Vehicle-road-cloud cooperative communication detection method and system

Abstract

The present application relates to the vehicle-road cloud integration cooperative communication and its safety transmission detection technical field, especially in vehicle-road cloud cooperative communication detection method and system, including the collection vehicle node sends baseband complex signal sequence and motion state vector, and combine the road space reflection map of cloud platform issue, according to motion state vector and road space reflection map extraction space propagation doppler frequency offset constraint condition.According to the constraint condition generates environment interference null feature dictionary matrix, the baseband complex signal sequence is executed orthogonal projection operation using environment interference null feature dictionary matrix, obtains the projection residual signal.From the residual signal, the isolated state phase sequence is extracted, and the projection residual energy ratio of the baseband complex signal sequence is calculated.Finally, according to the phase sequence and energy ratio, the abnormality judgment result is generated, the present application can effectively strip the deep overlap of nonlinear phase accumulation and environmental doppler shift, realize the accurate discrimination of hardware sudden failure and instantaneous environmental interference.

Description

Technical Field

[0001] This invention relates to the field of vehicle-road-cloud integrated collaborative communication and its secure transmission detection technology, and particularly to a vehicle-road-cloud collaborative communication detection method and system. Background Technology

[0002] Vehicle-road-cloud cooperative communication and detection technology is a collaborative communication system integrating vehicles, roadside units, and a cloud platform. It aims to monitor traffic environment and safety events through real-time data interaction. Vehicles transmit their status and perception information via onboard communication devices, roadside units collect local traffic data and interact with vehicles, and the cloud platform integrates multi-source information for fusion analysis and decision-making, thereby completing a comprehensive detection of traffic conditions. Based on a hierarchical collaborative architecture, the system expands detection coverage and improves detection accuracy and timeliness through efficient data exchange, providing fundamental support for intelligent traffic management. In scenarios where high-speed vehicle movement causes dynamic changes in network topology, the vehicle-road-cloud cooperative communication mechanism plays a crucial role in maintaining stable system operation.

[0003] Existing vehicle-road-cloud cooperative communication detection technologies suffer from the following technical challenges: In application scenarios of integrated vehicle-road-cloud cooperative secure transmission technologies with highly dynamic topology characteristics, the severe Doppler frequency shift caused by the high-speed displacement of vehicles leads to significant spectral broadening of the wireless link signal. Simultaneously, the inherent clock jitter and nonlinear RF characteristics of mobile node hardware cause a nonlinear phase deviation in the feedback signal that evolves over time during transmission. When the frequency drift caused by the Doppler frequency shift and the aforementioned nonlinear phase overlap deeply in the frequency domain, the composite signal characteristics received by the detection end exhibit extremely high similarity. This results in a lack of sufficient observation window to distinguish whether the signal distortion originates from a sudden hardware failure in the RF front-end or from random interference caused by a transient and harsh electromagnetic environment, even under ultra-low latency constraints. For example, during coordinated patrols in a platoon on a highway, when the roadside unit receives feedback link signals, the rapid fading caused by the high-speed movement of vehicles and the instantaneous frequency drift of the oscillator inside the vehicle communication terminal are very likely to overlap. This makes it impossible for the monitoring center to determine whether the degradation of the transmission link quality is due to hardware failure caused by component wear or to increased interference caused by multipath reflections from complex building clusters. Consequently, it may trigger incorrect emergency response decisions in a very short time, posing a potential threat to the overall operational safety of the system. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a vehicle-road-cloud cooperative communication detection method and system. This invention solves the technical problem that the detection end cannot effectively isolate sudden hardware failures and instantaneous environmental interference under ultra-low latency constraints due to the nonlinear phase accumulation generated by high-speed mobile nodes in the cooperative feedback link and the deep overlap of environmental Doppler frequency shift.

[0005] To solve the above-mentioned technical problems, the specific contents of the present invention are as follows: In a first aspect, the vehicle-road-cloud cooperative communication detection method provided by the present invention includes: Step 1: Collect the baseband complex signal sequence and motion state vector sent by the vehicle node, and obtain the road space reflection map sent by the cloud platform; Step 2: Extract spatial propagation Doppler frequency offset constraints based on the motion state vector and the road spatial reflection map, and generate an environmental interference null feature dictionary matrix based on the spatial propagation Doppler frequency offset constraints; Step 3: Perform orthogonal projection operation on the baseband complex signal sequence using the environmental interference null feature dictionary matrix to obtain the projected residual signal; Step 4: Extract the isolated state phase sequence from the projected residual signal, and calculate the projection residual energy ratio based on the projected residual signal and the baseband complex signal sequence; Step 5: Generate an anomaly determination result based on the ratio of the isolated phase sequence to the projected residual energy. The anomaly determination result includes environmental interference or hardware failure. Output communication link adjustment control signaling based on the anomaly determination result.

[0006] Furthermore, the vehicle-road-cloud cooperative communication detection method of the present invention, wherein the step of extracting spatial propagation Doppler frequency offset constraints based on the motion state vector and the road spatial reflection map, and generating an environmental interference null feature dictionary matrix based on the spatial propagation Doppler frequency offset constraints, includes: Extract the boundary three-dimensional coordinates and surface material reflectance coefficients from the road space reflection map; Based on the three-dimensional coordinates of the boundary, the reflectivity of the surface material, and the motion state vector, calculate the spatial reflection path arrival angle and the instantaneous Doppler frequency offset measurement value; The instantaneous Doppler frequency offset measurement value is determined as the spatial propagation Doppler frequency offset constraint condition, and the environmental Doppler constraint manifold matrix is ​​constructed using the spatial propagation Doppler frequency offset constraint condition. Perform singular value decomposition on the environmental Doppler-constrained manifold matrix to extract the noise subspace basis vector matrix; A projection matrix is ​​constructed using the noise subspace basis vector matrix, and the projection matrix is ​​determined as the environmental interference null feature dictionary matrix.

[0007] Furthermore, in the vehicle-road-cloud cooperative communication detection method of the present invention, the step of performing orthogonal projection operation on the baseband complex signal sequence using the environmental interference null feature dictionary matrix to obtain the projected residual signal includes: Convert the baseband complex signal sequence into a baseband complex signal vector; The baseband complex signal vector is multiplied with the environmental interference null feature dictionary matrix to extract the projection component of the baseband complex signal vector in the subspace spanned by the environmental interference null feature dictionary matrix. The projection component is determined as the projection residual signal.

[0008] Furthermore, in the vehicle-road-cloud cooperative communication detection method of the present invention, the step of extracting the isolated state phase sequence from the projected residual signal includes: Extract the complex sample points from the projected residual signal; The argument extraction calculation is performed on the complex sample points to output the original phase evolution sequence; Perform a phase unwinding operation on the original phase evolution sequence to output a continuous phase sequence; The continuous phase sequence is determined as the isolated phase sequence.

[0009] Furthermore, in the vehicle-road-cloud cooperative communication detection method of the present invention, the step of calculating the projection residual energy ratio based on the projected residual signal and the baseband complex signal sequence includes: Calculate the square value of the complex modulus of the projected residual signal, and determine the square value of the complex modulus of the projected residual signal as the null space projection energy value. Calculate the square value of the complex modulus length of the baseband complex signal sequence, and determine the square value of the complex modulus length of the baseband complex signal sequence as the total received energy value in the entire domain; Calculate the ratio of the zero-depression space projected energy value to the total received energy value over the entire area; The ratio value is determined as the projection residual energy ratio.

[0010] Furthermore, in the vehicle-road-cloud cooperative communication detection method of the present invention, the step of generating an anomaly determination result based on the ratio of the isolated state phase sequence to the projected residual energy includes: Obtain the preset alarm threshold; The projection residual energy ratio is compared with the preset alarm threshold. Perform a polynomial fitting operation on the isolated phase sequence and extract the coefficients of the quadratic term generated by the polynomial fitting operation; When the projection residual energy ratio is greater than the preset alarm threshold and the quadratic term coefficient is greater than 0, the abnormal judgment result is generated as a hardware fault.

[0011] Furthermore, the vehicle-road-cloud cooperative communication detection method of the present invention, wherein generating an anomaly determination result based on the ratio of the isolated state phase sequence to the projected residual energy, further includes: Obtain the preset discrete threshold; Calculate the phase variance of the isolated phase sequence; When the projection residual energy ratio is less than or equal to the preset alarm threshold, and the phase variance value is greater than the preset discrete threshold, the abnormal judgment result is generated as environmental interference.

[0012] Furthermore, in the vehicle-road-cloud cooperative communication detection method of the present invention, the step of outputting communication link adjustment control signaling based on the anomaly determination result includes: When the anomaly determination result is a hardware failure, the vehicle takeover warning signal is identified as the communication link adjustment control signal and sent to the vehicle node, and a hardware failure identifier is sent to the cloud platform; When the anomaly determination result is environmental interference, the air interface retransmission control signaling or the coding redundancy adjustment control signaling is determined as the communication link adjustment control signaling and sent to the vehicle node.

[0013] Furthermore, the vehicle-road-cloud cooperative communication detection method of the present invention, wherein the step of collecting the baseband complex signal sequence and motion state vector sent by the vehicle node and obtaining the road spatial reflection map issued by the cloud platform includes: Obtain the time reference; Based on the timing reference, perform timestamp comparison calculations on the baseband complex signal sequence, the motion state vector, and the road spatial reflection map; The baseband complex signal sequence after performing the timestamp comparison operation, the motion state vector, and the road spatial reflection map are spliced ​​together to generate a data matrix to be processed. Parse the target baseband complex signal sequence with the same timestamp, the target motion state vector, and the target road spatial reflection map from the data matrix to be processed; The target baseband complex signal sequence is determined as the baseband complex signal sequence, the target motion state vector is determined as the motion state vector, and the target road spatial reflection map is determined as the road spatial reflection map.

