CN-121541155-B - Clock synchronization and spatial registration integrated method for distributed networking radar

CN121541155BCN 121541155 BCN121541155 BCN 121541155BCN-121541155-B

Abstract

The invention provides a clock synchronization and spatial registration integrated method for a distributed networking radar, and relates to the technical field of radar signal processing. According to the method, the sub radar echo and pose data are acquired, the co-vision event is judged, time difference, frequency difference and angle difference measurement is formed, cross-node constraint is established, round trip time measurement, continuity and smooth prior are combined, and robust loss suppression anomaly measurement is adopted. And performing sparse nonlinear least square solution in the sliding window to obtain a unified time scale and a unified space coordinate system, correcting the transmitting time sequence, the sampling clock and the beam pointing in real time, and taking the result as the prior of the next window. When the residual exceeds the threshold, the observability is enhanced, the geometric condition is improved by adding the bidirectional time measurement and adjustment parameters, the high-precision synchronization and registration without external teaching are realized, and the consistency and the cooperative performance of the networking radar are ensured.

Inventors

Ran Jinhe
ZHANG HAIYAN
LIN ZHONGWEI
SHEN YANG
ZHANG YI
Qi Xunshuo

Assignees

中国人民解放军国防科技大学

Dates

Publication Date: 20260508
Application Date: 20260119

Claims (7)

