performance of the radar networking system is not considered, and the resource allocation is optimized only to reduce the tracking error at each moment. The specific solution process is similar to the cyclic minimum method used in this paper, except that the objective function is The second comparison method is the minimum interception probability mode, which does not consider the tracking error at each moment, and optimizes resource allocation only to reduce the interception probability of the radar networking system. and The solution process is the same as equation (59), the beam allocation matrix U k The third comparison method is random power + random duty cycle + random dwell time, and the beam allocation matrix U k The above three comparison methods all assume that the radar tracks the target randomly according to the beam upper limit. It can be seen that in the method of this paper and the minimum tracking error mode, the tracking error is always lower than the preset threshold, which meets the requirements of the radar network system for multi-target tracking accuracy. In contrast, in the minimum intercept probability mode and the random mode, except for the second target, the tracking accuracy of the remaining targets did not meet the task requirements.

[0177] The one used in Figure 6 represents the calculated or predicted tracking accuracy. Therefore, the root mean square error (RMSE) of each target can be used, that is, To express the actual tracking accuracy, it is defined as

[0178]

[0179] in, represents the target state estimate obtained in the nth Monte Carlo experiment, and Indicates the actual target state. Figure 7 It can be seen that as time goes by, It will gradually converge to PCRLB. Note that N MC =1000.

[0180] The low interception of the networked radar in the collaborative working mode proposed in the present invention is verified through simulation experiments. At each tracking moment, the reconnaissance device receives a pulse stream signal, which is formed by the confluence of pulses emitted by multiple radars. The present invention measures the anti-interception performance of the radar network by the difficulty of sorting out each radar pulse sequence from the pulse stream. Using pulse repetition interval (PRI) information to sort pulse sequences is a widely used technology, and this section adopts the PRI transformation method. For the convenience of quantitative comparison, it is assumed that there is a reconnaissance device in the center of the scene, that is, [15, 15] km. Under the resource allocation method proposed in the present invention, the low interception of the networked radar will be reflected in three aspects: first, by controlling the dwell time of each radar to emit a short pulse train; second, the reduction of radiation power reduces the probability of a single pulse being intercepted. Assume that at time k, the reconnaissance device can transmit P fa =10 -4 , P d =0.999 intercepts the pulse with the maximum power. Note that the maximum power here refers to Maximum, R In is the distance between the reconnaissance equipment and the radar n, and the transmission gain of each radar in the direction of the interception receiver and the receiving gain and processing gain of the interception receiver in the direction of each radar are the same. According to formulas (49, 50, 51, 52), the detection probability of each radar pulse by the interception receiver can be calculated, and then each radar pulse is selectively intercepted based on this probability. The interception here refers to detection. When the radiation power of a certain radar is small, its detection probability is relatively small, resulting in an increase in lost pulses; third, pulse overlap leads to an increase in pulse loss. According to the previous analysis, pulse loss occurs when the arrival time of the two pulse fronts is less than the width of the first pulse. In this simulation, it is assumed that the second pulse is lost. When the total duty cycle of the pulse stream increases, the pulses lost due to overlap increase accordingly.

[0181] Figure 8 describes the PRI spectra of the four resource allocation methods at the 10th frame. It can be seen that due to the large number of pulse losses (not detected or lost due to overlap), the PRI spectra of the method of the present invention and the minimum intercept probability mode show a smaller number of pulses compared with the other two methods, and can only detect one PRI. In other words, at most only one radar can be detected, while the minimum tracking error mode can detect all 6 simulated radars, and the random mode can detect 3 or 4 radars.

[0182] In order to quantitatively compare the anti-sorting performance of these resource control methods, this embodiment carried out 1000 Monte Carlo tests. The sorting probability of each radar is plotted in Figure 9. It can be seen that in the minimum tracking error mode, each radar has a very high sorting probability throughout the whole process. In contrast, under the method of the present invention, radar nodes 4 and 5 achieve stealth after the 12th frame, and in the minimum intercept probability mode, radar nodes 3, 4 and 5 can almost achieve stealth throughout the whole process. In the random mode, the sorting probability of each radar is high and low, and there is no regularity to follow, but it is still higher than the method of the present invention overall.

[0183] The LPI performance of the four methods can be measured by the number of pulses intercepted by the interception receiver. When the number of radiation sources is determined, the longer the intercepted pulse, the greater the possibility of finding the radar radiation pattern. Fig.10 It can be seen that the minimum tracking error mode radiates a large number of pulses during the entire tracking process. In contrast, the method of the present invention and the minimum intercept probability mode only emit a large number of pulses in the early stage of tracking, and significantly reduce the number of pulses in the later stage. Fig.11 It shows how the method of the present invention controls the length of the pulse train. In the early stage of tracking, the number of pulses transmitted and the radiation power are both large, so the number of pulses lost due to non-detection is relatively small, and the number of pulses lost due to overlap is relatively large; in the later stage, the number of transmitted pulses is small, and there is no need to further increase the pulse loss by controlling the radiation power and duty cycle.

[0184] In summary, the method of the present invention enables the radar network system to have lower interception performance, and can achieve a balance between the stealth performance and the detection performance of the radar networking system.

