A low-latency, low-communication-overhead navigation method for anti-burst link interruption

Abstract

This invention discloses a low-latency, low-communication-overhead navigation method for resisting sudden link interruptions, relating to the fields of wireless communication, cooperative positioning, and graph signal processing. The invention constructs a momentum-accelerated graph filter in a distributed network, accelerating convergence by introducing an inertial momentum term into node state updates, and adaptively adjusting the polynomial truncation order using residual energy detection to reduce latency and communication overhead. Simultaneously, a spatiotemporal two-dimensional trust region model is constructed, combining temporal consistency and spatial geometric verification to perform real-time trust scoring of neighboring node links. Finally, based on the trust score, a robust M-estimator based on the Huber kernel function is used to update the graph displacement operator. This invention solves the problems of slow convergence and high communication overhead in traditional algorithms, and can effectively cope with sudden link interruptions and measurement drift in highly dynamic scenarios, achieving high-precision, highly robust distributed cooperative navigation.

Description

Technical Field

[0001] This invention relates to the fields of wireless communication, cooperative positioning, and graph signal processing, and in particular to a low-latency, low-overhead navigation method for resisting sudden link interruptions. Background Technology

[0002] With the rapid development of emerging application scenarios such as the Internet of Things (IoT), the Internet of Vehicles (IoV), and the Industrial Internet, the trend of interconnectedness is becoming increasingly prominent. Distributed collaborative navigation technology, as the core foundation supporting the precise operation of various scenarios, is becoming increasingly important. In the field of intelligent transportation, the network topology of the IoV exhibits highly dynamic characteristics. The high-speed movement of vehicles, frequent lane changes, and complex fluctuations in traffic flow place stringent demands on the real-time performance and accuracy of navigation and positioning. Vehicles need to achieve advanced functions such as collaborative obstacle avoidance and autonomous driving through precise location information exchange. This requires the navigation system to be able to complete location updates and information synchronization within milliseconds. In Industry 4.0 scenarios, AGV robots, industrial robotic arms, and other equipment in smart warehouses need to perform high-precision collaborative operations in dense working environments. The precise handling of materials and the collaborative operation of equipment both rely on sub-meter or even centimeter-level positioning accuracy. Any positioning deviation may lead to production interruption or equipment damage.

[0003] Meanwhile, the iteration of communication technologies has also driven the continuous upgrading of positioning accuracy requirements. As the core body for setting global mobile communication standards, 3GPP has continuously improved the performance indicators of positioning technology: in the 4G era, positioning accuracy requirements reached 50 meters, meeting basic needs such as general navigation and location reporting; in the 5G era, the R16 standard improved the positioning accuracy for commercial applications to better than 3 meters, supporting in-depth applications in scenarios such as precise dispatching for ride-hailing services and smart city security monitoring; and the R17 standard further proposed positioning accuracy requirements better than 0.2 meters for industrial IoT applications, while the vehicle-to-everything (V2X) field further pursues ultra-high accuracy of 0.1 meters to adapt to advanced application scenarios such as autonomous driving and vehicle-to-infrastructure (V2I) communication. These increasingly stringent accuracy requirements pose unprecedented challenges to the performance of distributed cooperative navigation technologies.

[0004] However, existing collaborative positioning technologies face two major bottlenecks in practical applications, which severely restrict their deployment and application in complex scenarios.

[0005] First, the inherent contradiction between high accuracy and low latency is difficult to reconcile. Currently, widely used algorithms in distributed positioning, such as the Belief Propagation (BP) algorithm, have slow iterative convergence speeds, leading to a significant increase in positioning latency. Furthermore, the large number of message exchanges generated by multiple iterations consumes valuable bandwidth, increasing the risk of network congestion. Existing technologies lack effective mechanisms to accelerate convergence, making it difficult to achieve a high-accuracy steady-state level with very few iterations.

[0006] Secondly, there is a lack of effective mechanisms to address sudden link interruptions and measurement drift. In highly dynamic environments, non-line-of-sight propagation, electromagnetic interference, or attacks can cause severe deviations in ranging values. Existing methods are mostly based on the assumption of an ideal link. When a link malfunctions, the system cannot promptly identify the faulty link and dynamically adjust the weights, leading to a sharp drop in positioning accuracy and easy trajectory jumps.