1. The integrated method for clock synchronization and spatial registration for the distributed networking radar is characterized by comprising the following steps of: Acquiring local time scale echo and pose data of each sub-radar, performing cross-node correlation in a sliding window according to distance, angle and Doppler consistency, judging a common-view event, and calculating common-view measurement, wherein the common-view measurement comprises an arrival time difference, an arrival frequency difference and a sight angle difference respectively; Setting clock deviation, frequency drift, an extrinsic rotation matrix and extrinsic translation for each node, and fixing extrinsic parameters of a reference node to eliminate global uncertainty; forming a constraint set of a cross node by using the common view measurement, superposing constraint during round trip time, drift continuity priori and external parameter smooth priori, and inhibiting outlier measurement by using robust loss; performing sparse nonlinear least square solution on the constraint set in a sliding window to obtain clock and external parameter of each node, thereby forming a unified time scale and a unified space coordinate system; Correcting the transmitting time sequence, the sampling clock and the beam direction on line based on the clock and the external parameter, and taking the clock and the external parameter as the prior of the next window; Triggering observability enhancement when any one of the following conditions is met, wherein the root mean square of an arrival time difference residual error exceeds a preset first threshold value, or the root mean square of a sight angle difference residual error exceeds a preset second threshold value, or the chi-square statistics of a normalized residual error exceeds a preset third threshold value; Performing sparse nonlinear least square solution on the constraint set in a sliding window to obtain clock and external parameter of each node, thereby forming a unified time scale and a unified space coordinate system, comprising: parameter set to be estimated obtained by the sliding window As an initial value, linearizing all residual errors in the constraint set at the initial value to obtain a jacobian matrix And residual vector Wherein, the method comprises the steps of, As a coefficient of the frequency drift, In order for the clock to be biased, For the rotation matrix of the extrinsic parameters, Is an extrinsic translation vector; Construction of normal equations Wherein 、 Sparse decomposition is carried out according to the elimination sequence of node blocks, and the elimination time parameter is prioritized External parameters of the space for eliminating primordial energy again To reduce the filling and preserve key coupling items, find the increment Wherein, the method comprises the steps of, Is a matrix of coefficients of a normal equation, Is the right end item; Updating the extrinsic rotation matrix using exponential mapping to add up Judging convergence according to a preset increment norm threshold and a weighted residual error reduction threshold, and recalculating when the convergence is not achieved Continuing iteration, and updating the formula of the external parameter rotation matrix by adopting index mapping as follows: and normalize quaternions or hold Orthogonality, wherein the degree of orthogonality between the two adjacent light sources, In order to rotate the increment vector, Is a plum group Is an exponential mapping of (2); When the window slides, the marginalization is carried out on the parameters and measurement to be moved out of the window, and a priori factor is formed by using ShuerBu or equivalent information reduction and is incorporated into a constraint set of a new window; After convergence, to map Defining a unified time scale to Defining an extrinsic set from a local coordinate system of each node to a reference node coordinate system as a unified space coordinate system Providing for subsequent coherent processing and multi-base fusion, wherein, As a consequence of the local time-stamp, For a uniform time scale, The rotation matrix and the translation set are the extrinsic parameters of the nodes.
2. The integrated clock synchronization and spatial registration method for a distributed networking radar according to claim 1, wherein the co-view event is an associable observation of at least two nodes on the same scatterer in the same window.
3. The integrated clock synchronization and spatial registration method for distributed networking radars according to claim 1, wherein obtaining local time scale echo and pose data of each sub-radar, performing cross-node correlation according to distance, angle and doppler consistency in a sliding window, judging a co-view event, and calculating co-view measurement, namely an arrival time difference, an arrival frequency difference and a line-of-sight angle difference, respectively, comprises: Setting the length and the step length of a sliding window, respectively marking the sliding window as preset values, extracting target detection items comprising a local time mark, an inclined distance, a sight angle and Doppler frequency shift and corresponding pose data from nodes of each sub-radar, and forming a detection set in the window; performing propagation and motion compensation on each detection item based on the pose data, namely calculating propagation time and compensating arrival time according to the slope distance and the propagation speed of electromagnetic waves, compensating Doppler frequency shift according to the radial speed of a platform, and performing coordinate transformation on a line angle according to a reference node coordinate system; setting thresholds for distance, angle, doppler and arrival time respectively, and recording the thresholds as a distance threshold, an angle threshold, a Doppler threshold and a time threshold, wherein the thresholds are determined by combining a preset constant and a last window estimation result; recording candidate association pairs according to the conditions that the distance difference does not exceed a distance threshold, the angle difference does not exceed an angle threshold, the Doppler difference does not exceed a Doppler threshold and the arrival time difference does not exceed a time threshold between the detection of the nodes of different sub-radars, and constructing an undirected graph by taking the detection as a vertex and the candidate association pairs as edges; Extracting connected subgraphs from the undirected graph, selecting a combination of at most one detection component of each sub-radar node by minimizing a combined residual error formed by an inclined distance, a converted line of sight angle, compensated Doppler and a compensated arrival time in each connected subgraph, and judging as one common-view event when the number of the sub-radar nodes contained in the combination is not less than two; and designating a reference node in each common view event, and respectively calculating the time difference of arrival, the frequency difference of arrival and the line-of-sight angle difference by any participating node relative to the reference node.