[0014] Secondly, the vehicle-road-cloud cooperative communication detection system provided by the present invention is applied to the vehicle-road-cloud cooperative communication detection method as described above, including: The data acquisition module is used to acquire the baseband complex signal sequence, motion state vector, and road space reflection map sent by the vehicle node and the cloud platform. A manifold construction module is used to extract spatial propagation Doppler frequency offset constraints based on the motion state vector and the road spatial reflection map; The dictionary generation module is used to generate an environmental interference null feature dictionary matrix based on the spatial propagation Doppler frequency offset constraint condition. The projection mapping module is used to perform orthogonal projection operation on the baseband complex signal sequence using the environmental interference null feature dictionary matrix to obtain the projected residual signal; The feature extraction module is used to parse the isolated phase sequence from the projected residual signal and calculate the ratio of the projected residual energy of the projected residual signal to the baseband complex signal sequence. The state determination module is used to generate anomaly determination results based on the isolated state phase sequence and the ratio of projected residual energy. The instruction linkage module is used to issue communication link adjustment control signaling based on the anomaly determination result.

[0015] Beneficial effects of this invention: This invention establishes spatial propagation Doppler frequency offset constraints by introducing a road spatial reflection map and motion state vectors, achieving deep decoupling between physical spatial geometric priors and underlying communication physical layer characteristics. This effectively solves the technical pain point of difficulty in distinguishing anomaly sources caused by the deep overlap of Doppler frequency shift and hardware phase distortion in high-speed moving scenarios. By performing orthogonal projection operations on the baseband complex signal sequence using the environmental Doppler constraint manifold matrix and the environmental interference null feature dictionary matrix, environmental interference components conforming to physical propagation laws can be accurately suppressed, allowing the projected residual signal to truly represent the hardware nonlinear phase characteristics independent of environmental factors. Based on polynomial fitting operations and quadratic term coefficient extraction of the isolated state phase sequence, combined with the quantitative comparison of the projected residual energy ratio and the preset alarm threshold, a highly interpretable hardware fault determination mechanism is established, effectively avoiding false triggering by traditional energy detection methods in fast fading environments. By directly linking the anomaly determination results to the differentiated outputs of vehicle takeover warning signaling, air interface retransmission control signaling, and coding redundancy adjustment control signaling, the communication link adjustment actions can respond to the anomaly in a targeted manner, significantly reducing the probability of erroneous emergency response in vehicle-road-cloud integrated collaborative safe transmission scenarios, and improving the system's operational stability and the accuracy of emergency response in dynamic topology networks. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on the drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the vehicle-road-cloud collaborative communication detection method of the present invention. Detailed Implementation

[0018] To make the technical solution of the present invention clearer, the present invention will be clearly and completely described below with reference to specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. 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. The present invention provided by various embodiments will be described in detail below with reference to the accompanying drawings. To better understand the purpose of the present invention, the present invention will be described in further detail below.

[0019] Firstly, please refer to Figure 1 The vehicle-road-cloud cooperative communication detection method provided by the present invention includes: Step 1: Collect the baseband complex signal sequence and motion state vector sent by the vehicle node, and obtain the road space reflection map sent by the cloud platform; Step 2: Extract spatial propagation Doppler frequency offset constraints based on the motion state vector and the road spatial reflection map, and generate an environmental interference null feature dictionary matrix based on the spatial propagation Doppler frequency offset constraints; Step 3: Perform orthogonal projection operation on the baseband complex signal sequence using the environmental interference null feature dictionary matrix to obtain the projected residual signal; Step 4: Extract the isolated state phase sequence from the projected residual signal, and calculate the projection residual energy ratio based on the projected residual signal and the baseband complex signal sequence; Step 5: Generate an anomaly determination result based on the ratio of the isolated phase sequence to the projected residual energy. The anomaly determination result includes environmental interference or hardware failure. Output communication link adjustment control signaling based on the anomaly determination result.

[0020] The vehicle-road-cloud cooperative communication detection method is applicable to integrated vehicle-road-cloud cooperative secure transmission scenarios with highly dynamic topology characteristics. Application scenarios include high-speed vehicle travel, continuous switching of vehicle-roadside unit access relationships, the presence of building reflection interfaces around the road, occlusion switching areas, and multipath enhancement areas. Vehicle-mounted nodes continuously move along the road and send baseband complex signal sequences and motion status information to the roadside units via a wireless feedback link. The cloud platform maintains a road spatial reflection map and, based on the vehicle's current location or target road segment, distributes the corresponding road spatial reflection map to the roadside units or edge detection nodes. The detection function can be deployed inside the roadside unit or on an edge processor connected to the roadside unit, used to determine the source of anomalies under low-latency constraints and output communication link adjustment control signaling.

[0021] In one embodiment, the vehicle-mounted node may include a vehicle-mounted communication module, an inertial measurement unit, a satellite positioning receiving unit, and a local clock unit. The roadside unit may include a radio frequency transceiver module, a baseband processing module, an edge processor, and a timing module. The cloud platform may include a map maintenance unit, a road segment matching unit, a reflection information management unit, and a distribution interface. The vehicle-mounted communication module is responsible for forming and transmitting baseband complex signal sequences; the inertial measurement unit and the satellite positioning receiving unit jointly form a motion state vector; the timing module provides a unified timing reference to the roadside unit; and the edge processor performs manifold construction, dictionary generation, projection mapping, feature extraction, state determination, and command linkage processing. The road spatial reflection map can be generated by the cloud platform based on road surveying data, road boundary models, typical material reflection parameters, and historical maintenance information, and organized according to road segmentation.

[0022] In one embodiment, the baseband complex signal sequence is a set of complex samples arranged in order of sampling time, where each complex sample contains the in-phase component value and the quadrature component value at the same sampling time. The motion state vector includes at least vehicle position, vehicle speed, vehicle heading, and timestamp information. The road spatial reflection map includes at least the boundary three-dimensional coordinates and surface material reflection coefficients, along with road segment identification and timestamp information. To facilitate subsequent correlation processing, the baseband complex signal sequence, motion state vector, and road spatial reflection map are all organized using timestamps as a synchronization reference. The baseband complex signal sequence reflects frequency drift, phase evolution, and energy distribution information in the receiving link; the motion state vector is used to characterize the propagation path changes caused by vehicle motion; and the road spatial reflection map is used to characterize the spatial reflection boundary and the distribution of reflection strength. The above input data are used together to establish environmental propagation constraints, thereby distinguishing between environmental interference components and hardware fault components.

[0023] In one embodiment, the method first performs data acquisition and time alignment processing. After the roadside unit or edge processor acquires the timing reference, it reads the baseband complex signal sequence, motion state vector, and road spatial reflection map within the current processing window, and performs timestamp comparison operations on the three types of data. The timestamp comparison operation is used to filter data records that are at the same processing time or within the same tolerance range. The tolerance range can be set according to the wireless frame length, edge processing cycle, or timing accuracy. After the timestamp comparison is completed, the filtered data records are spliced ​​to generate a data matrix to be processed. In the data matrix to be processed, each record corresponds to at least one common timestamp and is associated with a target baseband complex signal sequence, a target motion state vector, and a target road spatial reflection map. After parsing, data with the same timestamp are identified as the baseband complex signal sequence, motion state vector, and road spatial reflection map used in subsequent processing. Through the above processing, propagation constraint offset caused by mixing data from different time slices can be avoided, improving the temporal consistency of subsequent anomaly source determination.

[0024] In one embodiment, the current processing window is organized using a sliding update method, with the window starting point continuously advancing according to the timing reference, and the window continuously containing multiple sets of complex sample points. Each complex sample point in the baseband complex signal sequence carries a sampling time label, each record in the motion state vector carries a location time label, and each boundary record in the road spatial reflection map carries a road segment identifier and a map version time label. The timestamp comparison operation includes the following processing: slicing the baseband complex signal sequence into windows according to the timing reference, extracting motion state vector records according to the same time range, and extracting road spatial reflection map records according to the road segment identifier corresponding to the vehicle's current position; for records with time differences exceeding the tolerance range, they are directly marked as mismatched records and removed from the data matrix to be processed; for records lacking time labels or with time label bounces, they are directly marked as invalid records and stopped from participating in the current processing window operation. Each record in the data matrix to be processed includes at least a baseband complex signal segment within the same processing window, a motion state vector record at the corresponding time, and a road space reflection map record for the corresponding road segment. This ensures that the spatial propagation Doppler frequency offset constraint, the projected residual signal, and the ratio of the isolated phase sequence to the projected residual energy all correspond to the real propagation environment within the same time slice.

[0025] In one embodiment, the baseband complex signal sequence can be directly output from the roadside unit receiving link, or it can be recovered at the roadside unit after feedback from the vehicle-mounted node. To improve the processability of the sequence, before entering projection processing, the baseband complex signal sequence can be vectorized and rearranged according to the processing window length, arranging the continuously sampled complex samples into a baseband complex signal vector in chronological order. The processing window length can be set according to low latency requirements, for example, organized according to a feedback frame, multiple feedback symbols, or a short observation segment. The motion state vector can be obtained by fusing the position and velocity information output by the satellite positioning receiving unit and the attitude change information output by the inertial measurement unit. The boundary three-dimensional coordinates in the road spatial reflection map can be used to represent the positional distribution of guardrails, building facades, sound barriers, or other fixed reflective boundaries on both sides of the road, and the surface material reflectivity can be used to represent the reflectivity of the corresponding boundary to the wireless propagation path. By unifying the motion state information and spatial reflection information into the same processing window, direct input can be provided for Doppler frequency offset constraint extraction.