[0185] The networked radar tracking multi-target resource control method provided by the present invention designs a resource control scheme in a multi-radar collaborative tracking multi-target tracking scenario from the perspective of active anti-interference, so as to reduce the probability of the radar system being intercepted. Compared with the existing RF stealth method that only considers the single pulse detection problem, the present invention considers the pulse sorting problem that is more closely related to the radiation source detection. By establishing and analyzing the signal model of multi-radar collaborative tracking of multiple targets, it is found that radar controllable variables such as radiation power, dwell time, duty cycle and radar beam allocation scheme will simultaneously affect the radar's tracking accuracy of the target and the interception probability of the reconnaissance equipment to the radar. By optimizing these variables, a balance between detection performance and stealth performance can be achieved.

[0186] The following is a device embodiment corresponding to the above method embodiment. This embodiment can be implemented in conjunction with the above embodiment. The relevant technical details mentioned in the above embodiment are still valid in this embodiment. In order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the above embodiment.

[0187] like Fig.12 As shown, another embodiment of the present invention provides a networked radar tracking multi-target resource control device 2, including the following modules: a signal model construction module 20, a resource control model construction module 22, a target tracking module 21 and a solution module 23, wherein the signal model construction module 20 is used to construct a networked radar signal model, establish a target motion model and a measurement model; the target tracking module 21 is used to construct a target tracking method, and determine the performance characterization of target tracking dynamic parameter estimation; the resource control model construction module 22 is used to establish a resource control model with radiation power, duty cycle, dwell time and beam allocation as optimization variables, with the pulse string length intercepted by the electronic reconnaissance equipment as the objective function, and with the target tracking accuracy as a constraint condition; the solution module 23 is used to solve the resource control model.

[0188] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the module described can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0189] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which belong to the protection scope of the present invention.

Claims

1. A network radar tracking multi-target resource control method, characterized in that: The following steps are involved: Build a networked radar signal model, and establish target motion model and measurement model; Construct a target tracking strategy and determine the performance characteristics of target tracking dynamic parameter estimation; A resource control model is established with radiation power, duty cycle, dwell time and beam allocation as optimization variables, the length of the pulse train intercepted by the electronic reconnaissance equipment as the objective function, and the target tracking accuracy as the constraint condition. The resource control model is solved.

2. The method according to claim 1, characterized in that: The networked radar signal model includes N co-located MIMO radar nodes, monitoring Q targets, and each of the co-located MIMO radar nodes operates in a wide-transmit and narrow-receive mode, where: Each of the co-located MIMO radar nodes transmits a wide beam to cover the entire surveillance area, and receives signals from different targets through multiple narrow beams.

3. The method according to claim 2, characterized in that The constructing of the networked radar signal model further includes: estimating the angle and Doppler frequency of the target by using the maximum likelihood estimation method.

4. The method according to claim 2, characterized in that: The motion model, i.e., the motion state equation, is: in, is the state vector of the qth target among the Q targets at the kth fusion sampling interval, is the state vector of the qth target at the k-1th fusion sampling interval, and F is the moving target state transfer matrix, expressed as: ΔT is the sampling interval, is zero-mean white Gaussian process noise, and the covariance matrix is: κ q is the noise intensity during the target q-state transition process.

5. The method according to claim 4, characterized in that Assume that the observation error is zero-mean Gaussian noise with covariance Then the measurement model of the nth radar node is: Where n represents one of the N radar nodes. Nonlinear function Defined as 6. The method according to claim 6, characterized in that: The construction target tracking strategy includes: The measurement models of multiple radar nodes are spliced ​​to form a global measurement model, and the global measurement model is: in Track the target at k time q A collection of radar nodes, represents the radar node at time k n To target q Take measurements, Representing a collection No. n elements; A nonlinear filter is used to estimate the target state of the global observation model.

7. The method according to claim 6, characterized in that The performance characterization of determining target tracking dynamic parameter estimation includes: PCRLB is used as the performance characterization of target tracking dynamic parameter estimation, and the PCRLB is defined as: in represents the target at time k q All measured values ​​of is a Bayesian information matrix about the target state, and the dynamic parameters include the radiation power, the duty cycle, the dwell time and the beam allocation.

8. The method according to claim 7, characterized in that The resource control model is as follows: in, Respectively k a control vector of the radiation power, the duty cycle and the dwell time of the radar network at a given moment, represents the matrix of the beam allocation and satisfies the following optimization problem: Among them, α1 and α2 are weight coefficients, Constrain the tracking error of each target not to exceed an upper limit and Respectively constraining the radiation power, the duty cycle, and the dwell time of each radar node to be within a certain range; It means that each radar node can form at most μ n receive beams; It is required that each of the targets be tracked by at least one of the radar nodes; Represents the number of pulse signals emitted by radar node n at time k.

9. The method according to claim 1, characterized in that: The resource control model is solved by using an alternating minimization method.

10. A networked radar tracking multi-target resource control device, characterized in that: A method for controlling multi-target resource tracking by a networked radar according to any one of claims 1 to 9, comprising: Signal model building module, used to build network radar signal model, target motion model and measurement model; The target tracking module is used to build the target tracking strategy and determine the performance characterization of the target tracking dynamic parameter estimation; A resource control model building module is used to establish a resource control model with radiation power, duty cycle, dwell time and beam allocation as optimization variables, the length of the pulse string intercepted by the electronic reconnaissance equipment as the objective function, and the target tracking accuracy as the constraint condition; A solution model is used to solve the resource control model.

Images