[0007] Therefore, how to overcome the limitations of existing technologies and design a distributed cooperative navigation system that can achieve low latency, low communication overhead, and strong anti-interference and anti-interruption capabilities while ensuring ultra-high accuracy has become a key technical challenge to support the high-quality development of fields such as the Internet of Things, the Internet of Vehicles, and the Industrial Internet. It has important theoretical research value and urgent engineering application needs. Summary of the Invention

[0008] In view of this, this invention proposes a low-latency, low-communication-overhead navigation method for resisting sudden link interruptions. This method introduces momentum acceleration graph filtering iteration, spatiotemporal two-dimensional trust region evaluation, and an adaptive weighting mechanism based on the Huber kernel function to achieve high-precision positioning in complex environments while significantly reducing system convergence latency and communication overhead.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0010] A low-latency, low-overhead navigation method for resisting sudden link interruptions includes the following steps:

[0011] Step 1: Deploy around and above the work area A reference node with a precisely known location assists in the work area. Navigate to the nodes to be navigated;

[0012] Step 2, construct the initial graph displacement operator;

[0013] Step 3: Perform a momentum-accelerated navigation process for each node to be navigated;

[0014] Step 4: Perform anti-interruption robust fusion processing and trust domain evaluation on the links between the navigation nodes to obtain the corresponding link trust scores;

[0015] Step 5: Calculate the current link information based on the reference node and the node to be navigated to construct a preliminary graph displacement operator, and perform a weighted update by combining the link trust score to obtain the current graph displacement operator;

[0016] Step 6: Calculate the graph filter response based on the current graph shift operator;

[0017] Step 7: Iterate through steps 3 to 6 until all nodes to be navigated are determined to have reached the navigation endpoint, thus completing low-latency, low-communication-overhead navigation to withstand sudden link interruptions.

[0018] Furthermore, the initial graph displacement operator in step 2 is: An identity matrix of order 1. .

[0019] Furthermore, step 3 is performed as follows:

[0020]

[0021] in, This represents the two-dimensional position vector of the i-th node to be navigated during the k-th iteration. Let be the known initial position of the i-th node to be navigated. i=1,2,3,……, , For the graph displacement operator in the k-th iteration, This is the initial graph displacement operator. That is The corresponding graphical filter response, ; , .

[0022] Furthermore, step 4 is specifically implemented as follows:

[0023] Step 401, Time Dimension Consistency Verification: For the i-th node to be navigated, select node j with a direct communication link with it, and calculate the change in monitoring signal strength between the i-th node to be navigated and node j. Deviation from physical radial velocity ;like And physical radial velocity deviation If so, then the instantaneous score is based on the time dimension. Otherwise, the instantaneous score in the time dimension is:

[0024]

[0025] in, The penalty factor has a range of values. , , , The measured radio frequency distance between the i-th node to be navigated and node j; and Let be the two-dimensional position vectors of the i-th node to be navigated and node j in the k-th iteration, respectively; The link reliability score is given during the k-th iteration. ;

[0026] Step 402, Spatial Dimension Geometric Verification: Calculate the Euclidean distance between the i-th node to be navigated and node j. If the Euclidean distance is less than a preset threshold... Then the instantaneous score of spatial dimension Otherwise, instantaneous scoring of spatial dimensions ;

[0027] Step 403, weighted fusion calculation of instantaneous score:

[0028] ; , ;

[0029] Step 404, calculate the latest link trust score:

[0030] , For smoothing weights, the range of values ​​is... .

[0031] Furthermore, the specific method in step 5 is as follows:

[0032] Step 501: Construct the preliminary graph displacement operator , for Square array ,correspond Reference nodes and One node awaiting navigation;

[0033]

[0034] , Indicates the relationship with node during the (k+1)th iteration. A set of neighboring nodes with direct communication links. That is, a set The number of elements in;

[0035] Step 502, if the link trust score ,but

[0036]

[0037]

[0038] Set linearity error value If the link trust score Then judge Is it true? If so, then let Otherwise

[0039]

[0040] If the link trust score ,but

[0041]

[0042] in, ;

[0043] Step 503, apply the preliminary graph displacement operator element values ​​in Replace with The current graph displacement operator is obtained. .

