A remote communication method and system for unmanned vehicles

Abstract

This invention relates to the field of vehicle communication technology and discloses a remote communication method and system for unmanned vehicles. The method includes: acquiring a set of effective interference points in the environment surrounding the unmanned vehicle and calculating a dynamic occlusion index; simultaneously acquiring underlying physical channel quality parameters and performing cross-layer bilateral determination based on the dynamic occlusion index and the underlying physical channel quality parameters; when the cross-layer bilateral determination meets the triggering conditions, entering a link protection state, performing flow interception operations on regular service flows, and implementing enhanced transmission strategies for core control commands; when preliminary recovery conditions are detected, initiating a backoff recovery window, performing progressive service recovery during the backoff recovery window period, and performing network topology rollback after the backoff recovery window expires. This invention improves the accuracy of link interruption prediction, ensures reliable transmission of critical commands when the channel deteriorates, and effectively prevents network state oscillations caused by frequent signal fluctuations in occlusion edge areas.

Description

Technical Field

[0001] This invention relates to the field of vehicle communication technology, specifically to a remote communication method and system for unmanned vehicles. Background Technology

[0002] During operation, autonomous vehicles need to maintain real-time communication with remote control centers or roadside units to receive core control commands and upload routine business data. In complex environments such as urban roads, vehicles often face physical obstructions from buildings, large vehicles, or other obstacles, leading to a sudden decrease in communication link quality or even complete disruption.

[0003] Existing remote communication scheduling strategies typically rely on feedback from underlying physical channel metrics to trigger network protection mechanisms. Due to the time overhead in the statistical and feedback process of these metrics, this mechanism exhibits a response lag when facing sudden physical disruptions, leading to a large backlog of pending data in the gateway's buffer, resulting in queue congestion and overall network latency. Some existing solutions attempt to introduce environmental awareness data to predict link disruptions in advance; however, relying solely on spatial awareness information is susceptible to interference from inherent sensor noise or differences in the reflective properties of the environmental medium, leading to false alarms and causing resource constraints and waste during normal communication.

[0004] During the communication link recovery phase, when the autonomous vehicle (RV) travels to the edge of or out of an obstructed area, short-term changes in the spatial environment can cause frequent fluctuations in physical signal strength. Existing communication systems, upon detecting a short-term rise in channel metrics, typically immediately restore normal data transmission and switch back to the original network connection topology. This instantaneous response to signal fluctuations leads to repeated switching of service scheduling rules and access paths, causing data queue fluctuations, disordered changes in transmission delay, and repeated reconstruction of control plane signaling, thus reducing the overall stability of the RV's remote communication system. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a remote communication method and system for unmanned vehicles. It solves the problems in existing remote communication systems for unmanned vehicles, such as the scheduling strategy relying on feedback from underlying channel indicators having a response lag, easily causing gateway buffer congestion when the physical link is blocked, and relying solely on environmental perception and prediction easily generating false alarms due to sensor noise or differences in media properties.

[0006] The first aspect of this invention provides a remote communication method for an unmanned vehicle, comprising: Obtain the effective set of interference points in the environment surrounding the unmanned vehicle, and calculate the dynamic occlusion index based on the effective set of interference points to characterize the degree of communication line-of-sight occlusion. Synchronously acquire the underlying physical channel quality parameters, and perform cross-layer bilateral determination by combining the dynamic occlusion index and the underlying physical channel quality parameters; When the cross-layer bilateral determination meets the triggering conditions, it enters the link protection state. In the link protection state, the regular service flow is intercepted and the core control command is enhanced by a transmission strategy. When the initial recovery conditions are detected, a backoff recovery window is initiated. During the duration of the backoff recovery window, a gradual service recovery is performed, and after the backoff recovery window expires, a network topology rollback is performed.

[0007] Furthermore, the calculation of the dynamic occlusion index, used to characterize the degree of communication line-of-sight occlusion, specifically includes: calculating a spatial weighting coefficient for each discrete point in the effective interference point set, wherein the spatial weighting coefficient is jointly determined by the orthogonal distance from the discrete point to the communication line-of-sight ray and the normalized reflection characteristic of the discrete point; combining the electromagnetic attenuation coefficient of each discrete point with its corresponding spatial weighting coefficient, and introducing an equivalent local volume element for discrete summation calculation to obtain the original occlusion index; and performing normalization processing on the original occlusion index using a logic function to generate a scalar value that continuously reflects the occlusion risk as the dynamic occlusion index.

[0008] Furthermore, the step of performing cross-layer bilateral determination by combining the dynamic occlusion index and the underlying physical channel quality parameters includes: calculating the temporal fading change rate of the signal-to-noise ratio measurement based on a set time step; when the dynamic occlusion index is greater than a preset occlusion warning threshold and the temporal fading change rate is lower than the maximum tolerable fading rate threshold, determining that the cross-layer bilateral determination meets the triggering condition.

[0009] Further, the process of throttling regular service flows includes: calculating a dynamic throttling factor for the regular service flows based on the dynamic occlusion index and the occlusion warning threshold; redirecting the regular service data packets to be sent to a suspended buffer or actively discarding them according to the throttling ratio represented by the dynamic throttling factor. The process of implementing an enhanced transmission strategy for core control commands includes: intercepting the core control commands to be sent and extracting the key information payload of the core control commands; downgrading the network transmission protocol layer and reconstructing and encapsulating the extracted key information payload into a connectionless User Datagram Protocol (UDP) message or a custom Ethernet frame for transmission; attaching a service layer sequence number to the reconstructed and encapsulated message; synchronously starting a retransmission timer; and reporting a transmission failure to the upper-layer application if no application layer status confirmation packet is received after reaching the maximum number of retransmissions or the maximum duration. Simultaneously, the enhanced transmission strategy also includes implementing an information redundancy strategy, a time redundancy strategy, or a multi-path redundancy strategy based on the risk level range of the dynamic occlusion index.

[0010] Further, initiating the backoff recovery window includes: calculating a normalized recovery determination quantity by weighted summation based on the current value of the dynamic occlusion index, the signal-to-noise ratio measurement value, and the confirmation success rate of the core control command; when the recovery determination quantity is higher than a preset recovery entry threshold within a set continuous sampling period, it is determined that the preliminary recovery condition is met, and the backoff recovery window is initiated; the duration of the backoff recovery window is calculated and set according to the current recovery determination quantity and the preset minimum recovery observation duration. Performing progressive service recovery during the duration of the backoff recovery window includes: constructing a recovery opening coefficient that gradually increases with the elapsed time of the backoff recovery window; calculating an actual gating coefficient based on the recovery opening coefficient to relax the interception operation on the regular service flow, and gradually reducing the redundant replica number of the core control command according to the recovery opening coefficient; during the duration of the backoff recovery window, if the dynamic occlusion index is detected to be higher than the deterioration determination threshold or the signal-to-noise ratio measurement value is detected to be lower than the lower limit threshold, the current backoff recovery window is terminated, and the system is reset to the link protection state.

[0011] Further, the step of performing network topology rollback after the backoff recovery window expires includes: extracting the baseline network state information saved before entering the link protection state, and calculating the comprehensive link utility of the baseline network state information; the comprehensive link utility is obtained by weighting the statistically normalized values ​​of the underlying physical channel quality parameters, one-way delay, acknowledgment success rate, and topology switching cost; when the comprehensive link utility of the baseline network state information is greater than the sum of the comprehensive link utility and utility hysteresis margin of the current temporary topology state, a rollback control command is triggered; the sub-item rollback operation is performed in the order of first restoring the receiving beam number and operating frequency, then restoring the target access node, and finally restoring the queue scheduling parameters; and the temporary network state records generated during the blocking period are cleared after the rollback result is confirmed to be stable.

[0012] A second aspect of the present invention provides a remote communication system for an unmanned vehicle, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the remote communication method for the unmanned vehicle described in the first aspect of the present invention.

[0013] This invention provides a remote communication method and system for unmanned vehicles. It has the following beneficial effects: 1. This invention performs cross-layer bilateral determination by calculating a dynamic occlusion index and combining it with the temporal fading change rate of the signal-to-noise ratio measurement. This combines the spatial occlusion risk assessed by the application layer with the actual channel fading status fed back by the physical layer. This technical feature effectively overcomes the response lag problem caused by relying solely on the feedback of lower-level indicators, while filtering out false alarms caused by sensor noise or differences in medium properties, improving the accuracy of link physical blockage prediction, and thus avoiding invalid congestion in the gateway buffer.

[0014] 2. When a communication link is deemed to be at risk of being blocked, this invention performs a flow-cutting operation on regular service flows through a dynamic truncation factor, and downgrades the transmission protocol level for core control commands, as well as employing a graded redundancy transmission strategy based on the level of obstruction risk. This technical feature can proactively control the occupancy of regular data packets on the wireless-side buffer during periods of deteriorating channel quality, utilizing the released network resources to prioritize low-latency and high-reliability transmission of critical control messages, thus maintaining the operational safety of unmanned vehicles in complex communication environments.