[0026] In one embodiment, the in-phase and quadrature component values ​​in the baseband complex signal sequence are collected according to a uniform sampling period and formed into a one-dimensional time series according to the sampling order. For complex sample points whose amplitudes significantly exceed the receiver's dynamic range, the edge processor marks the corresponding complex sample points as clipped sample points and replaces them using interpolation of adjacent valid sample points. For continuously missing complex sample points, the edge processor records the corresponding processing window as a low-confidence window and reduces the judgment priority of the current processing window during the anomaly judgment stage. The motion state vector is formed by fusing the position record output by the satellite positioning receiver unit and the attitude change record output by the inertial measurement unit. The fusion result includes at least vehicle position, vehicle speed, vehicle heading, and timestamp information. For a sudden change in vehicle speed but no corresponding attitude change record provided by the inertial measurement unit, the edge processor records the corresponding motion state vector as an abnormal motion record and stops participating in the calculation of the space propagation Doppler frequency offset constraint condition, thereby reducing the propagation of abnormal positioning errors to the path angle of arrival calculation results.

[0027] In one embodiment, the method enters the spatial propagation Doppler frequency offset constraint extraction stage. The edge processor first extracts the boundary 3D coordinates and surface material reflectance coefficients of the road segment corresponding to the current vehicle position from the road spatial reflection map. The boundary 3D coordinates are used to give the positional relationship of the spatial reflection boundaries, and the surface material reflectance coefficients are used to characterize the differences in the contribution of different boundaries to the amplitude of the reflection path. Subsequently, based on the boundary 3D coordinates, surface material reflectance coefficients, and motion state vectors, the edge processor calculates the spatial reflection path arrival angle and instantaneous Doppler frequency offset measurement value for each candidate spatial reflection path. The spatial reflection path arrival angle is used to characterize the angular relationship of the reflection path entering the receiver, and the instantaneous Doppler frequency offset measurement value is used to characterize the frequency shift that the corresponding propagation path may form at the current moment. Multiple candidate paths correspond to multiple instantaneous Doppler frequency offset measurement values, and these multiple instantaneous Doppler frequency offset measurement values ​​together constitute the spatial propagation Doppler frequency offset constraint conditions. Through these constraint conditions, the environmental propagation characteristics jointly determined by vehicle motion and the road reflection environment can be explicitly written into the subsequent projection space.

[0028] In one embodiment, the edge processor first performs road segment matching based on the vehicle's position in the motion state vector, and then extracts boundary records related to the forward propagation direction from the road spatial reflection map. The boundary record filtering rules include: the boundary's three-dimensional coordinates are within a preset distance range of the vehicle's current position; the surface material's reflectivity is higher than a preset lower limit; and the angle between the boundary orientation and the vehicle's heading is within a preset angle range. For boundary records exceeding the preset distance range, boundary records with excessively low surface material reflectivity, and boundary records located in the vehicle's rear blind spot, the edge processor does not include the corresponding records in the candidate spatial reflection path set. After the candidate spatial reflection path set is formed, the edge processor sequentially calculates the spatial reflection path arrival angle and instantaneous Doppler frequency offset measurement value for each candidate spatial reflection path. By setting boundary record filtering rules, the spatial propagation Doppler frequency offset constraint conditions can prioritize reflecting propagation paths with a strong impact on the receiving link within the current processing window, reducing the impact of invalid boundary records on the stability of the environmental Doppler constraint manifold matrix.

[0029] In one embodiment, the edge processor constructs an environmental Doppler constraint manifold matrix based on the spatial propagation Doppler frequency offset constraint. Each column in the environmental Doppler constraint manifold matrix represents the constraint direction corresponding to a candidate propagation path, and different columns collectively represent the range of environmental interference variations that may occur under the current road environment. After the manifold matrix is ​​constructed, the edge processor performs singular value decomposition on the environmental Doppler constraint manifold matrix and extracts the noise subspace basis vector matrix based on the decomposition result. The noise subspace basis vector matrix is ​​used to represent subspace directions that are orthogonal to or weakly correlated with the dominant environmental propagation direction. Subsequently, the edge processor constructs a projection matrix using the noise subspace basis vector matrix and determines the projection matrix as the environmental interference null feature dictionary matrix. The technical role of the environmental interference null feature dictionary matrix is ​​to suppress components highly correlated with environmental interference constraints to a low level, thereby creating conditions for subsequent identification of residual components related to hardware anomalies.

[0030] In one embodiment, the number of rows in the environmental Doppler constraint manifold matrix corresponds to the number of complex sample points in the current processing window, and the number of columns corresponds to the number of candidate spatial reflection paths. Each column records the phase evolution direction caused by a candidate spatial reflection path within the current processing window, and different columns collectively characterize the legal environmental disturbance range under the joint constraints of the road spatial reflection map and the motion state vector. After the singular value decomposition operation is completed, the edge processor divides the main direction component and the weak direction component according to the magnitude of the singular values, and extracts the noise subspace basis vector matrix from the column vectors corresponding to the weak direction components. The projection matrix and the environmental disturbance null feature dictionary matrix are set with the same dimensions and are used to directly perform multiplication operations with the baseband complex signal vector. By clarifying the correspondence between row dimensions and column dimensions, those skilled in the art can directly understand the construction order and data interface relationship between the environmental Doppler constraint manifold matrix, the noise subspace basis vector matrix, and the environmental disturbance null feature dictionary matrix.

[0031] In one embodiment, the method enters the orthogonal projection processing stage. The edge processor converts the aforementioned baseband complex signal sequence into a baseband complex signal vector, and then multiplies the baseband complex signal vector with the environmental interference null feature dictionary matrix. The result of the multiplication operation is used to extract the projection component of the baseband complex signal vector within the spanned subspace corresponding to the environmental interference null feature dictionary matrix. The projection component represents the residual components retained after mapping through the environmental constraint subspace. Since the dominant environmental propagation components have been constrained during the dictionary matrix construction stage, the information retained in the projection component is closer to the anomalous changes caused by non-environmental factors. The method identifies the projection component as the projected residual signal. The projected residual signal serves as the direct input for subsequent phase extraction and energy ratio calculation.

[0032] In one embodiment, the method extracts an isolated-state phase sequence from the projected residual signal. The edge processor first extracts complex sample points from the projected residual signal and performs an argument extraction calculation for each complex sample point, outputting an original phase evolution sequence. This original phase evolution sequence reflects the phase trajectory of the residual signal as it changes over time. Since phase jumps may occur at adjacent period boundaries, the edge processor continues to perform phase dewinding operations on the original phase evolution sequence, connecting the jumps across period boundaries into a continuous trajectory, outputting a continuous phase sequence. This continuous phase sequence is identified as the isolated-state phase sequence. The isolated-state phase sequence is used to reflect the phase evolution of residual anomalous components after environmental interference nulling. Hardware faults typically cause phase changes that accumulate continuously or exhibit a significant bending trend, while environmental interference typically causes phase fluctuations with higher dispersion and greater randomness. Therefore, the isolated-state phase sequence can provide a direct basis for subsequent judgments.

[0033] In one embodiment, the ratio of the isolated-state phase sequence to the projected residual energy is calculated based on the projected residual signal within the same processing window. The edge processor processes the complex sample points in the projected residual signal point by point in the sampling order, first forming a continuous phase sequence, and then performing trend analysis on the continuous phase sequence; simultaneously, it performs modulus square accumulation processing on the same batch of complex sample points to form the null space projected energy value. The total received energy value of the entire domain is statistically analyzed using the same window start and end points as the null space projected energy value, thereby ensuring that the isolated-state phase sequence, the null space projected energy value, and the total received energy value of the entire domain are consistent in time range. By unifying the statistical boundaries, it is possible to prevent deviations in the anomaly detection results caused by the isolated-state phase sequence and the projected residual energy ratio coming from different time periods.

[0034] In one embodiment, the method further calculates the projection residual energy ratio. The edge processor first calculates the squared magnitude of each complex sample in the projected residual signal and accumulates or sums the corresponding results to obtain the null space projection energy value. Subsequently, the edge processor calculates the squared magnitude of each complex sample in the baseband complex signal sequence and accumulates or sums the corresponding results to obtain the total received energy value across the entire domain. The edge processor continues to calculate the ratio of the null space projection energy value to the total received energy value across the entire domain and determines this ratio as the projection residual energy ratio. The projection residual energy ratio is used to represent the proportion of residual anomalous components in the total received energy after environmental constraint suppression. A higher projection residual energy ratio indicates a more prominent anomalous component inconsistent with environmental propagation constraints, and is more likely to correspond to abnormal changes at the hardware level; a lower projection residual energy ratio indicates that most of the received energy can still be explained by environmental propagation constraints, and is more likely to correspond to environmental interference dominating.

[0035] In one embodiment, the method enters the anomaly determination phase. The edge processor first obtains a preset alarm threshold and compares the projected residual energy ratio with the preset alarm threshold. The preset alarm threshold is used to distinguish between high residual ratio states and low residual ratio states. The preset alarm threshold can be set based on historical calibration data, link stress test results, or statistical results under different road levels. The edge processor also performs a polynomial fitting operation on the isolated phase sequence and extracts the quadratic term coefficient from the fitting result. The quadratic term coefficient is used to characterize whether the isolated phase sequence has a significant bending trend within the observation window. If the projected residual energy ratio is greater than the preset alarm threshold and the quadratic term coefficient is greater than zero, an anomaly determination result is generated as a hardware fault. The technical meaning of the above determination logic is that: high energy residuals indicate a decrease in environmental interpretability, while a positively bending phase trend indicates that the abnormal phase is not randomly discretely distributed, but rather a continuous accumulation or aggravation phenomenon over time, thus being closer to phase anomalies caused by hardware oscillation instability, radio frequency nonlinear mutations, or device failures.