[0044] Furthermore, the specific method in step 6 is as follows:

[0045] Step 601: Calculate the residual energy of the i-th node to be navigated. :

[0046]

[0047] Set the estimated high dynamic threshold ,like Then higher order is used. As a graph filter response If the total order is M, then use a lower order. As a graph filter response The total order M;

[0048] Step 602, Construct the graph filter response , The process of obtaining the value is as follows:

[0049] Set the target frequency response of the ideal low-pass filter. , for The eigenvalues ​​are then approximated using Chebyshev polynomials or least squares fitting methods. eigenvalue range Internal solution for optimal coefficients , making and The mean square error is the smallest;

[0050]

[0051] This is a preset feature value threshold.

[0052] Furthermore, the specific method in step 7 is as follows:

[0053] calculate Whether it is valid, Let be the coordinates of the navigation endpoint for the i-th node to be navigated. Let $\frac{i}{i}$ be the position convergence threshold. If this threshold is met, the i-th node to be navigated is determined to have reached the navigation endpoint, the iteration process of that node is terminated, and the position of the node to be navigated is saved. The process iterates over the remaining nodes that have not yet reached the navigation endpoint until all nodes are determined to have reached the navigation endpoint.

[0054] Due to the adoption of the above technical solution, the beneficial effects of this invention compared with the prior art are as follows:

[0055] This invention benefits from the momentum acceleration mechanism of fractional decay, and the system breaks through the traditional consensus algorithm. Convergence constraints, to achieve The system achieves accelerated convergence. Experimental data shows that, under the same accuracy requirements, its convergence time is reduced by 30% compared to the traditional BP algorithm. Furthermore, while significantly reducing the number of iterations, the decrease in positioning accuracy is strictly controlled within 20%, and the average positioning error under normal conditions is less than 1 meter. When faced with abnormal situations such as sudden link interruptions, non-line-of-sight (NLOS) occlusion, and position drift, the system utilizes a spatiotemporal dual-dimensional trust region assessment and a Huber robust M-estimation mechanism to perform adaptive weight reduction operations on abnormal links in real time. The system can tolerate a link anomaly rate as high as 20% and maintains a stable positioning error within the 1.2-meter range even under complex interference environments, effectively solving the problems of sudden failures and trajectory divergence in dynamic navigation scenarios. Attached Figure Description

[0056] Figure 1 This is a schematic diagram illustrating the interaction between cluster size and node deployment in an embodiment of the present invention.

[0057] Figure 2 This is a schematic diagram of the space-time reliability assessment and score decay curve in an embodiment of the present invention.

[0058] Figure 3 This is a schematic diagram comparing adaptive communication truncation and residual energy monitoring in an embodiment of the present invention. Detailed Implementation

[0059] The invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0060] A low-latency, low-overhead navigation method for resisting sudden link interruptions, such as... Figure 1 As shown, this embodiment is a collaborative navigation scenario of AGVs (Automated Guided Vehicles) in smart warehousing (Industry 4.0), and provides a detailed description of the distributed collaborative navigation and communication integration technology provided by the present invention.

[0061] Cluster Size and Node Deployment: The system comprises five precisely located reference nodes (anchor nodes / base stations), A1-A5 in the diagram, deployed around the perimeter and top of the work area, and 11 AGV agent nodes performing highly dynamic movement tasks. The overall system architecture and the interaction logic between nodes are shown in Figure 1. The anchor nodes serve as a global coordinate reference source, providing absolute position references for all AGV agent nodes. The AGV nodes, in turn, combine the reference coordinates of the anchor nodes, wireless ranging data between nodes, and distributed cooperative navigation algorithms to achieve high-precision self-positioning.

[0062] Hardware platform and communication front end: Each AGV node is equipped with a high-performance embedded processing platform (such as Zedboard or a similar FPGA+ARM architecture processor). The radio frequency transceiver front end adopts a UWB (ultra-wideband) or 5G integrated communication and navigation module, supporting millisecond-level ranging and data interaction between nodes within a 100-meter range.

