A whole vehicle satellite communication multi-dimension cooperative detection system and method

Abstract

The application discloses a kind of whole vehicle satellite communication multidimension collaborative detection system and method, the system includes: core equipment cluster, using time synchronization mode, realize the time alignment of multi-source data;Three-layer detection unit, through high-speed industrial bus and central control module interconnection, according to uniform time acquisition multi-source data and upload to central control module, realize scene switching and data intercommunication;Four-dimensional collaborative evaluation unit, embedded in central control module, to carry out data integration and cross-scene consistency processing, output uniform quantitative evaluation result, and the result is fed back to central control module to realize test parameter adaptive adjustment or terminal design iteration suggestion generation, form closed loop;Visual and feedback unit: with central control module two-way interaction, for presenting detection data, evaluation index and abnormal diagnosis report.The application can be widely applied to intelligent networked vehicle satellite communication terminal research and development verification, regulation compliance certification and mass production quality monitoring.

Description

Technical Field

[0001] This invention relates to the field of intelligent connected vehicle communication detection technology, and in particular to a multi-dimensional collaborative detection system and method for whole vehicle satellite communication. Background Technology

[0002] As the intelligent connected vehicle (ICV) industry accelerates its evolution towards an integrated space-ground communication architecture, the deep integration of low-Earth orbit (LEO) satellite constellations, 5G non-terrestrial networks (NTN), and BeiDou short message technology has become a core technological support for vehicles to overcome the limitations of terrestrial network coverage, achieve global interconnection, emergency communication, and cross-border operations. Vehicle satellite communication systems must continuously ensure communication continuity, high positioning accuracy, and service transmission security under extreme conditions such as areas without terrestrial network coverage, complex electromagnetic interference environments, and dynamic vehicle obstruction. The technical complexity of this system far exceeds that of traditional terrestrial cellular communication systems.

[0003] However, existing satellite communication detection technologies for whole vehicles have many technical pain points that urgently need to be addressed, as follows: 1. Incomplete test scenario coverage and lack of dynamic realism: Traditional testing methods are centered on static laboratory verification, which can only simulate a single satellite signal or ideal channel environment. They cannot reproduce real operating conditions such as signal attenuation caused by the metal body and glass coating during vehicle operation, multipath propagation in urban canyons and tunnels, and dynamic beam switching. For example, when using a satellite signal simulator alone for positioning testing in existing technologies, the lack of consideration for the vehicle's electromagnetic environment and dynamic state leads to test results deviating from the actual operating state of the vehicle by more than 30%, making it difficult to support accurate verification of product performance.

[0004] 2. Limited Evaluation Dimensions and Lack of Collaborative Verification Mechanisms: Existing testing solutions often focus on single-link performance indicators (such as bit error rate and transmission latency), neglecting the collaborative evaluation of positioning and timing accuracy, user experience (QoE), and network security capabilities. For example, GB / T40764-2021 only specifies the interface and transmission integrity requirements for BeiDou short messages, without addressing quantitative evaluation of user experience; while ISO / SAE21434 clarifies network security requirements, it fails to establish a linkage verification logic with communication performance indicators, resulting in test results that cannot comprehensively reflect the overall service capabilities of the vehicle's satellite communication system.

[0005] 3. Insufficient equipment compatibility and low data synchronization accuracy: Test equipment from different constellations (LEO / GEO / RDSS) and different communication standards (5G NTN / DVB-S2X) mostly operate independently, lacking a unified collaborative control architecture. Furthermore, existing testing systems do not employ high-precision time synchronization technology, resulting in time deviations exceeding 100ns between GNSS positioning data, communication link data, and CAN bus data. This leads to logical inconsistencies during multi-source data fusion analysis, making it impossible to accurately establish a correlation analysis model between "positioning error and communication delay."

[0006] 4. Poor standard compatibility and high cross-platform verification costs: International standards (3GPP Rel-17 / 18, UNECE R144) and domestic standards (GB / T40764-2021, GJB7669-2012) differ in the definition of test indicators and the design of verification processes. For example, 3GPP Rel-18 requires the beam switching delay of LEO communication to be ≤50ms, while GJB7669-2012 stipulates that the round-trip delay of short messages is ≤2s. Existing testing systems are difficult to quickly adapt to the requirements of multiple standards, resulting in multiple rounds of testing for models sold in different regions, increasing verification costs by more than 60%.

[0007] 5. Experience quality is difficult to quantify, and subjective and objective indicators are disconnected: Existing detection relies solely on objective QoS indicators (such as latency and packet loss rate), which cannot accurately reflect the user's true perceived experience of voice calls and short message transmission and reception. For example, under the same transmission latency (1.5s), different voice compression algorithms may lead to significant differences in user experience, but existing technologies lack a quantitative evaluation model that correlates objective performance indicators with subjective experience (such as MOS score).

[0008] Therefore, there is an urgent need to build a vehicle satellite communication collaborative testing system that covers all scenarios, multiple dimensions, and multiple standards to systematically solve the above-mentioned technical pain points and provide technical support for the reliable application of intelligent connected vehicle satellite communication systems. Summary of the Invention

[0009] The purpose of this invention is to provide a multi-dimensional collaborative detection system and method for whole vehicle satellite communication, so as to solve the technical problem of how to achieve full-scene data acquisition, multi-dimensional collaborative evaluation and intelligent analysis and decision-making.