[0036] In one embodiment, the method can also determine environmental interference. The edge processor first obtains a preset discrete threshold and calculates the phase variance value based on the isolated phase sequence. The phase variance value is used to characterize the dispersion of the isolated phase sequence within the observation window. The preset discrete threshold can be set according to different road segment types, multipath complexity, or historical environmental noise statistics. If the projected residual energy ratio is less than or equal to a preset alarm threshold, and the phase variance value is greater than the preset discrete threshold, an anomaly determination result is generated as environmental interference. The technical meaning of the above determination logic is that a low residual ratio indicates that most reception characteristics still conform to environmental propagation constraints, while high discrete phase fluctuations indicate the presence of highly random phase disturbances in a short period of time, which is more consistent with environmental interference characteristics caused by fast fading, multipath enhancement, or obstruction switching.

[0037] In one embodiment, the process of setting a preset alarm threshold includes: collecting the projection residual energy ratios corresponding to multiple normal communication state processing windows, grouping and statistically analyzing them according to road type, vehicle speed range, and access mode, and forming a corresponding threshold entry for each group of statistical results; during the operation phase, the edge processor calls the corresponding threshold entry based on the current location road type, current vehicle speed range, and current access mode, using it as the preset alarm threshold for the current processing window. The process of setting a preset discrete threshold includes: collecting the phase variance values ​​corresponding to multiple environmental interference-dominated state processing windows, grouping and statistically analyzing them according to the complexity of road segment reflections and the frequency of occlusion, and forming a corresponding phase discrete threshold entry; during the operation phase, the edge processor calls the corresponding threshold entry based on the boundary density and surface material reflectance coefficient distribution characteristics in the road spatial reflection map, using it as the preset discrete threshold for the current processing window. By using a grouped statistical method to set the preset alarm threshold and preset discrete threshold, the anomaly detection results can be adapted to different road environments and different operating speed conditions.

[0038] In one embodiment, if neither the projected residual energy ratio nor the isolated phase sequence meets the hardware fault determination criteria nor the environmental interference determination criteria, the current processing result can be recorded as a state to be observed further, and the above steps can be repeated in subsequent processing windows. This setting can be used to shorten the length of a single observation window while avoiding premature triggering of false judgments in boundary states. The state to be observed further may not trigger forced link adjustments, or it may only output a low-level monitoring identifier for the cloud platform to record. This processing method is an optional execution strategy that revolves around the process described in the claims.

[0039] In one embodiment, the "waiting for continued observation" state corresponds to an intermediate judgment state. After entering the waiting for continued observation state, the edge processor does not immediately output vehicle takeover warning signals, air interface retransmission control signals, or coding redundancy adjustment control signals. Instead, it writes the isolated state phase sequence summary value, projected residual energy ratio, and time tag corresponding to the current processing window into the state buffer. When subsequent processing windows arrive, the edge processor continuously compares the summary values ​​in multiple adjacent processing windows. If 3 to 5 consecutive processing windows approach the hardware fault judgment condition, it outputs the abnormal judgment result corresponding to the hardware fault. If 3 to 5 consecutive processing windows approach the environmental interference judgment condition, it outputs the abnormal judgment result corresponding to the environmental interference. If 3 to 5 consecutive processing windows return to the normal fluctuation range, the intermediate judgment record in the state buffer is cleared. By setting the waiting for continued observation state, the probability of misjudgment caused by single-window boundary noise can be reduced.

[0040] In one embodiment, the method enters the control signaling output stage. When the anomaly determination result is a hardware failure, the edge processor identifies the vehicle takeover warning signaling as a communication link adjustment control signaling and sends it to the vehicle node, while simultaneously sending a hardware failure identifier to the cloud platform. The vehicle takeover warning signaling is used to prompt the vehicle node to enter a higher security level processing state, and the hardware failure identifier is used by the cloud platform to perform fault recording, node status updates, or operation and maintenance linkage. When the anomaly determination result is environmental interference, the edge processor identifies the air interface retransmission control signaling or the coding redundancy adjustment control signaling as a communication link adjustment control signaling and sends it to the vehicle node. The air interface retransmission control signaling is used to trigger the link layer retransmission process, and the coding redundancy adjustment control signaling is used to improve the data anti-interference capability of the current feedback link. By mapping different anomaly sources to different control actions, the control link response can be kept consistent with the nature of the anomaly, thereby reducing erroneous emergency responses and lowering the probability of link mishandling.

[0041] In one embodiment, the vehicle takeover warning signal is sent immediately after the hardware fault anomaly determination result is formed and remains valid until the subsequent processing window returns to normal. After receiving the hardware fault identifier, the cloud platform records the corresponding vehicle node in the fault node list and reduces the priority of subsequent collaborative control task allocation. The air interface retransmission control signal is used to trigger the retransmission process of the currently lost data packet or low-reliability data packet, and the coding redundancy adjustment control signal is used to improve the anti-interference capability in the subsequent transmission cycle. When the environmental interference determination condition is no longer met for 3 to 5 consecutive processing windows, the edge processor stops sending the air interface retransmission control signal or the coding redundancy adjustment control signal and restores the normal communication link configuration. By supplementing the timing of the control signal activation and the recovery conditions, a closed-loop correspondence can be formed between the anomaly determination result and the communication link adjustment action.

[0042] In one embodiment, the above method is executed during the inference and detection phase, and the training phase is not a necessary condition. This is because the core of the method lies in establishing environmental Doppler constraints based on motion state vectors and road spatial reflection maps, and determining the source of anomalies through orthogonal projection, phase extraction, and energy ratio calculation. The entire processing flow does not rely on the black-box model training results, facilitating stable and interpretable execution in edge processing scenarios. For map maintenance and road segment reflection information updates on the cloud platform side, offline updates can be performed according to road inspection cycles, historical communication quality changes, or environmental modification events. The offline update results are distributed in the form of road spatial reflection maps during the detection phase.

[0043] Secondly, the vehicle-road-cloud cooperative communication detection system provided by the present invention is applied to the vehicle-road-cloud cooperative communication detection method as described above, including: The data acquisition module is used to acquire the baseband complex signal sequence, motion state vector, and road space reflection map sent by the vehicle node and the cloud platform. A manifold construction module is used to extract spatial propagation Doppler frequency offset constraints based on the motion state vector and the road spatial reflection map; The dictionary generation module is used to generate an environmental interference null feature dictionary matrix based on the spatial propagation Doppler frequency offset constraint condition. The projection mapping module is used to perform orthogonal projection operation on the baseband complex signal sequence using the environmental interference null feature dictionary matrix to obtain the projected residual signal; The feature extraction module is used to parse the isolated phase sequence from the projected residual signal and calculate the ratio of the projected residual energy of the projected residual signal to the baseband complex signal sequence. The state determination module is used to generate anomaly determination results based on the isolated state phase sequence and the ratio of projected residual energy. The instruction linkage module is used to issue communication link adjustment control signaling based on the anomaly determination result.

[0044] Embodiment 1 of this invention: In a highway platooning scenario, the vehicle-mounted node periodically sends feedback baseband complex signals to the roadside unit. The satellite positioning receiving unit and inertial measurement unit in the vehicle-mounted node synchronously output vehicle position, speed, and heading information, forming a motion state vector. The cloud platform sends a road space reflection map of the corresponding road segment to the roadside unit based on the road segment where the vehicle is located. The road space reflection map records the three-dimensional coordinates of the boundaries of guardrails, sound barriers, and adjacent building facades, and records the surface material reflection coefficients corresponding to different boundary materials. The timing module in the roadside unit provides a unified timing reference to the edge processor. Based on the unified timing reference, the edge processor performs timestamp comparison operations on the baseband complex signal sequence, motion state vector, and road space reflection map in the current processing window, and splices data with consistent timestamps into a data matrix to be processed. The edge processor then parses the target baseband complex signal sequence, target motion state vector, and target road space reflection map corresponding to the same timestamp from the data matrix to be processed.

[0045] After data alignment, the edge processor reads the boundary 3D coordinates and surface material reflectance coefficients from the target road spatial reflection map. Combining this with vehicle position, speed, and heading information from the target motion state vector, it calculates the spatial reflection path arrival angle and instantaneous Doppler frequency offset for each candidate spatial reflection path. Multiple instantaneous Doppler frequency offset measurements are organized into spatial propagation Doppler frequency offset constraints. Based on these constraints, the edge processor constructs an environmental Doppler constraint manifold matrix and performs singular value decomposition on it to extract the noise subspace basis vector matrix. Finally, the edge processor uses the noise subspace basis vector matrix to construct an environmental interference null feature dictionary matrix.

[0046] The edge processor converts the target baseband complex signal sequence into a baseband complex signal vector, and multiplies the baseband complex signal vector with the environmental interference null feature dictionary matrix to obtain the projection component. The projection component is determined as the projected residual signal. The edge processor extracts complex sample points from the projected residual signal, performs argument extraction calculation on each of the complex sample points, outputs the original phase evolution sequence, and then performs phase dewinding operation on the original phase evolution sequence to output a continuous phase sequence, which is determined as the isolated state phase sequence. Subsequently, the edge processor calculates the cumulative squared magnitude of the projected residual signal to obtain the null space projection energy value; calculates the cumulative squared magnitude of the target baseband complex signal sequence to obtain the total received energy value; and calculates the ratio of the two to obtain the projection residual energy ratio.