[0063] Spatiotemporal discretization operation logic: During the physical operation of the system, the coordinates of each reference node remain unchanged, and each proxy node (i.e., the node to be navigated) executes in parallel. The distributed iterative update gradually approximates the real physical coordinates through multiple rounds of local information exchange.

[0064] Specifically, the following steps are included:

[0065] Step 1: Deploy around and above the work area A reference node with a precisely known location assists in the work area. Navigate to the nodes to be navigated;

[0066] Step 2, construct the initial graph displacement operator;

[0067] Step 3: Perform a momentum-accelerated navigation process for each node to be navigated;

[0068] Step 4: Perform anti-interruption robust fusion processing and trust domain evaluation on the links between the navigation nodes to obtain the corresponding link trust scores;

[0069] Step 5: Calculate the current link information based on the reference node and the node to be navigated to construct a preliminary graph displacement operator, and perform a weighted update by combining the link trust score to obtain the current graph displacement operator;

[0070] Step 6: Calculate the graph filter response based on the current graph shift operator;

[0071] Step 7: Iterate through steps 3 to 6 until all nodes to be navigated are determined to have reached the navigation endpoint, thus completing low-latency, low-communication-overhead navigation to withstand sudden link interruptions.

[0072] Furthermore, the initial graph displacement operator in step 2 is: An identity matrix of order 1. .

[0073] Furthermore, step 3 is performed as follows:

[0074]

[0075] in, This represents the two-dimensional position vector of the i-th node to be navigated during the k-th iteration. Let be the known initial position of the i-th node to be navigated. i=1,2,3,……, Physically, this means that the algorithm's "virtual velocity" is zero at the initial moment, allowing the position estimation to start smoothly from rest and avoiding instantaneous coordinate jumps due to a lack of historical information. For the graph displacement operator in the k-th iteration, This is the initial graph displacement operator. That is The corresponding graphical filter response, ; , By leveraging the "inertia" of historical update directions, the network can quickly overcome error extremes, thus increasing the overall network consistency convergence rate from... Upgraded to .

[0076] Furthermore, during AGV operation, if measurement anomalies occur due to shelf obstruction or electromagnetic interference, the system will activate as follows: Figure 2 Robust processing flow shown:

[0077] Step 401, Time Dimension Consistency Verification: For the i-th node to be navigated, select node j with a direct communication link with it, and calculate the change in monitoring signal strength between the i-th node to be navigated and node j. Deviation from physical radial velocity ;like And physical radial velocity deviation If so, then the instantaneous score is based on the time dimension. Otherwise, the instantaneous score in the time dimension is:

[0078]

[0079] in, The penalty factor has a range of values. , , , The measured radio frequency distance between the i-th node to be navigated and node j; and Let be the two-dimensional position vectors of the i-th node to be navigated and node j in the k-th iteration, respectively; The link reliability score is given during the k-th iteration. If, during the k-th iteration, there is no direct communication link between the i-th node to be navigated and node j, then By using this exponential decay formula, the dynamic score of sudden interference links is reduced, abnormal data is avoided from participating in the positioning calculation, the stability and accuracy of AGV cluster positioning are guaranteed, and the real-time response of the system is also taken into account.

[0080] Wherein, the change in monitoring signal strength between the i-th node to be navigated and node j That is, the monitoring signal strength in two adjacent iterations. and The absolute value of the difference is used to determine if there is no direct communication link between the i-th node to be navigated and node j during the k-th iteration. Physical radial velocity deviation The calculation process is as follows: Obtain the measured radial velocity between the i-th node to be navigated and node j based on the time-of-flight (TOF) of the wireless ranging signal. Simultaneously, by combining the acceleration and angular velocity data collected by the AGV's own inertial measurement unit (IMU), the trajectory estimation (DR) algorithm is used to calculate the distance between the i-th node to be navigated and the node at the same time. Theoretical radial velocity Then calculate the physical radial velocity deviation. Measured radial velocity The distance measurement values ​​from two consecutive iterations , and the duration between two iterations Calculated, i.e. Theoretical radial velocity The moving velocity vector obtained by integrating the IMU acceleration of the i-th node to be navigated is then compared with the i-th node to be navigated and the node. The final result is obtained by performing a dot product operation on the unit radial vectors between them. It can quantitatively reflect the degree of deviation between the measured radial velocity and the physical laws of motion, providing a core basis for judgment on consistency verification in the time dimension.