[0015] 3. This invention introduces a backoff recovery window when the link shows initial signs of recovery. During the window's duration, progressive service recovery is performed based on the recovery openness coefficient. After confirming stable recovery, a partial network topology rollback is performed based on the overall link utility. This technical feature effectively distinguishes between short-term link recovery and stable recovery states, preventing repeated switching of access paths and service scheduling strategies due to frequent signal fluctuations when vehicles pass through obstructed edge areas. This reduces control plane reconstruction overhead and the probability of network state oscillations. Attached Figure Description

[0016] Figure 1 This is a diagram illustrating the architecture of a communication scheduling system according to an embodiment of the present invention. Figure 2 This is a flowchart of the communication scheduling method of the present invention; Figure 3 This is a schematic diagram illustrating the communication line-of-sight ray reconstruction principle of an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the principle of spatial interference envelope construction and point cloud screening in an embodiment of the present invention. Figure 5 This is a schematic diagram illustrating the principle of cross-domain mapping of electromagnetic properties of multi-dimensional point clouds in an embodiment of the present invention. Figure 6 This is a schematic diagram illustrating the principle of spatial integral calculation of the dynamic occlusion index in an embodiment of the present invention. Figure 7 This is a flowchart illustrating the cross-layer bilateral determination and regular business flow interception process according to an embodiment of the present invention; Figure 8 This is a flowchart illustrating the core control command protocol pass-through and redundancy in an embodiment of the present invention. Figure 9This is a flowchart illustrating the backoff recovery window and network topology rollback process in an embodiment of the present invention.

[0017] Among them, 10 is the communication baseband unit; 20 is the sensing unit; and 30 is the scheduling gateway unit. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] See attached document Figure 1 , Figure 1 This is a communication scheduling system architecture diagram according to an embodiment of the present invention. The present invention provides a remote communication system for unmanned vehicles, wherein the predictive communication scheduling system is deployed in the computing platform inside the vehicle.

[0020] The predictive communication scheduling system includes a communication baseband unit 10, which is used to transmit and receive radio frequency signals from external base stations. The communication baseband unit 10 parses the downlink reference signal sent by the base station, calculates the main diameter angle of arrival of the radio frequency signal using the phase difference of the receiving antenna array, and outputs the main diameter angle of arrival parameter and physical layer reference signal reception quality parameter to the outside.

[0021] The predictive communication and dispatch system includes a sensing unit 20, which contains a lidar device. The sensing unit 20 continuously outputs multi-dimensional point cloud data in the local coordinate system of the vehicle. The multi-dimensional point cloud data includes the spatial relative coordinates of each discrete point in the environment and the corresponding laser reflection intensity value.

[0022] The predictive communication scheduling system also includes a scheduling gateway unit 30, which is connected to the communication baseband unit 10 and the sensing unit 20 via an on-board wired network bus. The scheduling gateway unit 30 acquires the parameters output by the communication baseband unit 10 and the multi-dimensional point cloud data output by the sensing unit 20.

[0023] The scheduling gateway unit 30 contains a spatial calculation program, a material mapping data table, and a flow control script. The scheduling gateway unit 30 takes over the vehicle's application layer transmission protocol stack, controlling the transmission and reception of network data streams and queue allocation.

[0024] See attached document Figure 2 , Figure 2 This is a flowchart of a communication scheduling method according to an embodiment of the present invention. The present invention provides a remote communication method for an unmanned vehicle, comprising the following steps: S10, the scheduling gateway unit 30 extracts the principal diameter angle of arrival parameter output by the communication baseband unit 10, and reconstructs the communication line-of-sight ray in the vehicle local coordinate system with the static installation coordinates of the roof communication antenna as the starting point and in the direction corresponding to the principal diameter angle of arrival. S20, the scheduling gateway unit 30 extracts the multi-dimensional point cloud data output by the sensing unit 20, constructs a cylindrical geometric spatial interference envelope with the communication line-of-sight ray as the central axis, calculates the orthogonal distance from each discrete point in the multi-dimensional point cloud data to the communication line-of-sight ray, and extracts the discrete points located inside the spatial interference envelope as an effective interference point cloud subset. S30, the scheduling gateway unit 30 extracts the laser reflection intensity values ​​of each discrete point in the effective interference point cloud subset, inputs the laser reflection intensity values ​​into the preset function model, and obtains the corresponding electromagnetic attenuation coefficient values. The electromagnetic attenuation coefficient values ​​are used to characterize the amount of microwave signal penetration loss by obstacles at the corresponding spatial location. S40, the scheduling gateway unit 30 obtains the spatial relative projection distance of each discrete point in the effective interference point cloud subset along the communication line-of-sight ray dimension, introduces an exponential decay penalty function based on the spatial relative projection distance, performs spatial discrete integral calculation in combination with the corresponding electromagnetic attenuation coefficient value, and outputs the dynamic occlusion index under the current time period. S50, the scheduling gateway unit 30 compares the dynamic blocking index with the preset security blocking threshold, and synchronously compares the physical layer reference signal reception quality parameters output by the communication baseband unit 10 with the corresponding quality degradation threshold. Based on the comparison results, it redirects the regular service data under the transmission control protocol to the suspended ring buffer and performs network transmission port truncation. S60, the scheduling gateway unit 30 switches the vehicle core business data from the transmission control protocol to the user datagram protocol, and transmits the core business data by overlaying a forward error correction coding stream at the data link layer; S70, the scheduling gateway unit 30 continuously compares the dynamic occlusion index with the safety blocking threshold. After determining that the dynamic occlusion index value has fallen back to within the safety blocking threshold and meets the set duration requirement, the suspension state of the ring buffer is released and the queued data is sent in sequence to restore the initial network protocol transmission configuration of the system.

[0025] See attached document Figure 3 , Figure 3 This is a schematic diagram illustrating the principle of reconstructing a communication line-of-sight ray according to an embodiment of the present invention. In this embodiment, the scheduling gateway unit 30 extracts the principal diameter angle of arrival parameter output by the communication baseband unit 10, and reconstructs the communication line-of-sight ray in the vehicle local coordinate system, starting from the static installation coordinates of the roof-mounted communication antenna, in the direction corresponding to the principal diameter angle of arrival. The specific implementation of this process includes the following steps: S101, the communication baseband unit 10 receives the downlink reference signal sent by the external base station, obtains the signal phase difference between each element of the multi-antenna array, and then calculates and outputs the principal path angle of arrival based on the signal phase difference. As a preferred embodiment, the downlink reference signal includes a synchronization signal block or a channel state information reference signal. The communication baseband unit 10 parses the downlink reference signal, obtains the azimuth and elevation angles of the principal path signal arriving at the receiving antenna array, and combines the azimuth and elevation angles to form the principal path angle of arrival parameter. For the specific algorithm principle of calculating the signal angle of arrival using the antenna array phase difference, those skilled in the art can use a multiple signal classification algorithm or a rotation-invariant subspace algorithm for calculation. The signal phase extraction and angle estimation calculation process are well-known technologies in the field and will not be elaborated here. Subsequently, the scheduling gateway unit 30 reads the azimuth and elevation angle data output by the communication baseband unit 10 through the vehicle-mounted wired network bus.

[0026] S102, to align the physical space with the computational model, the scheduling gateway unit 30 establishes a vehicle local coordinate system, extracts the three-dimensional coordinates of the roof-mounted communication antenna within this system, and converts the azimuth and elevation angles into unit direction vectors in the vehicle local coordinate system. Specifically, the vehicle local coordinate system is established as a three-dimensional Cartesian coordinate system with the center of the vehicle chassis or the center of the rear axle as its origin. The scheduling gateway unit 30 reads the pre-calibrated and stored static installation coordinate vector of the roof-mounted communication antenna. The aforementioned static installation coordinate vector can be obtained through the external total station calibration at the vehicle's factory and written into the gateway configuration file. Its specific expression is: ; in, This represents the coordinate value of the communication antenna on the horizontal axis of the vehicle's local coordinate system. This represents the coordinate value of the communication antenna on the vertical axis of the vehicle's local coordinate system. This represents the coordinate value of the communication antenna on the vertical axis of the vehicle's local coordinate system. This indicates the matrix transpose.

[0027] After obtaining the static coordinates, the scheduling gateway unit 30 uses the extracted principal diameter signal azimuth angle... and pitch angle By calculating through spatial triangulation projection, a unit direction vector pointing towards the signal transmission source of the external base station is generated. The calculation formula is as follows: ; Unit direction vector The direction of linear physical propagation of microwave signals within the local space of the vehicle was determined.

[0028] S103, after obtaining the starting point and direction, the scheduling gateway unit 30 constructs a mathematical equation for the communication line-of-sight ray, using the static installation coordinate vector as the starting point and the unit direction vector as the extension direction. Considering the straight-line propagation characteristics of microwave signals and the detection boundaries of the vehicle-mounted hardware, in order to accurately locate obstacles that may interfere with the signal in subsequent processing, the scheduling gateway unit 30 reconstructs a geometric ray model representing the main propagation path of the signal in reverse within the local space of the vehicle. (Communication line-of-sight ray) The spatial parametric equation is expressed as: ; in, The scalar length parameter is defined along the direction of the line-of-sight ray, and its value range is set to... . This is the set maximum effective detection range limit. In this embodiment, The value of is determined by the effective ranging range of the lidar hardware in the sensing unit 20, typically between 100 and 200 meters. By limiting the range of the scalar length parameter, the scheduling gateway unit 30 only performs truncated calculations on the physical environment interference within a limited spatial range around the vehicle. This helps reduce the ineffective computational power consumption when the system calculates unknown spatial areas. The reconstructed communication line-of-sight ray provides a spatial alignment reference for subsequent matching of 3D point cloud data.