[0047] The edge processor reads a preset alarm threshold, compares the projected residual energy ratio with the preset alarm threshold, and simultaneously performs a polynomial fitting operation on the isolated phase sequence to extract the quadratic coefficient. If the projected residual energy ratio is greater than the preset alarm threshold and the quadratic coefficient is greater than zero, the edge processor determines the anomaly to be a hardware fault and sends a vehicle takeover warning signal to the vehicle node, while also sending a hardware fault identifier to the cloud platform. If the projected residual energy ratio is not greater than the preset alarm threshold, the edge processor further calculates the phase variance of the isolated phase sequence and compares it with a preset discrete threshold. If the phase variance is greater than the preset discrete threshold, the edge processor determines the anomaly to be environmental interference and sends an air interface retransmission control signal or a coding redundancy adjustment control signal to the vehicle node. Through this processing flow, environmental interference and hardware faults can be distinguished within a short observation window, and the control actions can correspond to two different technical processing links: interference suppression and safe takeover.

[0048] Embodiment 2 of this invention: In the scenario of urban expressways, there are dense buildings and viaducts on both sides of the road, resulting in a higher number of reflection boundaries than in open highway scenarios. The cloud platform can store road spatial reflection maps in a road segment grid format and pre-download the corresponding road spatial reflection maps to the roadside units before a vehicle enters the target grid. After reading the road spatial reflection map, the edge processor can prioritize selecting boundary records that are close to the vehicle's current position and have a high reflection coefficient to participate in the construction of spatial propagation Doppler frequency offset constraints, thereby controlling the scale of the environmental Doppler constraint manifold matrix and meeting the requirements for low-latency detection. Subsequent processing still includes timestamp comparison, data stitching, target data parsing, manifold matrix construction, singular value decomposition, dictionary matrix generation, orthogonal projection, isolated state phase sequence extraction, projection residual energy ratio calculation, anomaly detection, and communication link adjustment control signaling output. By filtering boundary records for urban multi-reflection boundary scenarios, the processing efficiency of the edge processor in complex scenarios can be improved without changing the technical steps defined in the claims.

[0049] Embodiment 3 of this invention: In the deployment mode of integrated detection function of roadside unit, the data acquisition module is responsible for acquiring the baseband complex signal sequence, motion state vector, and road space reflection map sent by the vehicle node and the cloud platform; the manifold construction module generates an environmental Doppler constrained manifold matrix based on the motion state vector and the road space reflection map; the dictionary generation module performs orthogonalization operation on the environmental Doppler constrained manifold matrix to generate an environmental interference null feature dictionary matrix; the projection mapping module performs orthogonal projection operation on the baseband complex signal sequence using the environmental interference null feature dictionary matrix to obtain the projected residual signal; the feature extraction module parses the isolated state phase sequence from the projected residual signal and calculates the ratio of the projected residual energy of the projected residual signal to the baseband complex signal sequence; the state determination module generates an anomaly determination result based on the ratio of the isolated state phase sequence to the projected residual energy; and the instruction linkage module issues communication link adjustment control signaling based on the anomaly determination result. Each module can be executed by the same processor or by multiple functional circuits working together. The module division is used to correspond to the functional implementation relationship of the method steps and does not constitute a limitation on the hardware configuration.

[0050] This invention constructs a highway simulation test system containing 100 vehicle-mounted nodes. The highway simulation test system sets the vehicle speed to 120 km / h. Test personnel randomly injected 50 instances of RF hardware phase-locked loop (PLL) lock-up failures and 50 instances of multipath fast fading interference into the baseband complex signal sequence. The edge processor used both a traditional energy detection algorithm and the method provided in this invention to diagnose the anomaly sources. Test results show that the traditional energy detection algorithm has an accuracy rate of only 65% ​​in identifying RF hardware PLL lock-up failures. The traditional energy detection algorithm has a misclassification rate of up to 40% for multipath fast fading interference misidentified as hardware failure. The method provided in this invention achieves an accuracy rate of 98% in identifying RF hardware PLL lock-up failures. The method provided in this invention reduces the misclassification rate of multipath fast fading interference misidentified as hardware failure to 2%. Simulation experimental data demonstrates that the method provided in this invention can effectively distinguish between sudden hardware failures and transient environmental interference.

[0051] This invention utilizes prior knowledge of physical space geometry and mathematical matrices to construct an environmental Doppler constraint manifold matrix as a physical mapping mathematical model. First, an edge processor extracts the boundary 3D coordinates and surface material reflection coefficients from the road spatial reflection map. Then, combining the instantaneous 3D coordinates and speed of the vehicle contained in the motion state vector, a ray tracing algorithm is used to deduce the spatial reflection path arrival angles for both line-of-sight and non-line-of-sight reflection paths. Based on this, the edge processor performs an inner product operation on the radio frequency carrier frequency, the speed vector, and the aforementioned spatial reflection path arrival angles to derive the instantaneous Doppler frequency offset measurement value, which is then used as the spatial propagation Doppler frequency offset constraint condition.

[0052] Finally, the edge processor constructs an environmental Doppler-constrained manifold matrix based on the spatial propagation Doppler frequency offset constraint. The row dimension of the environmental Doppler-constrained manifold matrix corresponds to the length of the time sampling sequence of the baseband signal, while the column dimension corresponds to the number of effective reflection paths existing in physical space. Each column vector within the matrix, in complex exponential form, precisely represents the signal phase rotation law driven by a specific Doppler frequency shift. Ultimately, the edge processor performs singular value decomposition on the environmental Doppler-constrained manifold matrix to generate an environmental interference null feature dictionary matrix, providing a foundation for subsequent signal decoupling. In one embodiment, the data acquisition module outputs a time-aligned target baseband complex signal sequence, a target motion state vector, and a target road spatial reflection map to the manifold construction module; the manifold construction module outputs an environmental Doppler-constrained manifold matrix to the dictionary generation module; the dictionary generation module outputs an environmental interference null feature dictionary matrix to the projection mapping module; the projection mapping module outputs the projected residual signal to the feature extraction module; the feature extraction module outputs the ratio of the isolated state phase sequence to the projected residual energy to the state determination module; and the state determination module outputs the anomaly determination result to the instruction linkage module. The modules exchange processing results via a shared storage area or message queue. The message content includes at least a processing window identifier, a time stamp, and the corresponding processing result, thus ensuring that the modules in the system claims not only have a name correspondence but also an executable data flow relationship.

[0053] As can be seen from the above specific implementation methods, the vehicle-road-cloud cooperative communication detection method and system revolves around the inherent correlation between the baseband complex signal sequence, motion state vector, and road spatial reflection map. First, it establishes spatial propagation Doppler frequency offset constraints using the road spatial reflection map and motion state vector. Then, it suppresses the dominant environmental propagation component using the environmental Doppler constraint manifold matrix and the environmental interference null feature dictionary matrix. Next, it extracts the ratio of the isolated state phase sequence to the projected residual energy based on the projected residual signal. Finally, it outputs hardware fault or environmental interference judgment results based on the polynomial fitting results, phase variance values, and preset thresholds, and outputs corresponding communication link adjustment control signals. Therefore, the anomaly source identification results can directly enter specific technical processing links such as vehicle takeover warning, air interface retransmission, or coding redundancy adjustment, thereby reducing the probability of erroneous emergency responses caused by misjudgments and improving the operational stability and targeted handling of the vehicle-road-cloud cooperative communication link in highly dynamic topology scenarios.

[0054] In the application scenario of vehicle-road-cloud integrated collaborative safe transmission technology with high dynamic topology characteristics, vehicles travel at high speeds on the road in platoons or independently, and on-board nodes continuously send baseband signals containing sensing information to roadside equipment. Due to the high-speed spatial displacement of vehicles, the wireless transmission link inevitably introduces strong Doppler frequency shift, which, combined with multipath reflections from surrounding buildings, results in severe spectral broadening at the receiving end. The RF front-end components of the on-board nodes are prone to phase-locked loop (PLL) lockout and voltage-controlled oscillator (VCO) temperature drift under complex environmental changes, directly manifesting as nonlinear phase accumulation mixed into the baseband signal. When the frequency domain drift caused by Doppler frequency shift and the nonlinear phase distortion caused by components deeply overlap in the frequency domain, roadside equipment faces a dilemma in determining the source of the anomaly, unable to distinguish whether the signal distortion originates from a sudden severe electromagnetic reflection environment or an inherent hardware failure of the on-board node. To address the technical pain point of deep overlap between nonlinear phase accumulation and environmental Doppler frequency shift, the edge computing center initiates a data parsing pipeline at the underlying communication physical layer.

[0055] The system's stability is determined by strict temporal alignment of multi-source state information. The edge processor acquires the timing reference from the local timing module and simultaneously collects the baseband complex signal sequence and motion state vector representing the vehicle's transient position sent by the vehicle-mounted nodes. The cloud platform pre-generates and distributes a road space reflection map based on the digital twin base. The edge processor performs timestamp comparison operations on the baseband complex signal sequence, motion state vector, and road space reflection map according to the timing reference, filtering data records within the same transient time slice from the memory pool. The baseband complex signal sequence, motion state vector, and road space reflection map after the timestamp comparison operation are input into a logic register for data concatenation operations. After the operation is completed, a data matrix to be processed is generated. The edge processor parses the target baseband complex signal sequence, target motion state vector, and target road space reflection map with the same timestamp from the data matrix to be processed, identifying the target baseband complex signal sequence as a baseband complex signal sequence, the target motion state vector as a motion state vector, and the target road space reflection map as a road space reflection map.