[0081] Step 402, Spatial Dimension Geometric Verification: Calculate the Euclidean distance between the i-th node to be navigated and node j. If the Euclidean distance is less than a preset threshold... (e.g., set according to IMU drift rate) (meters), then the instantaneous score of spatial dimension. Otherwise, instantaneous scoring of spatial dimensions ;

[0082] Step 403, weighted fusion calculation of instantaneous score:

[0083] ; , ;

[0084] In this embodiment, This achieves balanced verification across two dimensions.

[0085] Step 404, calculate the latest link trust score:

[0086] , For smoothing weights, the range of values ​​is... This mechanism ensures that the score recovers smoothly rather than abruptly after the link interference disappears, enhancing the robustness of the distributed navigation system in complex electromagnetic environments.

[0087] Furthermore, the specific method in step 5 is as follows:

[0088] Based on the robust M-estimation and adaptive weighting mechanism using the Huber kernel function, the system uses the aforementioned spatiotemporal frequency score. A refined physical fusion strategy is implemented for each communication link to eliminate the impact of non-line-of-sight (NLOS) errors or measurement anomalies on the positioning results.

[0089] Step 501: Construct the preliminary graph displacement operator , for Square array ,correspond Reference nodes and One node awaiting navigation;

[0090]

[0091] , Indicates the relationship with node during the (k+1)th iteration. A set of neighboring nodes with direct communication links. That is, a set The number of elements in;

[0092] Step 502, if the link trust score If the system determines that the link is in a normal state, it will perform information fusion using standard Gaussian weights and execute a positive enhancement strategy.

[0093]

[0094]

[0095] Set linearity error value (In this embodiment, The range is 0.2 to 0.5 meters (preferred example: 0.3 meters), if the link reliability score... If the system triggers a soft interrupt and activates the Huber kernel robust estimator, it will determine... If this is true, it means that although there is slight interference in the link, the ranging residual is within a reasonable range. Therefore, let... This indicates a significant measurement anomaly in the link, making

[0096]

[0097] If the link trust score The system determines that the link is in a hard interrupt state. To avoid abnormal data corruption and retain the link recovery capability, then...

[0098]

[0099] in, In this state, the link information is severely suppressed but not completely disabled.

[0100] Step 503, apply the preliminary graph displacement operator element values ​​in Replace with The current graph displacement operator is obtained. .

[0101] In this system, the node to be navigated communicates with other nodes to be navigated and reference nodes, while reference nodes only communicate with the node to be navigated; when The corresponding nodes are: when one is the node to be navigated and the other is the reference node,

[0102]

[0103] Through the aforementioned adaptive weighting, even if individual neighboring nodes transmit abnormal coordinates due to interference, the final fused navigation state (including position, speed, and heading angle) is still dominated by the high-scoring link and the results calculated by the local inertial navigation system. This ensures that the AGV's positioning trajectory in complex warehousing environments will not experience instantaneous jumps or divergence, achieving sub-meter level continuous and stable navigation.

[0104] Furthermore, the specific method in step 6 is as follows:

[0105] Step 601: Calculate the residual energy of the i-th node to be navigated. :

[0106]

[0107] like Figure 3 As shown, set the estimated high dynamic threshold. ,like Then higher order is used. As a graph filter response The total order M is the filter order; otherwise, a lower order is used. As a graph filter response The total order M; , ;

[0108] In this embodiment, Higher order That is Figure 3 High dynamic model 12, low order That is Figure 3 Steady-state mode 3 in the middle;

[0109] Step 602, Construct the graph filter response , The process of obtaining the value is as follows:

[0110] Set the target frequency response of the ideal low-pass filter. , for The eigenvalues ​​are then approximated using Chebyshev polynomials or least squares fitting methods. eigenvalue range Internal solution for optimal coefficients , making and The mean square error is the smallest;

[0111]

[0112] In this embodiment, a preset feature value threshold is used. The value ranges from 0.1 to 0.3, with 0.2 being preferred. S is a weight matrix dynamically constructed based on the wireless communication connectivity between nodes. This matrix is ​​dynamically constructed based on the real-time changes in the wireless communication connectivity between nodes. When the communication link between nodes is interrupted due to movement or obstruction, the distribution of non-zero elements of S will be updated synchronously, always consistent with the actual communication topology of the current cluster.