[0029] See attached document Figure 4 , Figure 4 This is a schematic diagram illustrating the construction and point cloud selection principle of a spatial interference envelope according to an embodiment of the present invention. In this embodiment, the scheduling gateway unit 30 extracts the multidimensional point cloud data output by the sensing unit 20, constructs a cylindrical geometric spatial interference envelope with the communication line-of-sight ray as the central axis, and calculates the orthogonal distance from each discrete point in the multidimensional point cloud data to the communication line-of-sight ray, thereby extracting the discrete points located inside the spatial interference envelope as an effective subset of the interference point cloud. The specific implementation of this process includes the following steps: S201, the scheduling gateway unit 30 acquires the multi-dimensional point cloud data scanned and output by the sensing unit 20 in real time through the vehicle-mounted wired network bus. The multi-dimensional point cloud data set contains multiple discrete environmental reflection points, each carrying three-dimensional spatial coordinates in the vehicle's local coordinate system and the corresponding laser reflection intensity value. For the coordinate system transformation and noise reduction preprocessing of the original point cloud of the lidar, those skilled in the art can use voxel filtering or statistical filtering algorithms, and the preprocessing process is a well-known technology in the field, which will not be described in detail here. Considering that microwave radio frequency signals do not propagate along an absolutely single-dimensional extremely thin straight line in real physical space, their main propagation energy is dispersed and covered within an ellipsoidal or cylindrical volume with the line-of-sight ray as the central axis. Therefore, in order to accurately define the core physical interference region in the microwave signal propagation process, the scheduling gateway unit 30 extracts the aforementioned reconstructed communication line-of-sight ray as the central principal axis and constructs a cylindrical geometric spatial interference envelope with a confined boundary.

[0030] As a preferred approach, the envelope radius of the spatial interference envelope... The maximum radius of the first Fresnel zone can be calculated based on the current operating carrier frequency band of the microwave communication system, or set as a fixed value between 0.5 meters and 2.0 meters according to the tolerance requirements of actual engineering applications. This envelope physically defines the three-dimensional spatial range that may substantially obstruct the communication link.

[0031] S202, after establishing the cylindrical spatial interference envelope, the scheduling gateway unit 30 traverses each discrete point in the multidimensional point cloud data set, calculating the orthogonal distance from these spatial points to the line containing the communication line-of-sight ray. Specifically, let any discrete point in the multidimensional point cloud data be... Its three-dimensional coordinate vector in the vehicle's local coordinate system is represented as: To calculate the relative spatial position, the scheduling gateway unit 30 combines the coordinate vector of the starting point of the communication line-of-sight ray generated above. Calculate discrete points Relative position vector relative to the physical installation center of the antenna The calculation formula is as follows: ; Combined with the unit direction vector of the communication line-of-sight ray The scheduling gateway unit 30 uses vector cross product operations to solve for discrete points. Orthogonal distance to the line-of-sight ray This orthogonal distance characterizes the vertical physical deviation of spatially scattered points from the central axis of signal propagation, and the corresponding calculation formula is: ; in, This represents the triple cross product operation of vectors. This represents the L2 norm of the oriented vector, i.e., the spatial magnitude of the resulting vector. Since... Having been normalized to a unit vector, the absolute value of the direct cross product calculated above represents the exact perpendicular distance from the point in space to the line.

[0032] S203, based on the calculated orthogonal distance parameters, the scheduling gateway unit 30 executes spatial location filtering logic to remove peripheral, non-associative point cloud data. The system introduces the envelope radius of the spatial interference envelope. and the maximum effective detection range limit of line-of-sight rays As a joint boundary constraint, to ensure that the extracted discrete points are not only close to the ray in radial distance but also within the effective detection range in axial depth, the scheduling gateway unit 30 calculates the relative position vector. unit direction vector Projected length The specific formula for calculating the projection length is as follows: ; Here, · represents the inner product operation of vectors. Since the direction vector and detection range are already known in the preceding sequence, the scheduling gateway unit 30 synchronously determines the orthogonal distance. With projection length Does the following set interference zone subordination conditions simultaneously meet: ; ; When a discrete point simultaneously satisfies the aforementioned dual constraints of radial distance and axial distance, the scheduling gateway unit 30 will allocate the discrete point... These are marked as valid retention points. After a full traversal and comparison, all retained points that meet the above spatial boundary conditions are integrated to form a subset of the valid interference point cloud. In the specific implementation of the algorithm, if no discrete points meeting the conditions are extracted after the traversal and comparison, the subset of the valid interference point cloud is recorded as an empty set to represent a line-of-sight state without physical obstacles within the current line-of-sight propagation area. This filtering operation can effectively prevent road surface reflection point clouds and distant unrelated building point clouds that do not affect microwave line-of-sight propagation from being mixed into the subsequent calculation process, thereby reducing the data throughput and system memory usage in the subsequent electromagnetic property mapping calculation process.

[0033] See attached document Figure 5 , Figure 5This is a schematic diagram of the principle of cross-domain mapping of electromagnetic properties of multi-dimensional point clouds according to an embodiment of the present invention. In this embodiment, the scheduling gateway unit 30 receives the effective interference point cloud subset obtained in step S20, extracts the laser reflection intensity data attached to each discrete point, and combines it with a pre-established calibration mapping relationship to convert the optical echo characteristics into electromagnetic attenuation parameters used to characterize the degree of microwave propagation obstruction. It should be noted that the laser point cloud reflects the spatial distribution and surface echo characteristics of obstacles in the near-infrared band, while the communication link focuses on the additional loss caused to microwave propagation by the same spatial location. Although they are in different bands, they are both affected by the material of the obstacle, surface condition, density of the medium, and water content. Therefore, a statistical correspondence can be established through offline calibration. During the online operation phase, the scheduling gateway unit 30 assigns electromagnetic properties to each point of the effective point cloud according to this correspondence, so that the geometric point set output in step S20 is converted into an attribute point set that can participate in wireless propagation analysis. The specific implementation of this process includes the following steps: S301, the scheduling gateway unit 30 reads each discrete point in the effective interference point cloud subset. Let the effective interference point cloud subset be denoted as... , of which The discrete points are denoted as... Its three-dimensional coordinate vector is denoted as The corresponding original laser reflection intensity is denoted as In this embodiment, The intensity field in the point cloud data output by the sensing unit 20 is used to characterize the amplitude of the laser echo after sampling and quantization by the receiving circuit. If the sensing unit 20 outputs an integer intensity value, the scheduling gateway unit 30 reads it according to the original quantized value; if the sensing unit 20 outputs a floating-point normalized intensity value, it reads it directly according to the device protocol. For different models of lidar, the underlying sampling bit width and quantization encoding method can be adapted by those skilled in the art according to the device communication protocol. The interface parsing process is a well-known technology in the field and will not be described in detail here.

[0034] Considering the possibility of missing, null, or outlier values ​​in the intensity field, the scheduling gateway unit 30 prioritizes performing validity checks on the read results. When a discrete point lacks an intensity field, or the intensity field is non-numerical, the discrete point can be marked as invalid and excluded from subsequent mapping calculations. In another possible implementation, a preset background attenuation coefficient can be assigned to the discrete point to maintain the integrity of the data structure. The purpose of this processing is to avoid abnormal inputs from disturbing subsequent mapping results.

[0035] Besides being affected by the surface material of obstacles, laser reflection intensity is also influenced by ranging distance, incident angle, transmit power drift, and receiver gain settings. Therefore, in a simplified implementation, under conditions of fixed installation orientation, distance compensation already completed by sensing unit 20, and minimal change in incident angle, laser reflection intensity can be directly used as the primary input for mapping. In implementations requiring higher mapping accuracy, the electromagnetic attenuation parameter can be determined by combining the ranging value or the incident angle correction. As a preferred approach, the scheduling gateway unit 30 first performs range truncation processing on the original laser reflection intensity. Let the minimum effective intensity value in the device protocol corresponding to sensing unit 20 be... The maximum effective strength value is The strength value after truncation is denoted as The calculation formula is as follows: ; in, For the first The truncated strength value at each discrete point This is the minimum effective strength boundary that the device can recognize. The maximum effective strength boundary that the device can identify. This indicates taking the smaller value. This indicates taking the larger value. and The settings can be directly based on the datasheet of sensing unit 20, or determined according to the quantile values ​​of the intensity distribution in the calibration sample, such as the 1st quantile and the 99th quantile. Using quantile values ​​can reduce the impact of a small number of abnormal high-brightness points or weak echo points on mapping stability. If the output of sensing unit 20 is an 8-bit integer intensity value, It can be 0. The value can be 255; if the output is 12-bit, 16-bit or other quantization formats, it can be set according to the corresponding quantization range.

[0036] S302, the scheduling gateway unit 30 performs normalization processing on the intensity values ​​participating in the mapping, forming a dimensionless reflection feature quantity. Let the normalized reflection feature quantity be... The intensity value participating in the normalization is denoted as The calculation formula is as follows: ; in, The value range is [0,1]. This represents the intensity value entering the normalization operation. When When the value is close to 0, it indicates that the echo reflection capability of the corresponding point is weak; when... A value close to 1 indicates a strong echo reflection capability at the corresponding point. After normalization, the intensity data from different equipment quantization intervals are converted into a unified dimension, facilitating processing using the same mapping relationship.