[0010] This invention is achieved using the following technical solution: a multi-dimensional collaborative detection system for whole-vehicle satellite communication, comprising a central control module, wherein the central control module serves as the core scheduling unit of the multi-dimensional collaborative detection system, and the multi-dimensional collaborative detection system further comprises: The core equipment cluster is bidirectionally connected to the central control module and uses a high-precision time synchronization method to achieve nanosecond-level time alignment of multi-source data across the entire system, providing a unified time base for the three-layer detection units. The three-layer detection unit is interconnected with the central control module via a high-speed industrial bus. It synchronously collects multi-source data based on the unified time base provided by the core equipment cluster and uploads it to the central control module. The three-layer detection unit includes a laboratory static verification subsystem, a road dynamic measurement subsystem, and a virtual simulation reproduction subsystem. The three subsystems are interconnected through the central control module to achieve scene switching, bidirectional data mapping, and parameter calibration. The four-dimensional collaborative evaluation unit, embedded in the central control module, performs parallel collaborative calculations on four dimensions—link performance, QoS / QoE, positioning and timing, and security and anti-interference—based on multi-source data aligned at the same time. It also establishes a correlation analysis model of positioning error, communication delay, beam switching status, and QoE experience. The data fusion processing unit is connected to the central control module. It receives multi-source data from the three-layer detection unit, completes data integration and cross-scenario consistency processing, and then inputs it into the four-dimensional collaborative evaluation unit to complete the quantitative evaluation. It outputs a unified quantitative evaluation result and feeds the result back to the central control module. The central control module adaptively adjusts the test parameters of the three-layer detection unit based on the result, or generates terminal design iteration suggestions, forming a closed loop of the entire process of "collection-evaluation-analysis-optimization-re-verification". Visualization and Feedback Unit: Interacts bidirectionally with the central control module to present detection data, evaluation indicators, correlation analysis results, and anomaly diagnosis reports.

[0011] Furthermore, the three-layer detection unit includes: The laboratory static verification subsystem, deployed in the OTA anechoic chamber, simulates multi-constellation signals, dynamic channel characteristics, and complex electromagnetic interference through multi-device linkage, conducts basic index tests, and collects core parameters to provide benchmark data for subsequent tests. The road dynamic measurement subsystem is used to conduct dynamic tests on a preset test route and simultaneously collect satellite signal strength, positioning data, beam switching status and communication interruption event parameters to verify the communication robustness in a real road environment. The virtual simulation reproduction subsystem is used to reproduce extreme working conditions and abnormal scenarios, conduct system redundancy and self-recovery capability tests, and support unlimited reproduction of test scenarios and iterative optimization of parameters. The three subsystems of the three-layer detection unit adopt a unified time base for synchronous acquisition, with an acquisition time deviation of ≤10ns and a sampling frequency of ≥1kHz. The three subsystems achieve bidirectional data mapping and calibration through a central control module: the benchmark parameters output by the laboratory static verification subsystem are automatically transmitted to the road dynamic measurement subsystem as a reference benchmark. The abnormal working condition data collected by the road dynamic measurement subsystem triggers the virtual simulation reproduction subsystem to perform 1:1 scene reproduction and parameter iteration. The optimized parameters of the virtual simulation reproduction subsystem are transmitted back to the laboratory static verification subsystem for verification, forming a cross-scene closed loop. The road dynamic measurement subsystem also includes a scene tag association module. The scene tag association module encodes the real-time collected road scene type information into scene tags, associates and models them with the synchronously collected link performance, positioning accuracy, and safety anti-interference indicators, outputs scene-level diagnostic conclusions, and synchronously feeds back the scene tags and abnormal event data to the virtual simulation reproduction subsystem.

[0012] Furthermore, the four-dimensional collaborative evaluation unit includes: The link performance evaluation subsystem is used to evaluate signal-to-noise ratio, bit error rate, short message transmission and reception success rate, beam switching success rate, and link interruption duration. The QoS / QoE collaborative evaluation subsystem adopts a QoS / QoE fusion evaluation model to achieve accurate quantitative mapping from objective QoS indicators to subjective experience quality. The positioning and timing evaluation subsystem is used to evaluate the positioning accuracy, first acquisition time, and time deviation of CEP95. The security anti-interference assessment subsystem is used to assess anti-interference capability thresholds, data transmission integrity, and tamper resistance.

[0013] Furthermore, the data fusion processing unit includes: The high-precision data synchronization subsystem adopts any one of the following protocols: PTP v2, GNSSDO, IEEE1588, or hard-triggered synchronization, to achieve time alignment of multi-source data. The intelligent data processing subsystem integrates an improved Kalman filter algorithm, an LSTM neural network anomaly detection model, and a Gaussian process regression deviation calibration model to achieve noise filtering, anomaly detection, and deviation correction for cross-scene data. The automated control subsystem automatically adjusts test parameters or generates terminal optimization suggestions based on the evaluation results, realizing a closed-loop iteration of test-analysis-optimization. The LSTM neural network anomaly recognition model of the intelligent data processing subsystem has a 64-dimensional input layer. The input features include time-series data of signal power, delay, packet loss rate, SNR, BER, and beam switching status, with a time step of 30 sampling periods. The hidden layer consists of 3 LSTM layers, each with 64 neurons, and a dropout rate of 0.2. The output layer outputs the anomaly type and confidence level. The anomaly judgment rules are as follows: power drift refers to a signal power change ≥ 0.5dB and a duration ≥ 5s; delay abrupt change refers to a delay change ≥ 50ms and a duration ≥ 3s; and link interruption refers to a signal power ≤ -140dBm and a duration ≥ 10ms.

[0014] Furthermore, it also includes an anomaly root cause inference module, which performs correlation analysis on power drift, time delay abrupt change, beam switching failure anomaly events with vehicle attitude, signal obstruction, and electromagnetic environment parameters, and outputs anomaly root cause diagnosis results.

[0015] Furthermore, the multi-dimensional collaborative detection system can be deployed in any of the following ways: Form 1: Vehicle-side data acquisition module + edge computing unit + cloud analysis platform. The vehicle-side data acquisition module collects multi-source data and uploads it to the edge computing unit for preprocessing. The edge computing unit then uploads the preprocessed data to the cloud analysis platform for fusion evaluation and closed-loop optimization. Form 2: Vehicle-side data acquisition module + factory server. The vehicle-side data acquisition module collects multi-source data and stores it locally, while simultaneously uploading it to the factory server for data fusion analysis, evaluation, and optimization suggestion generation. In Form 1, the vehicle-side acquisition module includes an integrated vehicle-mounted data acquisition terminal equipped with an FPGA chip, a GNSS high-precision positioning module, and an omnidirectional polarized satellite signal receiving antenna; the edge computing unit preprocesses the acquired data and uploads it to the cloud; in Form 2, the HIL hardware-in-the-loop test platform connects the vehicle ECU, communication module, and PSAP simulation platform via a CAN / LIN bus, with an event triggering delay ≤1ms.