[0056] Extracting legitimate Doppler features requires introducing geometric prior constraints in physical space as a computational benchmark. The edge processor extracts the boundary 3D coordinates and surface material reflection coefficients from the road spatial reflection map. Combining the instantaneous 3D coordinates and speed of the vehicle contained in the motion state vector, the edge processor uses ray tracing logic to calculate the spatial reflection path arrival angles and instantaneous Doppler frequency offset measurements for line-of-sight and non-line-of-sight paths. The instantaneous Doppler frequency offset measurements reflect the theoretically possible set of legitimate environmental frequency shift components under known geometric spatial relationships. The edge processor uses these instantaneous Doppler frequency offset measurements as the spatial propagation Doppler frequency offset constraint condition. Utilizing the spatial propagation Doppler frequency offset constraint condition, the edge processor constructs a matrix-dimensional environmental Doppler constraint manifold matrix based on antenna array manifold theory. The column vectors of the environmental Doppler constraint manifold matrix span the signal subspace where legitimate multipath interference resides. The edge processor performs singular value decomposition on the environmental Doppler-constrained manifold matrix, separating the main signal subspace representing the physical reflection law and the mutually orthogonal noise subspace. It then extracts the basis vectors located within the noise subspace to generate a noise subspace basis vector matrix. The edge processor uses this noise subspace basis vector matrix to construct a projection matrix, which is then used as the environmental interference null feature dictionary matrix.

[0057] The edge processor invokes its internal computing module to perform orthogonal dimensionality reduction and decoupling operations using the environmental interference null feature dictionary matrix. First, it converts the baseband complex signal sequence into a baseband complex signal vector. The edge processor then multiplies the baseband complex signal vector with the environmental interference null feature dictionary matrix. Since the energy distribution of the legitimate environmental Doppler shift is naturally constrained by the topological laws of physical space, signal components belonging to legitimate multipath refraction in the baseband complex signal vector are automatically filtered out under orthogonal projection. Finally, the edge processor extracts the projection components of the baseband complex signal vector within the subspace spanned by the environmental interference null feature dictionary matrix and determines these projection components as the projected residual signal.

[0058] The components retained in the projected residual signal exhibit anomalous evolution characteristics independent of the physical environment's frequency shift. The edge processor extracts complex sample points from the projected residual signal. For these complex sample points, it performs argument extraction calculations based on arctangent logic, outputting an original phase evolution sequence that records the phase trajectory. Due to trigonometric function truncation in digital signal processing, the original phase evolution sequence contains jumps that are integer multiples of pi. The edge processor performs phase dewinding operations on the original phase evolution sequence to compensate for these jumps and output a smooth, continuous phase sequence. The edge processor defines this continuous phase sequence as an isolated-state phase sequence and simultaneously performs calculations in the energy dimension, extracting the squared value of the complex modulus of the projected residual signal. This squared value is then used as the null space projection energy value. To provide a normalized reference, the edge processor calculates the squared value of the complex modulus of the baseband complex signal sequence without orthogonal projection processing, determining this squared value as the total received energy value across the entire domain. The bottom-level multiplier calculates the ratio of the projected energy value in the zero-disturbance space to the total received energy value in the entire domain, and determines the ratio as the projection residual energy ratio.

[0059] The state determination module acquires a preset alarm threshold and a preset discrete threshold, and compares the ratio of projected residual energy with the preset alarm threshold. For the purified isolated phase sequence, the state determination module performs a polynomial fitting operation to approximate the nonlinear evolution trajectory of the isolated phase sequence and extracts the quadratic coefficients generated by the polynomial fitting operation. The quadratic coefficients physically represent the monotonically increasing quadratic curve bending physical quantity caused by the loss of lock-in in the hardware phase-locked loop. When the ratio of projected residual energy is greater than the preset alarm threshold and the quadratic coefficient is greater than zero, the received signal contains a large amount of residual energy that cannot be explained by the spatial multipath environment and exhibits a distortion that monotonically increases over time. The state determination module generates an anomaly determination result of hardware fault. The diagnostic module synchronously calculates the phase variance value of the isolated phase sequence. When the ratio of projected residual energy is less than or equal to the preset alarm threshold and the phase variance value is greater than the preset discrete threshold, the overall energy of the signal is still controlled by the manifold constraint of the road reflection environment. The phase fluctuation mainly originates from the random fast fading disturbance caused by the reflecting building complex. The diagnostic module generates an anomaly determination result of environmental interference.

[0060] Based on the abnormal results determined at the hardware or environmental level, the edge processor triggers differentiated communication link adjustment mechanisms at its underlying layer. When the abnormality determination result is a hardware failure, the conventional link retransmission mechanism cannot recover the signal interruption caused by the RF front-end failure. The edge processor identifies the vehicle takeover warning signal as a communication link adjustment control signal and sends it to the vehicle node. After receiving the signal, the underlying processor of the vehicle node takes over the autonomous driving decision-making power, and the edge processor sends a hardware failure identifier to the cloud platform in parallel to update the global terminal health status record. When the abnormality determination result is environmental interference, the edge processor determines that the current channel has short-term random decay characteristics, identifies the air interface retransmission control signal or the coding redundancy adjustment control signal as a communication link adjustment control signal, and sends it to the vehicle node. After parsing the communication link adjustment control signal, the baseband chip of the vehicle node starts a data packet buffer retransmission operation or lowers the quadrature amplitude modulation mapping order.

[0061] The edge processor parses specific physical space parameters from the extracted boundary 3D coordinates and motion state vectors to perform underlying geometric mapping algebraic operations. The edge processor uses the vector inner product rule and ray tracing algorithm to calculate the spatial reflection path arrival angle and instantaneous Doppler frequency offset measurement. The specific spatial mapping algebraic physical equations are expressed as follows:

[0062] In the above space mapping algebraic physical equations, Representing the The instantaneous Doppler frequency offset measurement value corresponding to each reflection path. This represents the scalar value of the vehicle's velocity, which is extracted from the motion state vector. The carrier wavelength representing the radio frequency communication electromagnetic waves transmitted by the vehicle-mounted node. Representing the The spatial reflection path arrival angle corresponding to each reflection path. This represents the logic of cosine trigonometric function operations. In the formula, the operational domain is defined within a continuous three-dimensional Euclidean space. To ensure the uniformity of the physical dimensions of multiple sources, the system performs normalization processing on the vehicle speed scalar value and the carrier wavelength before performing the division operation, uniformly converting them to international standard units (meters / second and meters) to eliminate the order-of-magnitude error in frequency offset calculation caused by the difference in dimensions.

[0063] For all boundary 3D coordinates with effective surface material reflectance coefficients in the road spatial reflection map, the above spatial mapping algebraic physical equations are repeated, outputting a set containing all instantaneous Doppler frequency offset measurements. This set of instantaneous Doppler frequency offset measurements is defined as the spatial propagation Doppler frequency offset constraint condition. Based on the constructed spatial propagation Doppler frequency offset constraint condition, the edge processor converts the discrete frequency constraint parameters into a high-dimensional feature representation within the matrix manifold space. The specific algebraic equation for constructing the environmental Doppler constraint manifold matrix is ​​as follows:

[0064] In the specific algebraic equations for constructing the Doppler-constrained manifold matrix of the above-mentioned environment Represents the environmental Doppler-constrained manifold matrix. This represents the total number of valid spatial reflection paths that meet the physical space constraints. This represents the guide vector corresponding to the first spatial reflection path. This represents the guide vector corresponding to the second spatial reflection path. Representing the The guiding vector corresponding to each spatial reflection path. For any given spatial reflection path, the guiding vector... The expanded algebraic equation is:

[0065] In the expanded algebraic equation of the above guiding vector, Representing the The guide vector corresponding to the spatial reflection path, Represents the natural constant. Represents the imaginary unit of complex numbers. Represents the constant value of pi. Representing the The instantaneous Doppler frequency offset measurement value corresponding to each reflection path. The physical sampling time interval represents the baseband complex signal sequence. This represents the total number of complex sample points contained in the baseband complex signal sequence. This represents the algebraic operation of matrix transpose.

[0066] After constructing the environmental Doppler-constrained manifold matrix, the underlying edge processor calls a matrix factorization algorithm to perform singular value decomposition (SVD) on the environmental Doppler-constrained manifold matrix. The specific matrix factorization equation for the SVD operation is as follows:

[0067] In the specific matrix factorization equations of the singular value decomposition operation described above... Represents the environmental Doppler-constrained manifold matrix. This represents the left singular value orthogonal matrix produced by the decomposition. Represents a diagonal matrix containing all singular values. This represents the conjugate transpose of the right singular value orthogonal matrix produced by the decomposition. The edge processor extracts the left singular value orthogonal matrix. The middle corresponds to the diagonal matrix The set of column vectors containing zero or minimum values ​​is reorganized to construct a basis vector matrix for the noise subspace. Subsequently, the edge processor uses this noise subspace basis vector matrix to construct a projection matrix. The specific projection construction equation for generating the environmental interference null feature dictionary matrix is ​​as follows:

[0068] In the specific projection construction equation of the above-mentioned generation of the environmental disturbance null feature dictionary matrix, The dictionary matrix representing the zero-trap features caused by environmental disturbances. Represents the basis vector matrix of the noise subspace. This represents the conjugate transpose of the basis vector matrix of the noise subspace. As the environmental interference null feature dictionary matrix is ​​solidified, the edge processor initiates decoupling operations on the baseband complex signal sequence. The edge processor converts the time-domain arranged baseband complex signal sequence into a baseband complex signal vector, calls the hardware multiplier inside the field-programmable gate array to multiply the baseband complex signal vector with the environmental interference null feature dictionary matrix, extracts the projection component of the baseband complex signal vector within the subspace spanned by the environmental interference null feature dictionary matrix, and obtains the specific orthogonal projection algebraic equation of the projected residual signal:

[0069] In the specific orthogonal projection algebraic equations for obtaining the residual signal after projection, as described above... Represents the projected residual signal. The dictionary matrix representing the zero-trap features caused by environmental disturbances. This represents the baseband complex signal vector.