[0037] In some application scenarios, when obstacles are far from the sensing unit 20, the intensity of the laser echo will decrease due to propagation loss. If the sensing unit 20 does not perform distance compensation in the underlying driver, the scheduling gateway unit 30 can perform distance correction on the intensity data before normalization. Let the installation coordinates of the optical center of the sensing unit 20 in the vehicle's local coordinate system be... , No. The ranging value of each discrete point relative to the lidar is The calculation formula is as follows: ; in, Let L2 denote the L2 norm of a vector. Let the reference distance be . When the output intensity of sensing unit 20 approximately satisfies the squared attenuation relationship with the propagation distance, the intensity value after distance correction... It can be approximated as: ; in, This is the intensity value after distance correction. Units and Consistent, usually in meters. As a preferred method, A distance of 5m or 10m can be chosen, or an intermediate distance point from the calibration experiment can be selected to reduce the correction error at both the near and far distances. If the intensity output by the sensing unit 20 has already been compensated for by the underlying driver, then let... If distance correction is required, the corrected intensity value should be calculated according to the above squared attenuation formula. To avoid abnormal magnification at close range or overcompensation at long range, [the following can be done]: The upper limit truncation is executed again, and its processing method is the same as... The same applies. The distance correction described above is an optional step, and whether it is enabled is determined by the output definition of the sensing unit 20.

[0038] S303, after obtaining the normalized reflection characteristic, the scheduling gateway unit 30 calls the pre-stored electromagnetic attribute mapping relationship to... Converted to electromagnetic attenuation coefficient In this embodiment, Used to characterize the The attenuation capability per unit length of the microwave signal in the working frequency band corresponding to the local spatial location of each discrete point can be expressed in dB / m. The larger the value, the stronger the obstruction effect of the obstacle at that location on microwave propagation. It should be noted that this parameter is not used to determine the precise material type of the obstacle point by point, but rather to characterize the statistical obstruction capability of obstacles on line-of-sight propagation in a vehicle operation scenario. It can be subsequently combined with spatial weights, path length, or equivalent thickness for cumulative calculation.

[0039] Considering that the relationship between near-infrared reflection intensity and microwave attenuation is typically not simple linear for different materials, the scheduling gateway unit 30 preferably uses a nonlinear mapping function for conversion. As a preferred method, an S-shaped mapping function is used, with the calculation formula as follows: ; in, This is the boundary of the lowest attenuation coefficient. This represents the boundary of the highest attenuation coefficient. The slope parameter of the mapping curve. This represents the inflection point of the mapping curve. This represents the natural exponential function. and Units and Same. Parameters , , and All parameters are determined by the offline calibration process and stored in the parameter table of the scheduling gateway unit 30.

[0040] When the calibration data exhibits a monotonic trend, the aforementioned S-shaped mapping function is easy to implement and perform online calculations. This mapping method is chosen because common obstacles such as glass, metal, concrete, and vegetation may fall within similar ranges in laser intensity, but typically exhibit significant differences in microwave penetration loss. Direct linear conversion would easily compress the distinction between high-obstruction and low-obstruction materials; however, the S-shaped mapping limits the output results to the boundary range determined by calibration, thus improving parameter stability.

[0041] As another possible implementation, the scheduling gateway unit 30 can also use a segmented lookup table method to complete the mapping. Assume the system pre-stores... Each intensity segment boundary and the corresponding attenuation coefficient ,in, The values ​​are positive integers, and the boundaries of each segment are sorted in ascending order of normalization strength. When When falling within adjacent segmented intervals, linear interpolation is used to obtain... .when When a sample happens to be located on a segment boundary, it can be assigned to any adjacent interval according to preset rules, as long as the system implementation remains consistent. The segment interval can be set equally or based on the distribution density of the calibration samples in different intensity intervals. When the calibration samples exhibit a non-monotonic distribution in different material intervals, or when local dispersion is strong, the segmented lookup table method is preferred. Both the segmented lookup table method and the S-shaped mapping function are specific implementation methods of the "mapping relationship".

[0042] S304, the scheduling gateway unit 30 generates a set of valid interference points marked with electromagnetic properties. Let the result set after mapping be denoted as... Its expression is: ; in, Each element in the table must include at least the spatial coordinates of a discrete point. Normalized reflection characteristic and electromagnetic attenuation coefficient After this processing, the scheduling gateway unit 30 no longer treats the point cloud as merely a geometric point set, but instead uses it as an attribute point set that simultaneously carries spatial location and electromagnetic barrier information for subsequent calculations.

[0043] If the effective interference point cloud subset output in step S20 is an empty set, the scheduling gateway unit 30 will... This is denoted as an empty set, and the point-by-point mapping result at the current moment is marked as an unobstructed state. If the effective interference point cloud subset contains discrete points, but the intensity fields of all discrete points are invalid, then the scheduling gateway unit 30 can also... This is recorded as an empty set, or an attribute point set is generated according to a preset background attenuation coefficient. This processing avoids algorithm interruption in subsequent steps due to empty or abnormal input, and the corresponding dynamic occlusion index can enter the subsequent calculation process with a value of zero or the minimum background value set by the system.

[0044] S305, the electromagnetic attribute mapping relationship is established using an offline calibration method and loaded by the scheduling gateway unit 30 before operation. To ensure the reproducibility of the technical solution, the calibration process can be implemented as follows: Select several typical obstacle samples, with sample materials including at least metal plates, ordinary glass, laminated glass, concrete walls, asphalt pavements, tree branches and leaves, and vehicle exterior panels; under the same or comparable installation posture, use the sensing unit 20 to collect the laser reflection intensity data of each sample, and simultaneously measure the additional attenuation value of the corresponding sample to the microwave signal in the system's operating frequency band. As a preferred method, each type of sample is sampled separately under multiple ranging positions and multiple incident angles to reduce the influence of a single posture on the calibration results. For samples with different thicknesses, the measured additional attenuation value can be converted into an attenuation coefficient per unit length before being used in the fitting. The additional attenuation value can be obtained through a vector network analyzer, a channel detector, or a joint measurement by the base station and the vehicle. For the specific instrument connection method for microwave additional attenuation measurement, those skilled in the art can configure it according to the operating frequency band and test environment; the measurement process is well-known in the field and will not be described in detail here.

[0045] In one possible implementation, let the first... The normalized reflectance characteristic of each calibrated sample is The measured electromagnetic attenuation coefficient is The total number of samples is ,in, If the integer is positive, then the above mapping parameters can be fitted using the least squares criterion, and the objective function is: ; in, The objective function for parameter fitting is defined. The scheduling gateway unit 30 itself is not responsible for online training; the parameter file after fitting is distributed and stored in the form of a configuration table. This approach reduces the on-board online computational burden and facilitates the creation of parameter sets for different operating frequency bands or different models of sensing units 20.

[0046] To facilitate project implementation, the parameter determination method can be further explained as follows: and It can be determined by the minimum and maximum measured attenuation coefficients in the calibrated sample, or by taking its statistical quantile interval, such as the 5th percentile and the 95th percentile, to reduce the influence of outlier samples; The rate of change of the mapping curve is controlled by taking a positive value, preferably between 2 and 15, and can be determined based on the principle of minimizing the fitting error; This indicates the transition point where the intensity-to-attenuation relationship is more sensitive, with a value range of [0,1], which can be determined from the vicinity of the sample distribution center. If the system operates in the millimeter-wave band, the difference in transmission loss between different materials is usually more pronounced. A larger value can be selected; if the system operates in the Sub-6GHz frequency band, the parameter range can be reduced accordingly. When using the segmented lookup table method, the boundaries of each segment can be set at equal intervals or adaptively divided according to the calibration sample density. Through the above parameter determination method based on offline calibration and curve fitting, the system can stably map optical echo characteristics to the expected electromagnetic penetration loss during online operation.

[0047] Through the above processing, the scheduling gateway unit 30 converts the optical point cloud data output by the sensing unit 20 into a set of electromagnetic parameters that can quantify the degree of wireless propagation obstruction. This result can be directly used as the input for subsequent spatial weighted calculations.