4. The integrated method of clock synchronization and spatial registration for a distributed networking radar of claim 1, wherein setting clock bias, frequency drift, extrinsic rotation matrix and extrinsic translation for each node, and fixing extrinsic parameters of a reference node to eliminate global uncertainty, comprises: Setting an extrinsic rotation matrix of the reference node as an identity matrix, and setting an extrinsic translation vector as a zero vector to eliminate global uncertainty under a unified space coordinate system, wherein the unified space coordinate system is taken as a reference node coordinate system; for each non-reference node i, let a unified time scale With local time stamping Satisfy the following requirements Wherein As a coefficient of the frequency drift, Is clock deviation, the unified time mark Defining as a common time reference after frequency drift and clock bias compensation in a sliding window; for each node i, making the point coordinates in the unified space coordinate system With local coordinates Satisfy the following requirements Wherein Is an external reference rotation matrix, Is an extrinsic translation vector; Meets the conditions of orthogonalization and unit determinant; The arrival time difference, arrival frequency difference and line-of-sight angle difference of the common vision measurement are respectively compared with Establishing an explicit association equation to form a parameter set to be estimated The reference node externally references the fixed constraint.
5. The integrated clock synchronization and spatial registration method for a distributed networking radar according to claim 1, wherein the forming a constraint set across nodes by the common view measurement, stacking round trip time constraint, drift continuity prior and extrinsic parameter smoothing prior, and adopting robust loss suppression outlier measurement comprises: For each common view event, constructing a time residual error, a frequency residual error and an angle residual error between any participating node and a reference node according to an arrival time difference, an arrival frequency difference and a sight angle difference, substituting frequency drift coefficients, clock deviations, an external parameter rotation matrix and an external parameter translation vector into the time residual error, the frequency residual error and the angle residual error, and acquiring the residual error weights of the time residual error, the frequency residual error and the angle residual error from the confidence coefficient of the common view event; when the round trip time measurement records of the transmission and the echo of the two nodes exist, calculating a receiving-transmitting time difference under a unified time scale, and subtracting the geometric round trip time determined by the external parameter translation vector and the propagation speed to obtain a time consistency residual error which is used for enhancing the observability of time parameters; Establishing a differential residual error between adjacent sliding windows for the frequency drift coefficient and clock deviation of the same node, so that the frequency drift coefficient and the clock deviation are continuous and bounded in time, and jump between windows is limited; Establishing smooth constraint on an extrinsic rotation matrix and extrinsic translation of the same node between adjacent sliding windows, parameterizing the extrinsic rotation matrix by adopting a unit quaternion or an equivalent rotation matrix, calculating logarithmic mapping of relative rotation, taking a three-dimensional vector of the logarithmic mapping as an extrinsic rotation matrix difference, and taking the difference between extrinsic translation vectors of the two windows as an extrinsic translation difference; And applying a robust loss function with a bounded influence function to the time residual, the frequency residual, the angle residual, the time consistency residual, the difference residual, the extrinsic rotation matrix difference and the extrinsic translation difference, weighting according to event confidence and preset weight, and summarizing to obtain a constraint set of cross nodes, wherein the robust threshold of the robust loss function is set according to the root mean square of the previous window residual.
6. The integrated method of clock synchronization and spatial registration for a distributed networking radar according to claim 1, wherein online correcting the transmit timing, sampling clock and beam pointing based on the clock and extrinsic parameters, and taking the clock and extrinsic parameters as the next window prior, comprises: According to the unified time scale, mapping and adjusting the local transmission trigger time of each node to ensure that the transmission signals of each node are consistent under the unified time scale; When the hardware does not support continuous compensation, equivalent adjustment is realized by a numerical control oscillation or resampling mode; Converting the expected beam direction in the unified space coordinate system into an instruction direction under the local coordinate system, and adjusting phased array weighting or servo control according to the instruction direction so as to ensure that the beams are correctly aligned in the global coordinate; after correction, checking residual errors of the arrival time and the pointing angle by using first-batch observation data, and triggering re-fine adjustment of the corresponding module when the residual errors exceed a preset threshold value; and recording the clock and external parameter obtained by the current window and the confidence information as initial values of the next window so as to ensure continuity and stability in the sliding window.
7. The integrated clock synchronization and spatial registration method for a distributed networking radar according to claim 1, wherein the observability enhancement is triggered when any of the arrival time difference residual root mean square exceeds a preset first threshold, or the line of sight angle difference residual root mean square exceeds a preset second threshold, or the chi-square statistics of normalized residuals exceeds a preset third threshold, the observability enhancement comprising adding bidirectional time measurement and adjusting operating parameters comprises: After the current sliding window is solved, respectively calculating the chi-square statistics of the arrival time difference residual root mean square, the line of sight angle difference residual root mean square and the normalized residual, and comparing with a corresponding preset threshold value; when the observability is judged to be insufficient, an enhancement instruction is issued to each node in the network, and additional time measurement tasks and parameter adjustment strategies are appointed to be executed; under the enhanced instruction, each node initiates additional round trip time measurement operation according to a scheduling protocol and returns a time measurement result so as to increase the independent measurement quantity in a constraint equation; Aiming at the situation that the sight geometry is poor, the quantity of the co-view targets is increased or the angle distribution is more balanced by adjusting the transmitting power, the working beam coverage range or the transmitting pulse scheduling of part of nodes; and taking the newly added round trip time measurement result and new measurement generated by parameter adjustment into a constraint set of a next sliding window together as an enhanced observation condition.