[0070] The projected residual signal is stored in a register as a complex vector. The feature extraction module extracts complex sample points from the projected residual signal and performs argument extraction calculations on the complex sample points to output the original phase evolution sequence. The algebraic equation for the argument extraction calculation is as follows:

[0071] In the algebraic equations for the above argument extraction calculation, Represents the first phase in the original phase evolution sequence The original phase value corresponding to each sampling time. This represents the logic for calculating the arctangent trigonometric function. Represents the first term in the projected residual signal. The complex sample points corresponding to each sampling time. Represents the first term in the projected residual signal. The imaginary part physical value of the complex sample points corresponding to each sampling time. Represents the first term in the projected residual signal. The real part physical values ​​of the complex sample points corresponding to each sampling time point are obtained. A phase dewinding operation is performed on the original phase evolution sequence, and a continuous phase sequence is spliced ​​to output it. This continuous phase sequence is then defined as an isolated-state phase sequence. The algebraic equation for the phase dewinding operation is:

[0072] In the algebraic equations for the phase unwinding operation described above, Represents the first isolated phase sequence The continuous phase values ​​corresponding to each sampling time. Represents the first phase in the original phase evolution sequence The original phase value corresponding to each sampling time. Represents the constant value of pi. This represents the dewinding integer compensation constant dynamically adjusted based on the phase transition direction of adjacent sampling points. The system uses a parallel scheduling multiplication-accumulation operator to extract and calculate the energy feature dimension, performing complex modulus-squared numerical accumulation on both the projected residual signal and the baseband complex signal sequence. The algebraic equations for calculating the null space projected energy value and the total received energy value over the entire domain are as follows:

[0073]

[0074] In the algebraic equations for calculating energy above, Represents the projected energy value of the zero-depression space. Represents the logic of cumulative summation algebraic operations. This represents the index number of the currently accumulated sample sequence. This represents the total number of complex sample points. Represents the first term in the projected residual signal. The squared value of the complex modulus of a complex sample point. Represents the total energy received across the entire region. This represents the first baseband complex signal in the baseband complex signal sequence before it is converted into a baseband complex signal vector. The squared value of the complex modulus of each complex sample point. The ratio of the projected energy value in the null space to the total received energy value over the entire area is calculated, and this ratio is defined by the algebraic equation for the ratio of the projected residual energy:

[0075] In the algebraic equations above used to calculate the numerical value of the ratio, Represents the ratio of projected residual energy. Represents the projected energy value of the zero-depression space. This represents the total energy received across the entire region.

[0076] The core mechanism of anomaly diagnosis relies on quadratic morphological curve approximation of the isolated-state phase sequence. The state determination module performs polynomial fitting operations on the isolated-state phase sequence to construct the mathematical fitting algebraic equation as follows:

[0077] In the above mathematical fitting algebraic equation, Represents the first isolated phase sequence The continuous phase values ​​corresponding to each sampling time. This represents the approximate fitting logic under the least squares rule. The coefficients of the quadratic term generated by the polynomial fitting operation. The sequence number representing the sampling time. Represents the physical sampling time interval. Represents the coefficient of the linear term. This represents the constant term bias. It extracts the coefficients of the quadratic term generated from the polynomial fitting operation. The control determination is completed by combining the ratio of projected residual energy with the preset alarm threshold. To address the random phase jitter characteristics caused by environmental interference, the state determination module further calculates the discrete fluctuations of the isolated phase sequence. The specific statistical algebraic equation for calculating the phase variance is as follows:

[0078] In the specific statistical algebraic equations for calculating the phase variance mentioned above... Represents the phase variance value. This represents the total number of complex sample points contained in the isolated-state phase sequence. Represents the logic of cumulative summation algebraic operations. This represents the index number of the currently accumulated sample sequence. Represents the first isolated phase sequence The continuous phase values ​​corresponding to each sampling time. Represents the isolated phase sequence in all The phase statistical average value within the sampling point range. The state determination module performs cross-condition judgment based on the extracted projection residual energy ratio, quadratic term coefficient, and phase variance value, and accurately outputs the anomaly determination result caused by environmental interference or hardware failure.

[0079] The baseband complex signal sequence transmitted by the vehicle-mounted node is the most primitive physical layer observation data of the underlying radio frequency communication link. Specifically, it includes in-phase and quadrature components obtained by an analog-to-digital converter discretizing the received electromagnetic waves at a preset sampling frequency (e.g., between 10MHz and 20MHz). The sampled data is encapsulated and aligned in complex floating-point format. The in-phase and quadrature components accurately record the amplitude attenuation and phase deflection trajectories during wireless transmission. The motion state vector is a data set representing the absolute motion and attitude changes of the vehicle in three-dimensional physical space. It is generated by the fusion calculation of the vehicle-mounted satellite positioning receiver and inertial measurement unit. Internally, it encapsulates the instantaneous three-dimensional spatial coordinates of the vehicle, including longitude, latitude, and altitude, and simultaneously carries the velocity vector recording the driving speed and direction, as well as the heading angle data. The edge processor concatenates and binds the baseband complex signal sequence with a unified timestamp, the motion state vector, and the data sent from the cloud at the memory address level to generate a structured data matrix to be processed, ensuring that subsequent multi-dimensional spatial matrix calculations have a strict time-aligned physical basis.

[0080] The road spatial reflection map distributed by the cloud platform forms a static digital twin foundation for the physical propagation environment. The data structure not only includes the boundary 3D coordinate point cloud of building facades, noise barriers, and large traffic signs along the route, but also maps the surface material reflection coefficients characterizing electromagnetic wave reflection attenuation. This encompasses the drastically different energy absorption and specular reflection attenuation parameters of concrete, metal components, and glass curtain walls for radio waves. The edge processor establishes a spatial geometric vector relationship based on the boundary 3D coordinates and the vehicle's instantaneous 3D spatial coordinates. Combined with the vehicle's own velocity vector, a ray tracing algorithm is used to deduce the spatial reflection path arrival angle for both line-of-sight direct paths and non-line-of-sight reflection paths. The edge processor performs an inner product operation on the radio frequency carrier frequency, the velocity vector, and the spatial reflection path arrival angle to derive the instantaneous Doppler frequency offset measurement value generated at the receiver for each effective propagation path. All sets of physical multipath frequency offsets conforming to geometric optics laws are then uniformly determined as the spatial propagation Doppler frequency offset constraint conditions.

[0081] After extracting the spatial propagation Doppler frequency offset constraints, the underlying computing center calls array signal processing algorithms to construct the environmental Doppler constraint manifold matrix. The row dimension of the environmental Doppler constraint manifold matrix corresponds to the length of the baseband signal's time sampling sequence, and the column dimension corresponds to the number of effective reflection paths in physical space. Each column vector within the matrix represents the signal phase rotation law driven by a specific Doppler frequency shift in complex exponential form, collectively spanning a signal subspace encompassing all legitimate environmental multipath interference. The edge processor performs singular value decomposition on the environmental Doppler constraint manifold matrix, reducing its dimension to a main signal subspace with large singular values ​​and a noise subspace with singular values ​​approaching zero. The edge processor extracts the set of mutually orthogonal column vectors within the noise subspace, combines them to generate a noise subspace basis vector matrix, and uses the multiplication operation between the noise subspace basis vector matrix and its conjugate transpose to generate a projection matrix. The projection matrix orthogonal to the legitimate environmental interference signal subspace is determined as the environmental interference null feature dictionary matrix.

[0082] Before algebraic operations, the baseband complex signal sequence is reconstructed into a column vector form. The edge processor schedules the underlying hardware multiplier to perform an orthogonal projection operation, multiplying the baseband complex signal vector with the environmental interference null feature dictionary matrix. Under the mathematical transformation of the orthogonal projection operation, the environmental Doppler frequency shift energy falling within the legal manifold subspace of the baseband complex signal vector is forcibly filtered out to zero. The edge processor extracts the residual projection component of the baseband complex signal vector within the orthogonal complement space spanned by the environmental interference null feature dictionary matrix, and identifies the projection component as the projected residual signal. By stripping away the environmental Doppler interference-masked projected residual signal, the potential abnormal distortion baseband characteristics of the RF hardware are directly exposed.

[0083] The projected residual signal is a complex signal sequence composed of discrete complex sample points. The edge processor performs an argument extraction calculation involving arctangent trigonometric function logic for each complex sample point, obtaining the original phase evolution sequence confined to the numerical range of negative to positive pi. Due to the truncation effect of the trigonometric function range, the original phase evolution sequence exhibits abrupt numerical changes when crossing the pi boundary, disrupting the continuous accumulation trajectory of the true physical phase. The edge processor performs phase dewinding operations on the original phase evolution sequence, monitoring the phase difference between adjacent sampling points and automatically adding or subtracting twice the pi value at abrupt changes to splice together a continuous phase sequence reflecting the true evolution trajectory. This continuous phase sequence is then defined as the isolated phase sequence.