[0048] See attached document Figure 6 , Figure 6 This is a schematic diagram illustrating the principle of dynamic occlusion exponential spatial integration calculation according to an embodiment of the present invention. In this embodiment, the scheduling gateway unit 30 receives the set of effective interference points with electromagnetic attribute markers output in step S30. The system performs a spatially weighted integral on each discrete point in the point set, thereby converting the discrete attribute data into a quantitative indicator characterizing the current line-of-sight occlusion level. The above integral calculation is based on the fact that microwave signals propagate in space not as a straight ray, but rather their energy is non-uniformly distributed along the cross-section. Correspondingly, the attenuation effect of obstacles located within the interference envelope on the link is not equivalent; obstacles closer to the central axis of the line-of-sight ray contribute more significantly to occlusion, while those closer to the edge of the envelope have a relatively weaker blocking effect. Based on this phenomenon, this embodiment uses a spatially based weighted summation method as an effective approximation of continuous spatial integrals in discrete point cloud scenarios. The specific implementation of this process includes the following steps: S401, the scheduling gateway unit 30 is the effective set of interference points. Each discrete point in Calculate a spatial weighting coefficient This weighting coefficient is used to quantify the spatial contribution of a single discrete point within the overall occlusion assessment system. As a preferred approach, the spatial weighting coefficient... It is primarily determined by the orthogonal distance from the discrete point to the communication line-of-sight ray. Specifically, to simulate the physical characteristic of electromagnetic energy attenuating from the center to the edge, a Gaussian function is used to define this weight, and its calculation formula is as follows: ; in, For the first Spatial weighting coefficients for discrete points; The orthogonal distance from the discrete point to the communication line-of-sight ray has been calculated in step S20; The scaling parameter of the Gaussian function is used to control how quickly the weights decay with distance. This represents the natural exponential function. To ensure the weight distribution matches the communication interference region, the scale parameter... The value of is the same as the radius of the spatial interference envelope defined in step S20. Related. As one possible implementation method, Can be set to A fixed ratio, for example ,in This is a proportionality coefficient, preferably between 0.3 and 0.7. At that time, weight Taking the maximum value of 1 indicates that the point located on the central axis of the ray contributes the most; as Increase, weight Smooth descent.

[0049] As another preferred implementation, spatial weighting coefficient Furthermore, the laser reflection intensity information at discrete points can be comprehensively considered. Given that dense and impermeable obstacle surfaces often correspond to high optical reflection characteristics, the system can superimpose reflection feature weights on the spatial distance allocation. In this approach, the weighting coefficients... The calculation formula can be updated as follows: ; in, The normalized reflection characteristic quantity calculated in step S30; The intensity-weighted index is an adjustable hyperparameter, preferably ranging from 0.5 to 2.0. Points with stronger reflectivity will be assigned greater weight. Both of the above weight calculation methods are specific implementations of spatial weight coefficients in this application.

[0050] S402, the scheduling gateway unit 30 will set the electromagnetic attenuation coefficient of each discrete point. Its corresponding spatial weight coefficient By combining and introducing local spatial differential elements for discretization and summation, the original occlusion index is obtained. It should be noted that the distribution density of LiDAR point clouds typically decreases significantly with increasing distance. Directly summing the attributes of discrete points would lead to an abnormally amplified measurement result for densely distributed obstacle points at close range. Therefore, this embodiment introduces local volumetric element parameters that match the point cloud resolution. As a form of integral compensation, its calculation formula is as follows: ; in, The original occlusion index; summation symbol. Represents the set of effective interfering points Iterate through and accumulate all discrete points; For the first The equivalent local volume element corresponds to each discrete point. As one possible implementation, the scheduling gateway unit 30 uses a voxel filtering algorithm to divide the three-dimensional grid during the preprocessing stage. This refers to the physical statistical volume of the voxel grid where the point is located; as another simplified implementation, Approximate calculations can also be made based on the distance measurement value of the discrete point and the system angular resolution to balance the density differences caused by distance fluctuations. This weighted summation process is a specific implementation of the "spatial integration calculation" described in this invention. Its physical meaning is to integrate and aggregate the local attenuation capability of each discrete point to the microwave signal according to its spatial relative importance and the proportion of physical space it occupies, thereby obtaining a metric reflecting the overall obstruction of the link.

[0051] If the effective interference point set is output in step S30 An empty set indicates that no obstacle points fall within the current line-of-sight interference zone, and the scheduling gateway unit 30 directly sends the original occlusion index. Set to 0. This processing logic ensures that the algorithm can output results that match the actual situation in a see-through scenario, avoiding algorithm logic abnormalities or program interruptions caused by empty set input.

[0052] S403, Scheduling gateway unit 30 for original occlusion index Perform normalization processing to generate the final output dynamic occlusion index. .because The numerical range exhibited is not fixed, making it inconvenient to directly correlate with the threshold configuration of the underlying communication module. To establish a unified decision-making standard, it needs to be mapped to a standardized range.

[0053] As a preferred implementation, an S-shaped logic function is used. The normalization process is calculated using the following formula: ; in, The final output is the dynamic occlusion index, which ranges from (0,1). The slope parameter of the normalized curve controls the sensitivity of the mapping to moderate occlusion changes; This is the center offset parameter, corresponding to the original occlusion index value when the normalized output is 0.5. The specific method for determining these parameters is as follows: It is preferable to set the value between 3 and 10 to ensure that the curve has a good gradient in the transition region; Then, the average value at which the link signal-to-noise ratio begins to deteriorate significantly can be used as the basis for the offline calibration phase. The value is calibrated and written. As another possible implementation method, normalization can also be performed by linear piecewise mapping or maximum value truncation, which are all specific implementations of normalization processing in this application.

[0054] The final output dynamic occlusion index It is a dimensionless scalar value that continuously reflects the risk of obstruction on the communication line-of-sight ray at the current moment. When When the value is close to 0, it indicates that the communication link is unobstructed or that there is only a slight obstruction; when... When the value approaches 1, it indicates severe obstruction of the communication link, which may be accompanied by a sharp decline in signal quality. After this index is generated, the scheduling gateway unit 30 completes the information conversion from the original physical perception of the environment to the dimension of wireless link obstruction, and uses it as the core input basis for subsequent control of the communication terminal to perform handover or beam adjustment.

[0055] See attached document Figure 7 , Figure 7 This is a flowchart of cross-layer bilateral determination and regular service interception according to an embodiment of the present invention. In this embodiment, after obtaining the dynamic obstruction index output in step S40, the scheduling gateway unit 30 makes a joint decision based on the underlying physical channel status fed back by the communication baseband unit 10. When it is determined that the current communication link faces a physical blockage risk, the active transmission of non-priority services is restricted. Traditional network scheduling strategies usually rely on the feedback of underlying channel indicators, which is essentially a delayed response and is prone to brief buffer congestion when a physical link blockage occurs. While using environmental perception for prediction can provide forward-looking indicators, it can sometimes produce false alarms due to sensor noise or differences in medium properties. By cross-validating the environmental spatial dimension of the application layer with the electromagnetic transmission dimension of the physical layer, it helps to reduce the ping-pong effect while ensuring secure flow transmission. The specific implementation includes the following steps: S501, the scheduling gateway unit 30 reads the dynamic occlusion index of the current sampling period. Simultaneously, the underlying channel quality parameters are obtained from the communication baseband unit 10. As a preferred method, the underlying channel quality parameters can be expressed as signal-to-noise ratio (SNR) measurements, denoted as... The measurement and extraction of the aforementioned signal-to-noise ratio (SNR) is typically performed by the baseband chip of the communication unit during the parsing of pilot or synchronization signals. Those skilled in the art can utilize this method according to specific wireless access standards (such as 5G NR or C-V2X protocols), which are well-known technologies in the field and will not be elaborated upon here. To capture the dynamic changing trend of channel quality, the scheduling gateway unit 30 further calculates the timing fading rate of the SNR. Let the current time step be... The previous step was The time step is Then the rate of change of signal-to-noise ratio The calculation formula is: ; Among them, time step The feedback period can be set based on the channel state information of the communication baseband unit 10, preferably within the range of 10 milliseconds to 50 milliseconds. This indicates that the current physical link quality is showing a downward trend in the time domain. This indicator, as a reference dimension, provides early status input for subsequent joint analysis.

[0056] S502, after acquiring the aforementioned environmental perception and physical channel parameters, the scheduling gateway unit 30 executes cross-layer bilateral decision-making logic. If the occlusion index calculated by the isolated point cloud is used, the system may trigger unnecessary speed reduction operations when encountering occlusion from suspended objects such as rain or snow, or non-metallic media; conversely, if only the underlying layer is considered... Due to variations, the system struggles to distinguish whether signal fluctuations originate from conventional multipath fading caused by vehicle movement or from encountering large physical obstacles. Therefore, this embodiment employs the following bilateral triggering discrimination algorithm: ; in, The result is either true (1) or false (0) as a trigger flag. Preset occlusion warning thresholds for the environmental perception dimension; The maximum tolerable fading rate threshold for the physical channel dimension; logical symbol This represents a logical AND operation. Regarding the method of determining the parameters, The tolerance for occlusion can be set between 0.6 and 0.85, depending on the system's tolerance. Setting it to a negative value can be calibrated by combining it with the statistical envelope lower limit of multipath fast fading in actual measurements, for example, setting it in the range of -3dB / s to -10dB / s. When the environmental perception model outputs an occlusion warning, and the underlying physical channel synchronously exhibits significant fading, That is, it is set to 1. This double confirmation mechanism can effectively filter out interference caused by a single dimension.

[0057] S503, for when the trigger condition is activated (i.e. In the event of this situation, the scheduling gateway unit 30 takes over the upper-layer data flow scheduling authority and performs routine business flow interception operations on the gateway queue. It should be noted that once... When set to 1, the system enters and locks into a link protection state. Even if subsequent sampling periods no longer meet the dynamic triggering conditions in S502, the system maintains this protection state until the release condition in step S504 is met. When a vehicle is about to enter an electromagnetic shielding area formed by a building, tunnel, or large truck, the available channel capacity often shrinks rapidly. If the injection of regular data is not restricted, the network card's transmission queue may quickly become full, causing high-priority commands with low latency requirements to fail due to queuing.