[0113] Furthermore, the specific method in step 7 is as follows:

[0114] calculate Whether it is valid, Let be the coordinates of the navigation endpoint for the i-th node to be navigated. Let $\frac{i}{i}$ be the position convergence threshold. If this threshold is met, the i-th node to be navigated is determined to have reached the navigation endpoint, the iteration process of that node is terminated, and the position of the node to be navigated is saved. The system iterates through the remaining nodes that have not yet reached the navigation endpoint until all nodes are determined to have reached the navigation endpoint. The coordinates of the reference node remain constant throughout the iteration. The core criterion for terminating the iteration is the global convergence threshold of the AGV cluster position estimate, rather than the distance from a single node to the target point (because this invention focuses on globally consistent cluster positioning, not single-point path planning). In this embodiment, The value is 1 mm; the maximum change in the estimated position of all AGV nodes in two adjacent iterations is less than 1 mm, the cluster position has reached a globally stable state, and continuing the iteration will not significantly improve the positioning accuracy, but will increase the communication overhead and computation latency. This termination condition takes into account both the positioning accuracy and the real-time requirements of the system.

[0115] The core of this invention lies in designing a "distributed cooperative navigation and communication integration technology". By exploring the performance gains of communication and navigation integration, it can significantly reduce the convergence time while ensuring steady-state accuracy close to the existing optimal algorithm, and has strong fault tolerance in the event of link interruption or measurement anomaly.

[0116] Low-Delay Graph Filter Design Based on Momentum Acceleration and Adaptive Truncation: To overcome the bottleneck of slow convergence, this invention constructs a distributed graph filter that introduces a fractionally decaying inertial momentum term. Unlike traditional static weight design, this method introduces a momentum acceleration mechanism based on historical gradients in node state updates, utilizing "inertia" to quickly overcome local extrema, thereby improving the consistent convergence rate to [missing information]. Meanwhile, in conjunction with an adaptive polynomial truncation strategy based on residual energy detection, the communication order is automatically reduced when the iteration approaches a steady state, effectively reducing redundant message interactions and lowering positioning latency.

[0117] Spatiotemporal Trust Region Assessment and Robust M-Estimation Fusion: To address sudden link interruptions and malicious attacks, this invention departs from single hard-decision detection and introduces a "spatiotemporal dual-dimensional trust region" model. By combining the temporal continuity of the received signal with the spatial geometric consistency calculated by the local inertial navigation system, a real-time "soft score" of the trust level of neighboring nodes is performed. In the information fusion stage, a robust M-estimator based on the Huber kernel function replaces the traditional Gaussian product fusion. By limiting the weight of large residual messages, the impact of abnormal data on the positioning results is automatically suppressed, providing strong fault tolerance for sudden link interruptions and measurement drift.

[0118] Those skilled in the art will recognize that the described embodiments are intended to help readers understand the principles of the invention and should be understood as not limiting the scope of protection of the invention to the described embodiments. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the scope of the claims of the invention.

Claims

1. A low-latency, low-overhead navigation method for resisting sudden link interruptions, characterized in that, Includes the following steps: Step 1: Deploy around and above the work area A reference node with a precisely known location assists in the work area. Navigate to the nodes to be navigated; Step 2, construct the initial graph displacement operator; Step 3: Perform a momentum-accelerated navigation process for each node to be navigated; Step 4: Perform anti-interruption robust fusion processing and trust domain evaluation on the links between the navigation nodes to obtain the corresponding link trust scores; Step 5: Calculate the current link information based on the reference node and the node to be navigated to construct a preliminary graph displacement operator, and perform a weighted update by combining the link trust score to obtain the current graph displacement operator; Step 6: Calculate the graph filter response based on the current graph shift operator; Step 7: Iterate through steps 3 to 6 until all nodes to be navigated are determined to have reached the navigation endpoint, thus completing low-latency, low-communication-overhead navigation to withstand sudden link interruptions.