[0016] A multi-dimensional collaborative detection method for whole vehicle satellite communication, based on the aforementioned multi-dimensional collaborative detection system for whole vehicle satellite communication, includes the following steps: S1: Parameter configuration phase, configure the appropriate standard system according to test requirements, and set the satellite constellation type, communication standard, test scenario parameters and evaluation threshold range; S2: Multi-scenario data acquisition phase, the three-layer detection unit is started simultaneously to collect multi-source data, including: link and radio frequency parameters in the controlled environment, dynamic communication data, and system response data in abnormal scenarios; S3: Multi-dimensional collaborative evaluation stage, which uses a four-dimensional collaborative evaluation unit to process the collected data in parallel, including: calculating the signal-to-noise ratio and beam switching success rate, analyzing positioning accuracy and time deviation, detecting anti-interference capability and data integrity, and outputting latency, packet loss rate and experience index. S4: Data fusion and analysis stage, using AI algorithms to identify abnormal data, calibrate cross-scenario test deviations, and generate visual analysis results; S5: Result output and optimization stage. The test report is displayed through the visualization and feedback unit. For non-compliant indicators, feedback is sent to the central control module, and test parameters are adjusted adaptively or terminal optimization suggestions are generated.

[0017] Furthermore, the test scenario parameters include vehicle trajectory, satellite orbit parameters, signal power, elevation angle distribution, and interference source type, and support multi-constellation signal combination configuration of GPS / GLONASS / Galileo / BeiDou / LEO.

[0018] Furthermore, the AI ​​algorithm includes an anomaly detection model and a deviation calibration model. The anomaly detection model automatically detects power drift and latency mutation problems based on an LSTM neural network. The anomaly judgment rules are as follows: power drift refers to a signal power change ≥ 0.5dB and a duration ≥ 5s; latency mutation refers to a latency change ≥ 50ms and a duration ≥ 3s; link interruption refers to a signal power ≤ -140dBm and a duration ≥ 10ms. The deviation calibration model is based on Gaussian process regression to correct for environmental differences between laboratory and road testing.

[0019] Furthermore, the analysis of positioning accuracy and time deviation specifically employs a three-level process to verify the accuracy of positioning and timing, including: QoS metric extraction is based on the PTP v2 timestamp, calculating the difference between the packet's sending and receiving times using the following formula: Where: D is the end-to-end delay; Tsend is the timestamp when the message is sent from the test device; Trecv is the timestamp when the vehicle communication module receives and acknowledges the message; The QoE model is used to calculate the QoE health checkup quality, employing an exponential function combination model. The formula is as follows: , where α, β, γ, and δ are regression weights calibrated based on multi-scenario test data; For packet loss / frame loss rate, MOS is the voice quality score; The rating mapping is based on the QoE (Quality of Life) assessment output rating. The mapping rules are as follows: QoE ≥ 85 points is excellent; 70-84 points is good; < 70 points is substandard.

[0020] Furthermore, step S4 includes the following sub-steps: Data preprocessing: Time synchronization alignment, noise filtering, and feature parameter alignment are performed on the data collected by the three-layer detection unit; An LSTM neural network model is used to achieve automatic anomaly detection; Gaussian process regression algorithm is used to correct cross-scenario testing bias; Generate visualizations of signal coverage heatmaps, link reliability curves, and QoE distribution maps.

[0021] The beneficial effects of this invention are as follows: Full-scenario coverage significantly improves test realism: Through a three-layer collaborative framework of "laboratory static verification - road dynamic testing - virtual simulation reproduction", it covers more than 95% of typical working conditions such as static, dynamic and abnormal conditions, solving the technical defects of traditional testing that "deviates from the actual operating environment of the vehicle". The test results deviate from the actual vehicle operating state by ≤5%.

[0022] Multi-dimensional collaboration significantly enhances the comprehensiveness of the assessment: The four-dimensional collaborative assessment unit realizes integrated testing of link performance, positioning accuracy, user experience, and network security, filling the gap of existing technology's "single-dimensional assessment" and comprehensively reflecting the integrated service capabilities of the vehicle's satellite communication system.

[0023] Multi-standard compatibility effectively reduces verification costs: The system supports multiple mainstream standards such as 3GPP, GB / T, GJB, and UNECE. Through parameterized configuration modules, it can quickly adapt to different testing requirements, reducing cross-platform verification costs by more than 60% and improving testing efficiency by 3 times.

[0024] Precise quantification of experience, high degree of consistency between subjective and objective: The innovative QoS / QoE integrated evaluation model transforms subjective experience into a quantifiable QoE index with an evaluation accuracy of ≥90%, effectively solving the industry pain point of "unmeasurable experience".

[0025] Automation and intelligence significantly improve iteration efficiency: scripted task scheduling, AI anomaly identification and closed-loop optimization mechanism reduce manual intervention by more than 80%, shorten problem location time to minutes, and strongly support the rapid R&D iteration of vehicle satellite communication terminals. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0027] Figure 1This is a system block diagram of the present invention; Figure 2 Flowchart for QoS / QoE fusion evaluation model calculation; Figure 3 This is a timing diagram of data interaction between the three-layer detection units; Figure 4 This is a flowchart of the detection method. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0029] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0030] The following detailed description of some embodiments of the present invention is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0031] See Figures 1 to 4 A multi-dimensional collaborative detection system for whole vehicle satellite communication, comprising: The central control module serves as the core scheduling unit of the multi-dimensional collaborative detection system. The core equipment cluster includes at least two of the following: a communication tester, a satellite signal simulator, a simulation platform, a short message closed-loop test system, and a HIL hardware-in-the-loop test platform. The equipment time synchronization unit is configured to use at least one of PTP v2, IEEE1588, GNSSDO, or hard-triggered synchronization methods to achieve time alignment of multi-source data. The three-layer detection unit mainly includes a laboratory static verification subsystem, a road dynamic measurement subsystem, and a virtual simulation reproduction subsystem. The three-layer subsystem is interconnected with the central control module through a high-speed industrial bus, collects multi-source data according to a unified time base and uploads it to the central control module to realize scene switching and data communication. The four-dimensional collaborative evaluation unit, embedded in the central control module, mainly includes a link performance evaluation subsystem, a QoS / QoE collaborative evaluation subsystem, a positioning and timing evaluation subsystem, and a security anti-interference evaluation subsystem. It performs collaborative calculations based on multi-source data aligned at the same time. A data fusion processing unit, which is in communication connection with the central control module, receives multi-source data from the three-layer detection subsystem, performs data integration and cross-scene consistency processing, outputs unified quantitative assessment results, and feeds the results back to the central control module to realize adaptive adjustment of test parameters or generation of terminal design iteration suggestions, forming a closed loop; Visualization and feedback unit: which performs two-way interaction with the central control module and is used to present detection data, assessment indicators and abnormal diagnosis reports.