[0058] Specifically, the scheduling gateway unit 30 parses the header of the data packet to be sent and reads the relevant fields representing service priority. By extracting the Differentiated Services Code Point (DSCP) in the Internet Protocol header or the traffic category identifier supported by the lower-level protocol, the current service flow is divided into high-priority service flows and regular service flows. High-priority service flows are used to cover time-sensitive messages such as autonomous driving cooperative control commands and emergency collision avoidance warnings; regular service flows include messages such as in-vehicle infotainment video streams and background diagnostic data synchronization.

[0059] For identified regular service flows, the scheduling gateway unit 30 dynamically truncates them based on the severity of the obstruction. Perform proactive throttling. Remove throttling factors. The calculation formula is: ; in, This represents the rate-limiting ratio or drop probability of regular business data packets, and its value range is... ; The penalty constant is preferably... to real numbers; This indicates taking the smaller value. As the occlusion index... Towards rising, Approaching For intercepted regular service packets, the scheduling gateway unit 30 determines whether to redirect them to a suspended circular buffer or actively discard them, depending on the remaining capacity of its local buffer queue. This probabilistic discarding or buffer queuing approach allows limited bandwidth to be reserved for high-priority service flows during periods of poor communication conditions, maintaining the continuity of core control data communication. The aforementioned process of implementing traffic limiting using a truncation factor specifically supports the regular service throttling feature in this application.

[0060] S504, the scheduling gateway unit 30 continuously and periodically monitors the communication status and obstruction environment, and performs smooth link fallback processing when recovery conditions are met. It should be noted that when deep obstruction occurs, the signal-to-noise ratio may remain at an extremely low value and no longer decrease. Will to Convergence occurs, and the previous fading rate trigger condition is no longer met. To prevent the system from erroneously exiting the throttling state when in deep occlusion, this embodiment constructs a release logic based on absolute state preservation. Prerequisites for releasing the trigger flag. for: ; in, The hysteresis threshold for environmental occlusion recovery is set slightly below [value missing]. For example, it is advisable ; The signal-to-noise ratio threshold value used to indicate that communication quality has returned to normal. If and only if... When the result is true, it indicates that the vehicle has substantially left the obstructed area and the underlying link has been rebuilt. The scheduling gateway unit 30 initially determines that the conditions for restoring normal communication are met and triggers a subsequent backoff recovery window mechanism to perform gradual service recovery and network status assessment, replacing an immediate forced reset. The introduction of a closed-loop triggering assessment logic with hysteresis judgment and absolute state dependence can prevent environmental edge state machine oscillations and avoid the algorithm getting stuck in a state suspension. This cross-layer judgment and collaborative scheduling mechanism establishes a control path connecting environmental perception and network scheduling, which helps reduce the probability of transmission delay violations under complex road conditions.

[0061] See attached document Figure 8 , Figure 8 This is a flowchart illustrating the core control command protocol pass-through and redundancy according to an embodiment of the present invention. In this embodiment, when the scheduling gateway unit 30 confirms through the determination logic in step S50 that the communication link faces a risk of physical blockage (i.e. The system will execute enhanced transmission strategies for identified high-priority core control commands. This strategy proactively sacrifices some transmission efficiency to ensure the reliability of critical command transmission when link quality is expected to deteriorate, thus guaranteeing the real-time safe operation of the vehicle. The specific implementation of this process includes the following steps: S601, the scheduling gateway unit 30 determines the triggering condition. Upon verification, the system immediately identifies and intercepts the core control commands to be sent according to the business priority rules. Here, the core control commands specifically refer to the data streams marked as high priority in step S50. The scheduling gateway unit 30 then parses the data content of these commands, extracts their key information payloads, and reconstructs their encapsulation format to achieve network transmission protocol degradation and pass-through switching (hereinafter referred to as protocol pass-through).

[0062] As a preferred method, the specific operations of protocol pass-through include: Instruction simplification and encoding optimization: The scheduling gateway unit 30 parses the raw data payload of the core control instructions, removes unnecessary fields from the protocol header, and retains only the minimum set of information directly related to the control function. For example, for a cooperative driving instruction containing vehicle position, speed, acceleration, and desired trajectory, it can be converted from a flexible but redundant JSON format into a compact, predefined-length binary structure, thereby significantly reducing the data payload.

[0063] Lowering the protocol stack layer: The scheduling gateway unit 30 can directly encapsulate the simplified instruction payload into a lower protocol layer for transmission. For example, instructions originally transmitted via Transmission Control Protocol (TCP) can be switched to User Datagram Protocol (UDP) to eliminate mechanisms such as the three-way handshake; in a more extreme preferred approach, the transport layer and network layer headers can even be further stripped off and directly encapsulated into a custom Ethernet frame for transmission. This "protocol pass-through" aims to bypass TCP's three-way handshake, congestion control, and other mechanisms that may introduce additional latency, striving for faster transmission opportunities when link quality is uncertain.

[0064] S602, the scheduling gateway unit 30 implements a redundant transmission strategy for the core control commands after protocol pass-through. As a preferred dynamic strategy, the scheduling gateway unit 30 can adjust the transmission based on the dynamic occlusion index. Different redundancy mechanisms are activated based on the severity of the problem. For example: Information redundancy (medium risk zone): when At a moderate level (e.g., 0.6 to 0.8), the scheduling gateway unit 30 primarily enables information redundancy. Specifically, it can instruct the communication baseband unit 10 to apply a lower-order modulation and coding scheme (MCS) to the core control commands to be transmitted at the physical layer, i.e., to employ stronger forward error correction (FEC) coding. This increases the redundant bits in the data, helping the receiver correct more transmission errors when the signal-to-noise ratio decreases.

[0065] Time redundancy (high-risk area): when When the threshold is high (e.g., greater than 0.8), the scheduling gateway unit 30, in addition to enabling information redundancy, also implements a time redundancy strategy. Specifically, the system repeatedly transmits a simplified version of the same core control command multiple times within a short period. For example, for each command, it is transmitted 3 to 5 times consecutively at intervals of 1 to 5 milliseconds. Multiple transmissions can effectively combat sudden deep fading of the channel and increase the probability that at least one copy is successfully received.

[0066] Multi-path redundancy (extreme risk area): If the vehicle terminal integrates multiple communication interfaces (such as 5G NR and DSRC), when When the threshold approaches 1.0, indicating an impending complete link disruption, the scheduling gateway unit 30 can activate the highest level of multipath redundancy strategy. Specifically, it sends instruction copies simultaneously to the transmission queues of both the primary communication interface (e.g., 5GNR) and the backup communication interface (e.g., DSRC), attempting parallel transmission through physically independent channels.

[0067] The aforementioned tiered strategy increases the risk level progressively, ensuring reliability while also considering resource consumption, thus constituting a specific implementation of the redundant transmission described in this application.

[0068] S603, while executing the aforementioned redundant transmission strategy, the scheduling gateway unit 30 monitors the transmission status of the core control command. To avoid a logical deadlock due to infinite retransmissions after a complete link interruption, the scheduling gateway unit 30 synchronously starts a maximum retransmission counter or timer when redundant transmission begins. Since the underlying protocol has been switched to connectionless User Datagram Protocol or custom Ethernet frames, the scheduling gateway unit 30 includes the service layer sequence number in the simplified payload of the core control command and requests the receiver to return a lightweight application layer status acknowledgment packet (ACK). If no acknowledgment is received from the receiver based on the above mechanism within the maximum number of retransmissions (e.g., 5 times) or the preset maximum duration (e.g., 20 milliseconds), the scheduling gateway unit 30 will stop redundant transmission of the command and may report the transmission failure to the upper-layer application. Once the command is confirmed as successfully transmitted (e.g., receiving an ACK), or the communication link status returns to normal (determined by the release mechanism in step S504, triggering the release flag), the redundancy is resolved. If the condition is true and the link has completed a smooth rollback, the scheduling gateway unit 30 will immediately exit the enhanced transmission mode and resume the normal protocol encapsulation and non-redundant transmission strategy for subsequent instructions. This closed-loop control logic, which includes a forced exit condition, ensures the robustness of the special processing mechanism and helps to provide enhanced transmission guarantees for high-priority services in harsh communication environments.