Description

Clock synchronization and spatial registration integrated method for distributed networking radar Technical Field The invention relates to the technical field of radar signal processing, in particular to a clock synchronization and spatial registration integrated method for a distributed networking radar. Background The distributed networking radar is cooperated with a plurality of sub-radar nodes to improve the target detection distance, the angle resolution and the anti-interference capability and support the omnidirectional coverage and the information redundancy under a multi-base system. In engineering, an external time reference is generally adopted to carry out clock synchronization, a calibration target or an external reference measurement flow is adopted to carry out space registration, each sub-radar independently operates under a local time scale and a local coordinate system, and then an upper fusion link is adopted to carry out time and space alignment. With the increase of multi-platform, maneuvering and elastic networking applications, the network topology is dynamically changed, the target scene is more complex, the availability of available links between nodes and external absolute time references is unstable, the time drift, frequency deviation and node external parameter change are required to be processed simultaneously under the online running condition, and a unified time scale and a unified space coordinate system are continuously maintained in a sliding time domain so as to ensure that cross-node coherent accumulation, angle measurement and ranging joint inversion and target track stability association are ensured. In the prior art, clock synchronization and spatial registration splitting are implemented, external reference or off-line calibration is relied on, and the coupling between time errors and external reference errors is not utilized by modeling, so that when the external reference fails, the calibration frequency is insufficient or the scene changes rapidly, the reliability of cross-node coherent processing and track splicing is reduced. The existing method lacks an online integrated mechanism for simultaneously estimating a clock and an external parameter based on co-view target observation under the condition of not depending on an external absolute time reference, and is difficult to ensure continuous and steady network-level cooperative perception quality. Disclosure of Invention In order to overcome the defects of the prior art, the invention aims to provide the clock synchronization and space registration integration method for the distributed networking radar, which constructs cross-node constraint through common view measurement and combines prior and robust optimization, so that the distributed networking radar can still complete high-precision clock synchronization and space registration under the conditions of no external time service and dynamic networking, thereby guaranteeing the long-term consistency and collaborative detection performance of the whole network. In order to achieve the above object, the present invention provides the following solutions: A clock synchronization and spatial registration integration method for a distributed networking radar comprises the following steps: Acquiring local time scale echo and pose data of each sub-radar, performing cross-node correlation in a sliding window according to distance, angle and Doppler consistency, judging a common-view event, and calculating common-view measurement, wherein the common-view measurement comprises an arrival time difference, an arrival frequency difference and a sight angle difference respectively; Setting clock deviation, frequency drift, an extrinsic rotation matrix and extrinsic translation for each node, and fixing extrinsic parameters of a reference node to eliminate global uncertainty; forming a constraint set of a cross node by using the common view measurement, superposing constraint during round trip time, drift continuity priori and external parameter smooth priori, and inhibiting outlier measurement by using robust loss; performing sparse nonlinear least square solution on the constraint set in a sliding window to obtain clock and external parameter of each node, thereby forming a unified time scale and a unified space coordinate system; Correcting the transmitting time sequence, the sampling clock and the beam direction on line based on the clock and the external parameter, and taking the clock and the external parameter as the prior of the next window; The observability enhancement is triggered when either the root mean square of the arrival time difference residual exceeds a preset first threshold, or the root mean square of the line of sight angle difference residual exceeds a preset second threshold, or the chi-square statistics of the normalized residual exceeds a preset third threshold, including increasing the bi-directional time measurement and adjusting the operating