[0032] In this embodiment, the laboratory static verification subsystem is deployed in an OTA anechoic chamber, wherein the electromagnetic shielding efficiency of the OTA anechoic chamber is ≥90dB, the reflection loss is ≥15dB@800MHz-6GHz, and the EIRP measurement accuracy is ±1dB; the subsystem constructs a controllable test environment jointly by a communication tester, a satellite signal simulator and a short message closed-loop test system, simulates multi-constellation signals, dynamic channel characteristics and electromagnetic interference, and collects core indicators such as effective isotropic radiated power (EIRP), bit error rate (BER), and short message transceiving success rate. The core device configuration includes: communication tester: supports multi-standard communication protocols of 2G-5G NR, Wi-Fi 6, Bluetooth 5.2, has a built-in PSAP (Public Safety Answering Point) simulation function, can generate and decode eCall Minimum Set of Data (MSD), the test range of voice quality MOS is 1-5 points, and the measurement accuracy of bit error rate (BER) is ≤10 -6 ; satellite signal simulator: supports combined output of multi-constellation signals of GPS / GLONASS / Galileo / BeiDou / LEO, orbital parameters can be flexibly configured (altitude 500-2000km, inclination 0-90°), the signal power adjustment range is -170dBm to -100dBm, and the test resolution of positioning accuracy (CEP95) is 0.1m; channel simulation platform: the frequency bandwidth supports 10MHz-100MHz, can accurately simulate Doppler frequency shift (±10kHz), multipath fading (Rayleigh / Rician channel model), interference injection (narrowband / co-channel / adjoining channel interference), and the path fading modeling accuracy is ±0.5dB; Beidou short message closed-loop test system: supports S1 / S2 system, the message length can be configured (1-140 bytes), the round-trip delay measurement accuracy is ±10ms, and the bit error rate is ≤10 -5 , and has the functions of message transceiving receipt and transmission integrity check. The core functions include: simulating multi-constellation signals, dynamic channel characteristics and complex electromagnetic interference through multi-device linkage, carrying out basic indicator tests such as link stability, radio frequency consistency, short message transceiving and voice communication, collecting core parameters such as effective isotropic radiated power (EIRP), total isotropic sensitivity (TIS) and beam switching success rate, and providing benchmark data for subsequent tests.

[0033] In this embodiment, the hardware components of the road dynamic measurement subsystem include: an onboard integrated data acquisition terminal: equipped with a high-performance FPGA chip (sampling rate ≥ 1 GHz), a GNSS high-precision positioning module (positioning accuracy ≤ 0.5 m, data update rate 10 Hz), and an omnidirectional polarized satellite signal receiving antenna (gain ≥ 15 dBi), supporting local large-capacity storage (capacity ≥ 1 TB) and real-time cloud upload (5G / NB-IoT dual-mode communication); and a scene perception module: integrating millimeter-wave radar (detection distance 0-200 m) and a high-definition camera (resolution 1080 P), to collect real-time road scene types (such as tunnels, urban canyons, and open areas) and environmental parameters. Its core functions include: conducting dynamic tests on preset test routes (covering typical scenarios such as open areas, mountains, tunnels, and urban canyons), with vehicle speeds ranging from 30-120 km / h, simultaneously collecting parameters such as satellite signal strength, positioning data, beam switching status, and communication interruption events to verify the communication robustness in real road environments.

[0034] In some embodiments, the road dynamic measurement subsystem also includes a scene label association module, which encodes the scene information output by the scene perception module into scene labels, associates them with link performance, positioning accuracy, and security anti-interference indicators to form a model, and outputs scene-level diagnostic conclusions.

[0035] In this embodiment, the virtual simulation reproduction subsystem includes a HIL hardware-in-the-loop test platform and a digital twin mapping module. The HIL platform connects the vehicle ECU, communication module, and PSAP simulation platform via a CAN / LIN bus, injecting abnormal events such as signal interruption and power fluctuation, with an event triggering delay of ≤1ms. The digital twin mapping module constructs a bidirectional dynamic model of "virtual vehicle-satellite channel" to realize the reproduction of test scenarios and iterative optimization of parameters. Specifically, the core components of the virtual simulation reproduction subsystem include: The HIL (Hardware-in-the-Loop) test platform: based on a CAN / LIN bus (transmission rate 500kbps-1Mbps), it achieves closed-loop connection between the vehicle ECU, satellite communication module, and PSAP simulation platform, supporting the injection of abnormal events such as signal interruption, power fluctuations (±10% voltage change), and module failures, with an event triggering delay ≤1ms; The digital twin mapping module: constructs a bidirectional dynamic mapping model of "virtual vehicle-satellite channel," where the virtual vehicle model integrates dynamic characteristics (acceleration / braking / steering) and electromagnetic shielding effects, and the satellite channel model is based on the ITU-R M.2101 standard, supporting accurate simulation of Doppler migration, rain attenuation, and shading effects. Its core functions include: reproducing extreme operating conditions and abnormal scenarios, conducting system redundancy and self-recovery capability tests, supporting unlimited reproduction of test scenarios and iterative parameter optimization, and significantly reducing the cost and risk of real vehicle testing.