2. The navigation method for resisting sudden link interruptions with low latency and low communication overhead according to claim 1, characterized in that, The initial graph displacement operator in step 2 is: An identity matrix of order 1. .

3. The navigation method for resisting sudden link interruptions with low latency and low communication overhead according to claim 1, characterized in that, The specific method for step 3 is as follows: in, This represents the two-dimensional position vector of the i-th node to be navigated during the k-th iteration. Let be the known initial position of the i-th node to be navigated. i=1,2,3,……, , For the graph displacement operator in the k-th iteration, This is the initial graph displacement operator. That is The corresponding graphical filter response, ; , .

4. The navigation method for resisting sudden link interruptions with low latency and low communication overhead according to claim 1, characterized in that, The specific method for step 4 is as follows: Step 401, Time Dimension Consistency Verification: For the i-th node to be navigated, select node j with a direct communication link with it, and calculate the change in monitoring signal strength between the i-th node to be navigated and node j. Deviation from physical radial velocity ;like And physical radial velocity deviation If so, then the instantaneous score is based on the time dimension. Otherwise, the instantaneous score in the time dimension is: in, The penalty factor has a range of values. , , , The measured radio frequency distance between the i-th node to be navigated and node j; and Let be the two-dimensional position vectors of the i-th node to be navigated and node j in the k-th iteration, respectively; The link reliability score is given during the k-th iteration. ; Step 402, Spatial Dimension Geometric Verification: Calculate the Euclidean distance between the i-th node to be navigated and node j. If the Euclidean distance is less than a preset threshold... Then the instantaneous score of spatial dimension Otherwise, instantaneous scoring of spatial dimensions ; Step 403, weighted fusion calculation of instantaneous score: ； ， ； Step 404, calculate the latest link trust score: , For smoothing weights, the range of values ​​is... .

5. A low-latency, low-communication-overhead navigation method for resisting sudden link interruptions according to claim 4, characterized in that, The specific method in step 5 is as follows: Step 501: Construct the preliminary graph displacement operator , for Square array ,correspond Reference nodes and One node awaiting navigation; , Indicates the relationship with node during the (k+1)th iteration. A set of neighboring nodes with direct communication links. That is, a set The number of elements in; Step 502, if the link trust score ,but Set linearity error value If the link trust score Then judge Is it true? If so, then let Otherwise ; If link trust score ,but in, ; Step 503, apply the preliminary graph displacement operator element values ​​in Replace with The current graph displacement operator is obtained. .

6. The navigation method for resisting sudden link interruptions with low latency and low communication overhead according to claim 1, characterized in that, The specific method in step 6 is as follows: Step 601: Calculate the residual energy of the i-th node to be navigated. : Set the estimated high dynamic threshold ,like Then higher order is used. As a graph filter response If the total order is M, then use a lower order. As a graph filter response The total order M; Step 602, Construct the graph filter response , The process of obtaining the value is as follows: Set the target frequency response of the ideal low-pass filter. , for The eigenvalues ​​are then approximated using Chebyshev polynomials or least squares fitting methods. eigenvalue range Internal solution for optimal coefficients , making and The mean square error is the smallest; This is a preset feature value threshold.

7. The navigation method for resisting sudden link interruptions with low latency and low communication overhead according to claim 1, characterized in that, The specific method in step 7 is as follows: calculate Whether it is valid, Let be the coordinates of the navigation endpoint for the i-th node to be navigated. Let $\frac{i}{i}$ be the position convergence threshold. If this threshold is met, the i-th node to be navigated is determined to have reached the navigation endpoint, the iteration process of that node is terminated, and the position of the node to be navigated is saved. The process iterates over the remaining nodes that have not yet reached the navigation endpoint until all nodes are determined to have reached the navigation endpoint.

Images