[0084] To quantify the severity of abnormal signals, the underlying multiply-accumulate unit calculates the sum of squares of the real and imaginary parts of all complex sample points in the projected residual signal and the baseband complex signal sequence, respectively, generating the null space projected energy value representing the abnormal residual energy and the total received energy value representing the initial reception state. The underlying hardware calculates the ratio of the null space projected energy value to the total received energy value through a divider, and determines this ratio as the projected residual energy ratio, reflecting the proportion of abnormal energy that violates the physical reflection laws. The system is pre-configured with preset alarm thresholds and preset discrete thresholds for state determination. The preset alarm threshold is an energy baseline threshold calibrated based on the noise floor level under historical normal communication conditions and the equipment aging experience model, while the preset discrete threshold is a phase fluctuation tolerance lower limit set based on the statistical characteristics of fast fading channels under different road conditions.

[0085] The anomaly detection mechanism relies heavily on quantitative analysis of the mathematical morphology of the isolated-state phase sequence. The diagnostic module performs second-order or higher-order polynomial fitting operations on the isolated-state phase sequence to find a mathematical curve that approximates the current phase change trend and extracts the quadratic coefficients generated by the polynomial fitting operation. When the hardware RF phase-locked loop suddenly loses lock or the voltage-controlled oscillator experiences severe temperature drift, the resulting phase deviation exhibits a monotonically accelerating quadratic parabolic shape in the time domain. The quadratic coefficients with values ​​greater than zero are extracted, and combined with the excessive projection residual energy ratio, the detection module is triggered to lock in the hardware fault. For the detection of instantaneous environmental interference, the diagnostic module calculates the phase variance value of the isolated-state phase sequence deviating from the mean. Sudden complex multipath overlap and high-speed vehicle traffic cause high-frequency random phase jitter, which manifests as a large jump in the phase variance value. Based on the anomaly source location obtained from the calculation, the edge processor issues communication link adjustment control signaling at different protocol layers. For hardware failures, it triggers the vehicle's underlying drive-by-wire chassis control system to generate a vehicle takeover warning signaling. For environmental interference, it generates air interface retransmission control signaling at the media access control layer to reschedule data packet transmission or reduces the coding redundancy adjustment control signaling at the physical layer to reduce the modulation and coding strategy order.

[0086] The preset alarm threshold is set between 0.65 and 0.85. In open highway sections, the edge processor typically sets the preset alarm threshold to 0.75. When the projected residual energy ratio is greater than 0.75, the edge processor determines that the received signal contains a large amount of residual energy that cannot be explained by spatial multipath interference. The preset discrete threshold is set between 0.25 radians squared and 0.45 radians squared. In densely built-up urban areas, the edge processor typically sets the preset discrete threshold to 0.35 radians squared. When the phase variance of the isolated phase sequence is greater than 0.35 radians squared, the edge processor determines that the baseband complex signal sequence is subject to severe multipath interference. The inventors set specific preset values ​​to enable the edge processor to accurately distinguish between environmental interference and hardware faults.

Claims

1. A method for detecting vehicle-road-cloud cooperative communication, characterized in that, include: Step 1: Collect the baseband complex signal sequence and motion state vector sent by the vehicle node, and obtain the road space reflection map sent by the cloud platform; Step 2: Extract spatial propagation Doppler frequency offset constraints based on the motion state vector and the road spatial reflection map, and generate an environmental interference null feature dictionary matrix based on the spatial propagation Doppler frequency offset constraints; Step 3: Perform orthogonal projection operation on the baseband complex signal sequence using the environmental interference null feature dictionary matrix to obtain the projected residual signal; Step 4: Extract the isolated state phase sequence from the projected residual signal, and calculate the projection residual energy ratio based on the projected residual signal and the baseband complex signal sequence; Step 5: Generate an anomaly determination result based on the ratio of the isolated phase sequence to the projected residual energy. The anomaly determination result includes environmental interference or hardware failure. Output communication link adjustment control signaling based on the anomaly determination result.

2. The vehicle-road-cloud cooperative communication detection method according to claim 1, characterized in that, The step of extracting spatial propagation Doppler frequency offset constraints based on the motion state vector and the road spatial reflection map, and generating an environmental interference null feature dictionary matrix based on the spatial propagation Doppler frequency offset constraints, includes: Extract the boundary three-dimensional coordinates and surface material reflectance coefficients from the road space reflection map; Based on the three-dimensional coordinates of the boundary, the reflectivity of the surface material, and the motion state vector, calculate the spatial reflection path arrival angle and the instantaneous Doppler frequency offset measurement value; The instantaneous Doppler frequency offset measurement value is determined as the spatial propagation Doppler frequency offset constraint condition, and the environmental Doppler constraint manifold matrix is ​​constructed using the spatial propagation Doppler frequency offset constraint condition. Perform singular value decomposition on the environmental Doppler-constrained manifold matrix to extract the noise subspace basis vector matrix; A projection matrix is ​​constructed using the noise subspace basis vector matrix, and the projection matrix is ​​determined as the environmental interference null feature dictionary matrix.

3. The vehicle-road-cloud cooperative communication detection method according to claim 2, characterized in that, The step of performing orthogonal projection operation on the baseband complex signal sequence using the environmental interference null feature dictionary matrix to obtain the projected residual signal includes: Convert the baseband complex signal sequence into a baseband complex signal vector; The baseband complex signal vector is multiplied with the environmental interference null feature dictionary matrix to extract the projection component of the baseband complex signal vector in the subspace spanned by the environmental interference null feature dictionary matrix. The projection component is determined as the projection residual signal.

4. The vehicle-road-cloud cooperative communication detection method according to claim 3, characterized in that, Extracting the isolated-state phase sequence from the projected residual signal includes: Extract the complex sample points from the projected residual signal; The argument extraction calculation is performed on the complex sample points to output the original phase evolution sequence; Perform a phase unwinding operation on the original phase evolution sequence to output a continuous phase sequence; The continuous phase sequence is determined as the isolated phase sequence.

5. The vehicle-road-cloud cooperative communication detection method according to claim 4, characterized in that, The step of calculating the projection residual energy ratio based on the projected residual signal and the baseband complex signal sequence includes: Calculate the square value of the complex modulus of the projected residual signal, and determine the square value of the complex modulus of the projected residual signal as the null space projection energy value. Calculate the square value of the complex modulus length of the baseband complex signal sequence, and determine the square value of the complex modulus length of the baseband complex signal sequence as the total received energy value in the entire domain; Calculate the ratio of the zero-depression space projected energy value to the total received energy value over the entire area; The ratio value is determined as the projection residual energy ratio.

6. The vehicle-road-cloud cooperative communication detection method according to claim 5, characterized in that, The step of generating an anomaly determination result based on the ratio of the isolated state phase sequence to the projected residual energy includes: Obtain the preset alarm threshold; The projection residual energy ratio is compared with the preset alarm threshold. Perform a polynomial fitting operation on the isolated phase sequence and extract the coefficients of the quadratic term generated by the polynomial fitting operation; When the projection residual energy ratio is greater than the preset alarm threshold and the quadratic term coefficient is greater than 0, the abnormal judgment result is generated as a hardware fault.

7. The vehicle-road-cloud cooperative communication detection method according to claim 6, characterized in that, The step of generating an anomaly determination result based on the ratio of the isolated phase sequence to the projected residual energy further includes: Obtain the preset discrete threshold; Calculate the phase variance of the isolated phase sequence; When the projection residual energy ratio is less than or equal to the preset alarm threshold, and the phase variance value is greater than the preset discrete threshold, the abnormal judgment result is generated as environmental interference.

8. The vehicle-road-cloud cooperative communication detection method according to claim 7, characterized in that, The step of outputting communication link adjustment control signaling based on the anomaly determination result includes: When the anomaly determination result is a hardware failure, the vehicle takeover warning signal is identified as the communication link adjustment control signal and sent to the vehicle node, and a hardware failure identifier is sent to the cloud platform. When the anomaly determination result is environmental interference, the air interface retransmission control signaling or the coding redundancy adjustment control signaling is determined as the communication link adjustment control signaling and sent to the vehicle node.

9. The vehicle-road-cloud cooperative communication detection method according to claim 8, characterized in that, The acquisition of baseband complex signal sequences and motion state vectors sent by the vehicle-mounted nodes, and the obtaining of road spatial reflection maps from the cloud platform, includes: Obtain the time reference; Based on the timing reference, perform timestamp comparison calculations on the baseband complex signal sequence, the motion state vector, and the road spatial reflection map; The baseband complex signal sequence after performing the timestamp comparison operation, the motion state vector, and the road spatial reflection map are spliced ​​together to generate a data matrix to be processed. Parse the target baseband complex signal sequence with the same timestamp, the target motion state vector, and the target road spatial reflection map from the data matrix to be processed; The target baseband complex signal sequence is determined as the baseband complex signal sequence, the target motion state vector is determined as the motion state vector, and the target road spatial reflection map is determined as the road spatial reflection map.

10. A vehicle-road-cloud cooperative communication detection system, applied to the vehicle-road-cloud cooperative communication detection method as described in any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to acquire the baseband complex signal sequence, motion state vector, and road space reflection map sent by the vehicle node and the cloud platform. A manifold construction module is used to extract spatial propagation Doppler frequency offset constraints based on the motion state vector and the road spatial reflection map; The dictionary generation module is used to generate an environmental interference null feature dictionary matrix based on the spatial propagation Doppler frequency offset constraint condition. The projection mapping module is used to perform orthogonal projection operation on the baseband complex signal sequence using the environmental interference null feature dictionary matrix to obtain the projected residual signal; The feature extraction module is used to parse the isolated phase sequence from the projected residual signal and calculate the ratio of the projected residual energy of the projected residual signal to the baseband complex signal sequence. The state determination module is used to generate anomaly determination results based on the isolated state phase sequence and the ratio of projected residual energy. The instruction linkage module is used to issue communication link adjustment control signaling based on the anomaly determination result.

Images