[0036] In this embodiment, the core evaluation metrics of the link performance evaluation subsystem include: signal-to-noise ratio (SNR), bit error rate (BER), short message transmission and reception success rate, beam switching success rate, and link downtime. The key calculation method includes: SNR = 10log 10 (P signal / P noise ), where P signal The target satellite signal power (precisely measured using a communication tester, with an accuracy of ±0.1 dBm), P noise Equivalent noise power (calibrated via channel simulation platform, range -174dBm / Hz to 10dBm / Hz); Beam switching success rate = (number of successful switches / total number of switches) × 100%, successful switch is defined as link interruption duration ≤ 50ms during the switch; Standard adaptation basis: 3GPP Rel-18 (beam switching delay ≤ 50ms), GB / T40764-2021 (short message transmission and reception success rate ≥ 99%). In this embodiment, the QoS / QoE collaborative evaluation subsystem establishes a quantitative mapping relationship between objective QoS indicators and subjective experience indicators, outputting a QoE index and corresponding level evaluation. The quantitative mapping relationship can be any one of exponential function combinations, linear combinations, or machine learning models. Specifically, the following QoS / QoE fusion evaluation model is used (as a preferred implementation): Where α, β, γ, and δ are regression weights calibrated based on 1000 sets of multi-scenario test data (α=0.35, β=0.002, γ=0.4, δ=0.25). This model can achieve accurate quantitative mapping from objective QoS indicators to subjective experience quality. For packet loss / frame loss rate, MOS is the voice quality score.

[0037] In this embodiment, the core evaluation indicators of the positioning and timing evaluation subsystem are: CEP95 positioning accuracy, time to first acquisition (TTFF), and time deviation. Standardized testing methods: CEP95 positioning accuracy: 1000 valid positioning points are collected, and the circular probability error at 95% confidence level is calculated, requiring ≤2m (compliant with the BeiDou-3 Open Service Performance Specification); Time to First Acquisition (TTFF): The time from receiving satellite signals to completing positioning in a cold start state, requiring ≤30s; Time deviation: Measured using a high-precision time synchronization protocol, requiring ≤±20ns; Standard adaptation basis: BeiDou-3 Open Service Performance Specification and GNSS international standards.

[0038] In this embodiment, the security anti-interference assessment subsystem injects multi-constellation interference signals through a channel simulation platform to test the tamper resistance, data integrity, and system self-recovery capability of the communication module, meeting the testing requirements for network security and electromagnetic compatibility. The channel simulation platform supports interference injection and path fading modeling with a bandwidth ≤100MHz. Specifically, the core evaluation indicators of the security anti-interference assessment subsystem are: anti-interference capability threshold, data transmission integrity, and tamper resistance. The systematic testing method includes: anti-interference capability: injecting narrowband interference (power -100dBm to -80dBm) and co-channel interference (frequency difference between interference signal and satellite signal ≤1MHz) through the platform, and measuring the interference threshold when SNR attenuation ≤3dB; data integrity: using the SHA-256 hash verification algorithm to verify the consistency of data before and after short message transmission, with an error rate ≤10%. -6 Anti-tampering: Simulated data tampering attack (modification of message field content), the abnormal alarm response time of the communication module is ≤100ms; Standard adaptation basis: UN ECE R10 (electromagnetic compatibility), ISO / SAE21434 (network security).

[0039] In this embodiment, the data fusion processing unit includes: a high-precision data synchronization subsystem, an intelligent data processing subsystem, and an automated control subsystem.

[0040] Specifically, the high-precision data synchronization subsystem can use any one of the following protocols: PTP (Precision Time Protocol) v2, GNSSDO, IEEE1588, or hard-triggered synchronization, to achieve time alignment of multi-source data.

[0041] The intelligent data processing subsystem integrates filtering algorithms, feature extraction algorithms, AI anomaly detection, and deviation calibration algorithms to achieve accurate data analysis. Specifically, a distributed acquisition architecture can be adopted: data from each subsystem is preprocessed in real-time by an FPGA module, and an improved Kalman filter algorithm is used to eliminate noise interference (filter coefficient 0.01-0.1) to accurately extract characteristic parameters such as signal amplitude, phase, and time delay. AI anomaly detection can be performed using AI algorithms, such as LSTM neural networks and convolutional neural networks, while deviation calibration models can employ algorithms such as Gaussian process regression and support vector machines. The specific algorithm selection can be flexibly configured according to the test scenario and accuracy requirements. This may include: Anomaly Detection Model: Based on an LSTM neural network, it automatically identifies anomalies such as power drift (≥0.5dB), latency spikes (≥50ms), and link interruptions by inputting time-series data such as power, latency, and packet loss rate, with an accuracy rate of ≥95%. Deviation Calibration Model: Employing a Gaussian process regression algorithm, it corrects for test deviations caused by environmental differences based on the correlation of multi-source data (e.g., the correlation between signal strength in laboratory and road scenarios), achieving a data error of ≤5% after calibration.

[0042] The automated control subsystem automatically adjusts test parameters or generates terminal optimization suggestions based on evaluation results, achieving a closed-loop iteration of test-analysis-optimization. Specifically, it supports scripting languages, has built-in standard test templates (compatible with mainstream standards such as 3GPP, GB / T, and GJB), can simultaneously schedule up to 10 parallel test tasks, and automatically adjusts test parameters based on evaluation results (such as increasing interference intensity or modifying satellite orbit parameters) or generates terminal optimization suggestions (such as adjusting antenna installation angle or optimizing algorithm parameters), achieving a closed-loop iteration of test-analysis-optimization.

[0043] In this embodiment, the visualization and feedback unit can output data in various formats, including: KPI data dashboards, link reliability curves, signal coverage heatmaps, QoE distribution maps, and anomaly diagnosis reports. In addition, it also supports interactive functions, including: multi-dimensional data comparison (such as QoE index comparison in different scenarios), historical test data tracing, and export of test reports (PDF / Excel format).

[0044] Furthermore, the device control interface in this embodiment adopts a unified instruction set and script task orchestration mechanism, supporting multi-device collaborative scheduling and unified parameter configuration to achieve automated execution of the testing process. The data model structure employs a hierarchical design, including a base layer (timestamps, scene tags, device identifiers), a raw data layer (raw waveforms, raw acquired data), a feature layer (signal amplitude, phase, delay, and other feature parameters), and an indicator layer (link performance, positioning accuracy, QoE index, and other quantitative indicators). The report structure is standardized, including compliance conclusions, detailed indicator compliance status, test evidence chain data, and traceable data indexes, ensuring the verifiability and traceability of the test results.