[0069] See attached document Figure 9 , Figure 9This is a flowchart of the backoff recovery window and network topology rollback according to an embodiment of the present invention. In this embodiment, after completing the cross-layer bilateral determination in step S50 and the core control command enhancement transmission in step S60, the scheduling gateway unit 30 preferably does not immediately cancel the aforementioned protection action when the link index shows a short-term rebound, but instead introduces a controlled recovery observation mechanism. When a vehicle passes through the edge of a building, the rear of a large vehicle, or the area blocked by a curve, the link quality may fluctuate briefly. If a channel rebound is detected at any time and normal transmission is immediately resumed, redundancy is removed, and the original network connection is switched back, the service scheduling and access path may switch repeatedly, thereby causing queue fluctuations, latency changes, and repeated reconstruction of the control plane. Based on this, a backoff recovery window is set in the link recovery process to distinguish between the two different states of "the link is starting to improve" and "the link has been stably restored". After confirming that the recovery is stable, the communication link, forwarding path, or access relationship temporarily established during the blocking period is gradually rolled back to the normal topology. In this embodiment, the backoff recovery window refers to the duration during which the system maintains observation and bandwidth limiting recovery before lifting traffic cutoff, removing redundancy, or canceling temporary routes; network topology rollback refers to selectively restoring temporary access nodes, temporary relay paths, temporary forwarding entries, temporary beam configurations, or temporary frequency point configurations established during the blocking period to the target network connection state recorded before the blocking. The specific implementation includes the following steps: S701, when the scheduling gateway unit 30 enters the link protection state, it synchronously saves the baseline network state information before the blocking occurred, as a target reference for subsequent rollback. This baseline network state information can be denoted as... It includes at least the identifier of the current access node. Current valid forwarding paths Current transmit beam number Current working frequency and the current business queue scheduling parameter set .in, This indicates the roadside unit or base station identifier to which the communication baseband unit 10 was connected before the blockage occurred; This indicates the predetermined forwarding path from the scheduling gateway unit 30 to the target network exit; This indicates the transmit or receive beam number currently used by the communication baseband unit 10; Indicates the main operating frequency used; This indicates the bandwidth allocation weight, transmission gating ratio, and redundancy level parameters related to the service queue. As a preferred method, the scheduling gateway unit 30 triggers a flag in step S50. A state snapshot is performed when the value changes from 0 to 1, and this snapshot is not overwritten before the end of the current protection state, so as to ensure... Maintain the baseline state as it was before this round of blocking occurred. If the system subsequently transitions from a normal monitoring state to a new protection state, a new state can be generated. For forwarding table caching, adjacency table maintenance, and access node identifier reading, those skilled in the art can implement these functions using the existing software architecture of vehicle networking gateways. The specific implementations are well-known technologies in this field and will not be elaborated upon here.

[0070] S702, after detecting preliminary signs that the link has been unprotected, the scheduling gateway unit 30 does not directly perform state restoration, but first establishes a backoff recovery window. As a preferred method, the prerequisite for establishing the backoff recovery window is the use of a recovery decision quantity. The measurement is performed using the following formula: ; in, To restore the judgment quantity, its value range is preferably normalized to [value range]. The dynamic occlusion index is output in step S40; The current signal-to-noise ratio measurement value fed back by the communication baseband unit 10; and These are the lower and upper bounds used for signal-to-noise ratio normalization, respectively. The signal-to-noise ratio value can be taken from near the lowest reliable demodulation threshold of the corresponding communication standard. High-quality signal-to-noise ratio percentile values ​​obtained from long-term statistics under open road conditions can be used, such as the 90th percentile signal-to-noise ratio value; both are used as normalization reference boundaries, rather than real-time measurements. To prioritize the success rate of message acknowledgments within the most recent statistical window, it is preferable to prioritize acknowledgments based on the most recent data. The statistics of the core control messages were obtained. You can choose 5 to 20, or choose the most recent duration. The high-priority control messages within the message were statistically analyzed. The time can range from 50 milliseconds to 200 milliseconds. , and For the weighting coefficients, satisfying As a preferred method, A value of 0.3 to 0.5 is acceptable. A value of 0.3 to 0.5 is acceptable. A value between 0.1 and 0.3 is acceptable. This recovery criterion is chosen because the occlusion index reflects whether the spatial environment has been substantially opened up, the signal-to-noise ratio reflects whether the physical layer link has recovered, and the confirmation success rate reflects whether the service layer has achieved stable transmission capabilities. Using all three together helps reduce the probability of false recovery caused by short-term fluctuations in a single indicator.

[0071] when Persistently above the recovery entry threshold When the scheduling gateway unit 30 initiates the backoff recovery window, the initiation condition can be expressed as follows: ; in, To restore the window startup flag; To restore the entry threshold, a value of 0.6 to 0.8 is preferred; To continuously satisfy The number of sampling times; The minimum number of consecutive samples required to enter the recovery window is preferably between 3 and 10. The purpose of setting the number of consecutive samples is to filter out occasional rises within a single sampling period. If... The scheduling gateway unit 30 records the current time as And start timing the backoff to restore the window length. As a preferred method, It can be adaptively determined based on the current recovery stability, and its calculation formula is as follows: ; in, Duration of the backoff recovery window; For the minimum recovery observation time, it is preferably 50 to 200 milliseconds; This is the time scaling factor, preferably between 100 milliseconds and 500 milliseconds. When When the value is low but has reached the recovery threshold, the system can appropriately extend the observation period; when When the threshold is high, the system can quickly enter the subsequent recovery process. With this setting, the recovery logic is no longer triggered at a single point, but rather has a time-based stability assessment. In S703, during the backoff recovery window, the scheduling gateway unit 30 performs gradual service recovery instead of withdrawing all protection actions at once. In this embodiment, this can be achieved by implementing tiered backoff for the regular service throttling in step S50 and the core control redundancy in step S60. For the recovery of regular services, the scheduling gateway unit 30 defines a recovery open coefficient. The calculation formula is as follows: ; in, To restore the current sampling time within the window The corresponding reopening coefficient has a range of values. To avoid the start time of the recovery window; The duration of this recovery window. When, it indicates that the restrictions in the triggered protection state are still maintained; when When the window for recovery expires, normal service can resume. Correspondingly, the scheduling gateway unit 30 can adjust the actual threshold coefficient for regular services. Set to: ; in, This represents the percentage that is allowed to be sent during regular business operations, and its value range is [value range missing]. This is the regular business truncation factor calculated in step S50. Therefore, it can be seen that when the recovery window is first started... ;along with Gradually increase As the threshold approaches 1, the gating for routine business operations is gradually relaxed. This gradual recovery approach helps avoid a sudden influx of too much data into the queue after temporary relaxation, thereby reducing the probability of renewed congestion in the wireless buffer.

[0072] For redundant recovery of core control commands, the scheduling gateway unit 30 does not require the immediate cancellation of all duplicate transmissions and low bit rate protection at the start of the recovery window. Instead, it follows the recovery open coefficient. Gradually reduce the redundancy level. As a preferred approach, the number of copies sent can be... Defined as: ; in, For a moment The number of copies to be sent for a single core control command; This represents the number of replicas under normal communication conditions, preferably 1. The preferred number of copies in the protected state is 3 to 5; This indicates rounding up. While the recovery window is still open, the system retains some redundancy until the link stabilizes before reverting to normal single-copy transmission. Modulation-coding level callbacks, forward error correction redundancy reduction, and beam parameter gradual changes can be implemented by those skilled in the art according to specific communication protocols; the control methods are well-known in the field and will not be elaborated upon here.

[0073] S704, within the backoff recovery window, the scheduling gateway unit 30 continuously checks whether the recovery status has been disrupted again to prevent the link from deteriorating again before the recovery process is complete. Therefore, this embodiment introduces a recovery cancellation determination condition. Its expression is: ; in, To restore the flag indicating the revocation of the judgment; Indicates a logical OR operation; The threshold for determining occlusion degradation should preferably be higher than the recovery hysteresis threshold. And can be taken as close to or equal to the occlusion threshold used when entering the protection state. This creates a hysteresis band between recovery and reprotection; The lower limit threshold for signal-to-noise ratio can be determined by raising it by 1 to 3 dB based on the lowest reliable demodulation threshold. To confirm the lower limit threshold for the success rate. As a preferred method, You can click on the most recent Statistics on the confirmation status of core control messages, among which The optimal value is 5 to 20. If any condition worsens again, the scheduling gateway unit 30 immediately terminates the current recovery window and... The settings are reset to 0, restoring the service throttling and instruction redundancy configuration to the protected state. Furthermore, to prevent rollback from occurring before the recovery window expires at the edge of the protection state, a recovery maintenance threshold can be further set. and demanded that Still meet the requirements upon expiration Only after that does the subsequent topology rollback process begin. As a preferred approach, It is acceptable to be no less than The value. This backoff hold logic forms a reprotection loop during the recovery process, which helps reduce the likelihood of frequent oscillations in the occlusion edge region of the system.

[0074] S705, when the backoff recovery window has completely ended and no recovery cancellation judgment condition has been triggered during the window period, the scheduling gateway unit 30 begins to execute network topology rollback. In this embodiment, the network topology includes access node selection relationships, heterogeneous network interface connection relationships (such as cellular and direct communication interfaces), beam pointing relationships, and operating frequency point relationships; the network connection state temporarily adjusted during the blocking period to maintain communication can all be used as rollback objects. To ensure the operability of the rollback operation, the scheduling gateway unit 30 records the current network state as The baseline network state saved in step S701 is recorded as follows: And calculate their combined link utility respectively. As a preferred method, any candidate topology state Comprehensive link effectiveness It can be calculated using the following formula: ; in, Candidate topological state The overall link utility; This represents the normalized average signal-to-noise ratio under this topology. Normalized average one-way delay; To normalize the success rate; This is a normalized amount of switching cost or path hop cost; , , and For non-negative weights, the preferred option satisfies... The optimal values ​​for each of the above normalization quantities are all within the observation window. Internal statistics are calculated and mapped to the interval [0,1], where, The time can be from 100 milliseconds to 500 milliseconds. It can be determined by at least one of the following: number of handovers, number of path hops, cross-site handover signaling overhead, and control plane reconstruction delay, or a weighted combination thereof. As a preferred approach, when the system prioritizes link reliability, and It can be increased appropriately; when the system is more concerned with switching stability, The size can be increased appropriately. Next, the scheduling gateway unit 30 determines whether to roll back to the baseline state according to the following rules: ; in, This is a topology rollback trigger flag; To provide a utility hysteresis margin and prevent frequent switching when the utilities of two topologies are close, a value of 0.02 to 0.1 is preferred. To avoid rolling back to a failed original connection state, the scheduling gateway unit 30 can also set a validity flag for the baseline state. Only when Only then can the above be executed. Determine; if If so, the current temporary topology is maintained, or the candidate topology selection process is restarted. This can be determined based on whether the original access node is still reachable, whether the original frequency point is still available, and whether the original forwarding path still exists. When this indicates that the baseline network state has regained sufficient advantage, the scheduling gateway unit 30 sends a rollback control command to the communication baseband unit 10 to restore the state to normal. , , , as well as The corresponding network connection configuration; when In such cases, the system continues to retain the current temporary topology to reduce the possibility of prematurely switching back when residual occlusion or local weak coverage still exists.