[0045] In some embodiments, the data fusion processing unit is also used to perform time alignment, feature alignment, and deviation calibration on laboratory, road, and virtual simulation data to generate a comparable unified index set and achieve cross-scenario data consistency. The time synchronization method is PTP v2, and the acquisition time deviation is ≤10ns. Feature alignment includes a unified statistical window / resampling strategy for latency, power, and packet loss rate.

[0046] In some embodiments, an anomaly root cause inference module may be added. The anomaly root cause inference module will perform correlation analysis with abnormal events such as power drift, time delay abrupt change, and beam switching failure, and output anomaly root cause diagnosis results.

[0047] The vehicle satellite communication multi-dimensional collaborative detection system can be deployed in any of the following configurations: Configuration 1: Vehicle-side data acquisition module + edge computing unit + cloud analysis platform. The vehicle-side data acquisition module collects multi-source data and uploads it to the edge computing unit for preprocessing. The edge computing unit then uploads the preprocessed data to the cloud analysis platform for fusion evaluation and closed-loop optimization. Configuration 2: Vehicle-side data acquisition module + factory server. The vehicle-side data acquisition module collects multi-source data and stores it locally, while simultaneously uploading it to the factory server for data fusion analysis, evaluation, and optimization suggestion generation.

[0048] This invention also provides a multi-dimensional collaborative detection method for whole-vehicle satellite communication, which includes the following steps: S1: Parameter configuration phase, configure the appropriate standard system according to test requirements, and set the satellite constellation type, communication standard, test scenario parameters and evaluation threshold range; S2: Multi-scenario data acquisition phase, the three-layer detection unit is started simultaneously to collect multi-source data, including: link and radio frequency parameters in the controlled environment, dynamic communication data, and system response data in abnormal scenarios; S3: Multi-dimensional collaborative evaluation stage, which uses a four-dimensional collaborative evaluation unit to process the collected data in parallel, including: calculating the signal-to-noise ratio and beam switching success rate, analyzing positioning accuracy and time deviation, detecting anti-interference capability and data integrity, and outputting latency, packet loss rate and experience index. S4: Data fusion and analysis stage, using AI algorithms to identify abnormal data, calibrate cross-scenario test deviations, and generate visual analysis results; S5: Result output and optimization stage. The test report is displayed through the visualization and feedback unit. For non-compliant indicators, feedback is sent to the central control module, and test parameters are adjusted adaptively or terminal optimization suggestions are generated.

[0049] In this embodiment, step S1: parameter configuration stage (preferably taking ≤10 minutes) specifically includes: Standard compatibility selection: Select the appropriate standard system based on the test objectives. Preferred standard systems include, but are not limited to, 3GPP Rel-17 / 18, GB / T40764-2021, GJB7669-2012, UN ECE R144, etc. Fine-tuning of scene parameters: Satellite parameters: constellation type (LEO / GEO / RDSS), orbital altitude (500-2000km), signal power (-170dBm to -100dBm), elevation angle distribution (0°-90°); Test scenarios: laboratory scenario (obstruction rate 0%-50%, interference type and intensity), road scenario (test route, driving speed range), virtual scenario (abnormal event type, trigger time and duration); Evaluation threshold settings: Preferred evaluation thresholds include short message round-trip delay ≤ 2s, CEP95 positioning accuracy ≤ 2m, voice quality MOS ≥ 4, and QoE index ≥ 85.

[0050] In this embodiment, step S2: the multi-scenario data acquisition stage (preferably with a time consumption of ≤2h for laboratory data acquisition, ≤4h for road data acquisition, and ≤1h for virtual simulation data acquisition) specifically includes: Laboratory subsystem startup: After the OTA anechoic chamber environmental parameters are calibrated, the communication tester generates an eCall MSD dataset, the satellite signal simulator outputs a multi-constellation combination signal, the simulation platform simulates dynamic channels and interference signals, the RDSS closed-loop system performs 1000 short message transmission and reception tests, and simultaneously collects data such as EIRP, TIS, BER, and beam switching status at a sampling frequency of 1kHz. The road subsystem is activated: the vehicle travels along the preset route, the on-board data acquisition terminal synchronously records the satellite signal strength, positioning data, and communication interruption events, the environmental perception module collects scene information in real time, and the data is uploaded to the cloud database in real time through dual-mode communication; Virtual simulation subsystem startup: The HIL platform injects abnormal events according to the preset scheme (such as 3 signal interruptions, each lasting 1 second). The digital twin module synchronously simulates the dynamic changes of the satellite channel and collects system response data and fault recovery time.

[0051] In this embodiment, step S3: multi-dimensional collaborative evaluation stage (preferably taking ≤30 minutes) specifically includes: Link performance evaluation: Calculates metrics such as SNR, BER, and short message transmission / reception success rate to determine if they meet preset standard thresholds. The core process revolves around "signal power, noise separation, metric calculation, and threshold determination," specifically including: Data preprocessing: Bandpass filtering is performed on the acquired raw signal to remove out-of-band interference and extract the time-domain waveform of the target signal; Key performance indicator calculation: Signal-to-noise ratio (SNR): Calculated using the power spectral density method, as shown in the following formula: ;wherein: P signalThe target satellite signal power (unit: dBm) is measured directly using the power detection module of the communication tester, with a measurement accuracy of ±0.1 dBm; P noise The equivalent noise power (unit: dBm) is obtained by measuring the average noise power within the same bandwidth when there is no satellite signal input. The measurement time is ≥1s and the statistical step size is 10ms.

[0052] Bit Error Rate (BER): For digitally modulated signals (such as QPSK, 16QAM), it is the ratio of the number of erroneous bits in the received data to the total number of bits. The formula is as follows: , where: N error The number of error bits is obtained through bit-level comparison after frame synchronization; N total The total number of bits transmitted, with a value ≥ 1 × 102 6 Bits are used to ensure statistical significance.

[0053] Short message transmission and reception success rate: The ratio of the number of valid short messages received to the number of messages sent, calculated using the following formula: Wherein: N send N represents the total number of short messages sent. recv_valid The number of messages successfully received and whose data integrity verification passes is recorded. The integrity verification uses the CRC-32 algorithm.

[0054] Beam handover success rate: The ratio of the number of times the link was not interrupted during beam handover to the total number of handovers, as shown in the following formula: , where: N beam_total The total number of beam switching operations is ≥100; N beam_valid This refers to the number of times the link is interrupted for a duration of ≤50ms during the handover process. The interruption is determined based on a signal power of ≤-140dBm and a duration of ≥10ms.

[0055] Threshold determination: Compare the calculation results with preset thresholds (e.g., SNR≥12dB, BER≤1×10). -6 R success Compare with ≥99% and output the status of each individual indicator meeting the standard.

[0056] Positioning and Timing Evaluation: CEP95 positioning accuracy, TTFF, and time deviation are statistically analyzed to verify positioning and timing accuracy. A three-level process is employed: "QoS indicator extraction, QoE model calculation, and QoS level mapping," specifically including: QoS indicator extraction: End-to-end delay (D): Based on the PTP v2 timestamp, calculate the difference between the message sending time and the receiving time, using the following formula: Where: Tsend is the timestamp of the message being sent from the test device; Trecv is the timestamp of the vehicle communication module receiving and acknowledging the message, with a timestamp accuracy of ≤1ns.

[0057] Packet loss rate (P) loss The ratio of lost packets to total sent packets is calculated using the following formula: , where: N recv The number of successfully received messages, including messages that were successfully retransmitted.

[0058] Voice Quality (MOS): Adopted according to ITU-T P.863 standard, it is measured by objective instruments and outputs a quantitative score of 1-5. The higher the score, the better the voice quality.

[0059] QoE fusion model calculation: This invention preferably employs an exponential function combination model, as shown in the following formula: Where: α=0.35, β=0.002, γ=0.4, δ=0.25 are regression weights, obtained through 1000 sets of multi-scenario (laboratory / road / virtual) test data, calibrated using the least squares method, with a goodness of fit R^2≥0.92; D is in ms, with a value range of 0-1000ms; Ploss ranges from 0 to 1; MOS ranges from 1 to 5; QoE index ranges from 0 to 100, and the calculation results are rounded to one decimal place.

[0060] Rating Mapping: The rating is output based on the QoE index. The mapping rule is: QoE ≥ 85 is excellent, 70-84 is good, and < 70 is unacceptable.

[0061] Security anti-interference assessment: Analyzes anti-interference thresholds, data integrity verification results, and anti-tampering response time to comprehensively evaluate security protection capabilities. A closed-loop process of "interference injection, performance monitoring, and indicator calculation" is adopted, specifically including: Anti-interference capability test: Interference injection: Through the channel simulation platform, narrowband interference (center frequency and satellite signal frequency difference ≤ 1MHz) or co-channel interference (frequency difference = 0) is injected, and the interference power is gradually increased from -100dBm to -80dBm in 5dBm increments; Performance monitoring: Under each interference power, the SNR change is continuously monitored, and the interference power (Pjamming) when the SNR attenuation is ≤ 3dB is recorded, which is defined as the anti-interference threshold.

[0062] Data integrity testing: Raw data hash value calculation: For the sent short message data (length 1-140 bytes), the hash value H is calculated using the SHA-256 algorithm. send Received data hash value calculation: For received short message data, the same algorithm is used to calculate the hash value H. recv Integrity determination: If H send =H recvIf the data is complete, it is considered complete; otherwise, it is considered incomplete. Calculate the data integrity error rate: ,Require .

[0063] Anti-tampering test: Tampering simulation: Injecting tampered messages into the transmission link (modifying message fields ≥ 1 byte); Response monitoring: Recording the abnormal alarm response time T of the vehicle communication module. response T is required response ≤100ms.

[0064] In this embodiment, step S4: the data fusion and analysis stage (preferably taking ≤20 minutes) has "data preprocessing, anomaly identification, deviation calibration, and visualization output" as its core process, and the specific implementation process is as follows: Data preprocessing: Time synchronization alignment, noise filtering, and feature parameter alignment are performed on the data collected by the three-layer subsystem; Time synchronization alignment: Based on PTP v2 timestamps, the collected data from the laboratory, roads, and virtual subsystems are mapped to a unified time axis with a time synchronization error of ≤10ns. Linear interpolation was used to fill in missing data (missing rate ≤ 5%), with an interpolation error ≤ 5%. Noise filtering: An improved Kalman filter algorithm is used to filter noisy data. The algorithm formula is as follows: State prediction equation: ; Prediction error covariance equation: ; Kalman gain equation: ; State update equation: ; Error covariance update equation: ; in: The filtered state value at time k; This represents the original collected value at time k; A=1 (state transition matrix), B=0 (control matrix), H=1 (observation matrix); Q=0.01 (process noise covariance), R=0.1 (observation noise covariance), determined through offline calibration.

[0065] Feature alignment: Unified statistical window: For indicators such as latency, power, and packet loss rate, a 1-second statistical window is used for resampling to ensure the comparability of data across scenarios; Feature standardization: Normalizes indicators of different dimensions to the [0,1] interval, as shown in the following formula: Where: x is the original eigenvalue; x min x max These are the historical minimum and maximum values ​​for this feature.

[0066] Intelligent anomaly detection: An automatic anomaly detection is achieved using an LSTM neural network model, specifically including: Model structure: Input layer: Dimension is 64. Input features include time series data of power, latency, packet loss rate, SNR, BER, and beam switching status. Time step = 30. Hidden layers: 3 LSTM layers, 64 neurons per layer, dropout rate = 0.2 to prevent overfitting; Output layer: Fully connected layer, outputs anomaly types (power drift / delay mutation / link interruption / packet loss / beam switching failure) and confidence levels; Anomaly detection rules: Power drift: Signal power change ≥ 0.5dB and duration ≥ 5s, confidence level ≥ 95%; Delay mutation: Delay change ≥ 50ms and duration ≥ 3s, confidence level ≥ 95%; Link interruption: Signal power ≤ -140dBm and duration ≥ 10ms, confidence level ≥ 99%; Output results: Generate an abnormal event statistics table, including the abnormal type, occurrence time, duration, confidence level, and associated scenario tags.