[0075] S706, when performing network topology rollback, the scheduling gateway unit 30 preferably uses a phased rollback rather than a full simultaneous rollback. Access nodes, routes, beams, and frequency points are typically coupled; if all are switched at the same time, it may introduce control plane reconstruction delay in a short period. As a preferred method, the scheduling gateway unit 30 restores the network topology in batches according to the order of radio access parameters, link forwarding parameters, and service scheduling parameters. Specifically, the scheduling gateway unit 30 can first send a low-level resource preference indication to the communication baseband unit 10, triggering the baseband chip to preferentially search for or request beam numbers. and working frequency After the communication baseband unit 10 completes link quality confirmation for several consecutive sampling cycles under this access condition, the target access node is then restored. Alternatively, use the primary communication interface to finally restore the queue weight set. The above-mentioned recovery order is a preferred implementation; for different wireless access protocol stacks or network control architectures, those skilled in the art can also adjust the recovery order according to the control plane dependency, and such adjustment still falls within the scope of the sub-item rollback described in this application. To determine whether the sub-item rollback has been stably completed, the scheduling gateway unit 30 can set an acknowledgment count for each sub-item. Only in consecutive If the second sample meets the quality requirements, proceed to the next step. As a preferred method, A value of 2 to 5 is acceptable. Adopting this segmented rollback mechanism helps reduce the probability of a new round of business fluctuations triggered by a one-time rollback.

[0076] In step S707, after the network topology rollback is completed, the scheduling gateway unit 30 preferably clears the temporary state records generated during the blocking period, including temporary access identifiers, temporary routing table entries, temporary redundancy parameters, and recovery window timers, after the rollback result has been confirmed as stable. It then switches the system state machine back to normal monitoring mode. If any step in the rollback process triggers the link risk determination in step S50 again, the scheduling gateway unit 30 immediately terminates the rollback process and restores or retains the temporary topology that was still valid during the blocking period to ensure uninterrupted basic communication. Through the above overall steps, link recovery is not directly equivalent to immediately returning to the pre-blocking state. Instead, a controlled rollback is implemented after confirming link recovery stability and the original baseline topology is available again, and a system-level stability confirmation is performed again after the rollback is completed. If the environment deteriorates again, the system can still return to the aforementioned protection and enhancement transmission process. The above-mentioned topology rollback mechanism based on backoff recovery window and hysteresis constraints effectively prevents control plane signaling overload and queue congestion caused by frequent signal fluctuations in the obstruction edge area of ​​communication equipment.

[0077] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A remote communication method for an unmanned vehicle, characterized in that, include: Obtain the effective set of interference points in the environment surrounding the unmanned vehicle, and calculate the dynamic occlusion index based on the effective set of interference points to characterize the degree of communication line-of-sight occlusion. Synchronously acquire the underlying physical channel quality parameters, and perform cross-layer bilateral determination by combining the dynamic occlusion index and the underlying physical channel quality parameters; When the cross-layer bilateral determination meets the triggering conditions, it enters the link protection state. In the link protection state, the regular service flow is intercepted and the core control command is enhanced by a transmission strategy. When the initial recovery conditions are detected, a backoff recovery window is initiated. During the duration of the backoff recovery window, a gradual service recovery is performed, and after the backoff recovery window expires, a network topology rollback is performed. The underlying physical channel quality parameters include signal-to-noise ratio measurements; the cross-layer bilateral determination based on the dynamic occlusion index and the underlying physical channel quality parameters includes: Based on the set time step, calculate the temporal fading rate of the signal-to-noise ratio measurement; When the dynamic occlusion index is greater than the preset occlusion warning threshold and the temporal fading change rate is lower than the maximum tolerable fading rate threshold, the cross-layer bilateral determination is determined to meet the triggering condition. The startup backoff recovery window includes: The normalized recovery determination quantity is calculated by weighting and summing the current value of the dynamic occlusion index, the signal-to-noise ratio measurement value, and the confirmation success rate of the core control command. When the recovery determination quantity is higher than the preset recovery entry threshold within the set continuous sampling period, it is determined that the preliminary recovery condition is met, and the backoff recovery window is started; Based on the current recovery determination quantity and the preset minimum recovery observation time, calculate and set the duration of the backoff recovery window; The gradual service recovery performed during the duration of the backoff recovery window includes: Construct a recovery opening coefficient that gradually increases over time as the retreat recovery window expires; The actual gating coefficient is calculated based on the recovery open coefficient to relax the interception operation of the regular business flow, and the number of redundant copies of the core control command repeatedly sent is gradually reduced according to the recovery open coefficient. If, during the backoff recovery window, the dynamic occlusion index is detected to be higher than the degradation judgment threshold or the signal-to-noise ratio measurement is detected to be lower than the lower limit threshold, the current backoff recovery window is terminated and the system is reset to the link protection state.

2. The remote communication method of an unmanned vehicle according to claim 1, wherein, The calculation of the dynamic occlusion index, which characterizes the degree of line-of-sight occlusion in communication, based on the effective set of interference points includes: For each discrete point in the effective interference point set, a spatial weighting coefficient is calculated. The spatial weighting coefficient is determined by the orthogonal distance from the discrete point to the communication line-of-sight ray and the normalized reflection characteristic of the discrete point. The electromagnetic attenuation coefficient of each discrete point is combined with its corresponding spatial weight coefficient, and an equivalent local volume element is introduced for discrete summation calculation to obtain the original occlusion index. The original occlusion index is normalized using a logic function to generate a scalar value that continuously reflects the occlusion risk as the dynamic occlusion index.

3. The remote communication method of an unmanned vehicle according to claim 1, wherein, The process of intercepting regular business flows includes: Based on the dynamic occlusion index and the occlusion warning threshold, calculate the dynamic truncation factor for the regular service flow; According to the throttling ratio represented by the dynamic throttling factor, the regular service data packets to be sent are redirected to the buffer in the pending state or actively discarded.

4. The remote communication method of an unmanned vehicle according to claim 1, wherein, The enhanced transmission strategy for core control commands includes: Intercept the core control command to be sent and extract the key information payload of the core control command; The network transmission protocol level is lowered, and the extracted key information payload is reconstructed and encapsulated into connectionless user datagram protocol messages or custom Ethernet frames for transmission; The service layer sequence number is included in the reconstructed and encapsulated message, and a retransmission timer is started synchronously. If no application layer status confirmation packet is received after reaching the maximum number of retransmissions or the maximum duration, a transmission failure is reported to the upper layer application.

5. The remote communication method of an unmanned vehicle according to claim 4, wherein, The enhanced transmission strategy for executing core control commands also includes enabling a graded redundancy mechanism based on the risk level range of the dynamic occlusion index: When the dynamic occlusion index is in the medium risk range, an information redundancy strategy is implemented, and the core control command is sent using a low-order modulation and coding scheme that includes forward error correction bits. When the dynamic occlusion index is in the high-risk range, a time redundancy strategy is superimposed on the information redundancy strategy to repeatedly send the same core control command multiple times at a set time interval. When the dynamic occlusion index is in an extreme risk range, a multi-path redundancy strategy is executed, and copies of the core control command are simultaneously sent to primary and backup communication interfaces of different standards for parallel transmission.

6. The remote communication method of an unmanned vehicle according to claim 1, wherein The step of performing network topology rollback after the backoff recovery window expires includes: Extract the baseline network state information saved before entering the link protection state, and calculate the comprehensive link utility of the baseline network state information; the comprehensive link utility is obtained by weighting the statistically normalized values ​​of the underlying physical channel quality parameters, one-way delay, acknowledgment success rate, and topology switching cost. When the overall link utility of the baseline network state information is greater than the sum of the overall link utility and the utility hysteresis margin of the current temporary topology state, a rollback control command is triggered. The rollback operation is performed in the following order: first, restore the receiving beam number and operating frequency; then, restore the target access node; and finally, restore the queue scheduling parameters. After the rollback result is confirmed to be stable, the temporary network status records generated during the blocking period are cleared.

7. A remote communication system for an unmanned vehicle, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the remote communication method for the unmanned vehicle as described in any one of claims 1 to 6.

Images