[0067] Deviation calibration optimization: A Gaussian process regression algorithm is used to correct cross-scenario test deviations, specifically including: Deviation Modeling: Using laboratory data as a benchmark (deviation = 0), we construct deviation prediction models for road scenes and virtual scenes. The input features are scene labels and environmental parameters, and the output is the deviation value (Δx) of each indicator. Calibration formula: , where: x calibrated The raw measurements for the road / virtual scene; x measured These are the calibrated, unified index values.

[0068] Calibration accuracy: The relative deviation of cross-scenario data after calibration is ≤5%, ensuring data consistency.

[0069] Visualization output: Generates visualization results such as signal coverage heatmap, link reliability curve, and QoE distribution map.

[0070] Signal coverage heatmap: Using latitude and longitude as coordinates, the color intensity represents the SNR magnitude: red (SNR≥18dB), yellow (12dB≤SNR<18dB), and blue (SNR<12dB). Link reliability curve: with test time on the horizontal axis and SNR on the vertical axis, BER and packet loss rate curves are superimposed, and the location of abnormal events is marked. QoE Distribution Map: With scene type as the horizontal axis and QoE index as the vertical axis, it shows the distribution of experience in different scenes; Abnormal diagnostic report: includes the type of abnormality, preliminary root cause analysis, and scope of impact.

[0071] In this embodiment, step S5: the result output and optimization stage (preferably taking ≤10 minutes) specifically includes: Test report generation: Outputs test reports that meet the target standards, clearly indicating whether compliance verification has been passed, and listing in detail the scores and compliance status of each dimension indicator; Optimization suggestion feedback: For non-compliant indicators, conduct in-depth analysis of the root causes and generate terminal optimization suggestions; Closed-loop iterative testing: Adjust test parameters or terminal design scheme according to optimization suggestions, repeat steps S1-S4 until all indicators meet the standards.

[0072] Taking the testing of the "LEO + Beidou RDSS" integrated communication system of a certain intelligent connected vehicle as an example, the target standards are 3GPP Rel-18 and GB / T40764-2021. The specific implementation process is as follows: S1: Parameter Configuration Standard selection: 3GPP Rel-18 (LEO communication performance verification) + GB / T40764-2021 (BeiDou RDSS short message function verification); Satellite parameters: LEO orbital altitude 550km, BeiDou RDSS S1 system, signal power -150dBm; Test scenarios: Laboratory: 30% obstruction rate, injected narrowband interference (power -95dBm); Road: Urban canyon + tunnel route, driving speed 30-60km / h; Virtual: 2 signal interruptions (1.5s each), power fluctuation +5%; Evaluation thresholds: Short message transmission and reception success rate ≥99%, CEP95≤2m, MOS≥4, QoE≥85, anti-interference threshold ≥-90dBm.

[0073] S2: Data Acquisition Laboratory: The RDSS closed-loop system performed 1000 short message transmission and reception tests, achieving a success rate of 99.8%, an average latency of 1.2s, a measured voice MOS of 4.2, and a recorded anti-interference threshold of -88dBm; Road: A vehicle traveled 20km along a preset route, achieving a short message transmission and reception success rate of 98.5%, a CEP95 of 1.8m, and a beam switching success rate of 99.2%, with 3 communication interruption events occurring in the tunnel; Virtual: The HIL platform experienced a signal interruption as preset, with a system recovery time of 0.7s, and no errors were found in the data integrity check.

[0074] S3: Multi-dimensional assessment Link performance: Short message average success rate ((99.8%+98.5%) / 2=99.15%), SNR=18dB, both meeting the preset thresholds; Positioning and timing: CEP95=1.8m, TTFF=25s, both meeting standard requirements; Security and anti-interference: Anti-interference threshold -88dBm, data integrity 100%, meeting security protection requirements; QoS / QoE: D=1.4s, P loss =1.2%, MOS=4.2, substituting into the fusion model, we get QoE=0.35e^(-0.002×1400)+0.4×(1-0.012)+0.25×4.2≈89, which reaches the excellent level.

[0075] S4: Data Fusion Analysis Anomaly Detection: AI algorithms detected two short message losses in the road tunnel, the root cause of which was signal obstruction causing SNR to attenuate to 12dB, with anomaly detection accuracy ≥95%; Deviation Calibration: Based on the correlation data of signal strength in the laboratory and road scenarios, the test deviation caused by environmental differences was corrected, and the data consistency after calibration was ≤5%; Visualization Output: A heat map of signal coverage in the urban canyon scene was generated, clearly identifying the tunnel as a weak signal area.

[0076] S5: Results Output and Optimization Test report: The "LEO + Beidou RDSS" integrated communication system of this vehicle model complies with the requirements of 3GPP Rel-18 and GB / T40764-2021 standards; Optimization suggestion: For the problem of weak signal in tunnels, performance can be improved by optimizing antenna gain (increasing to 18dBi) or increasing satellite beam coverage.

[0077] This invention, through an innovative design of "three-layer collaborative detection + four-dimensional quantitative evaluation + intelligent fusion of multi-source data," systematically solves the core technical pain points in whole-vehicle satellite communication testing, such as incomplete scenario coverage, single evaluation dimensions, difficulty in quantifying user experience, and poor compatibility with multiple standards. This system and method can be widely applied to the R&D verification, regulatory compliance certification, and mass production quality monitoring of satellite communication terminals for intelligent connected vehicles, providing standardized and intelligent testing technology support for the large-scale application of "space-ground integrated" vehicle-to-everything (V2X) communication. In the future, it can be further integrated with cutting-edge technologies such as 6G NTN and AI large-scale models to expand advanced functions such as multi-constellation collaborative detection and predictive maintenance, continuously supporting the technological evolution from GEO emergency communication to LEO broadband routine services, and promoting the high-quality development of the intelligent connected vehicle industry towards global interconnection and intelligence.

[0078] For the foregoing embodiments, in order to simplify the description, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to this application.

[0079] The above embodiments describe the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Modifications and variations made by those skilled in the art without departing from the spirit and scope of the invention should be within the protection scope of the appended claims.

Claims

1. A multi-dimensional collaborative detection system for whole vehicle satellite communication, comprising a central control module, wherein the central control module serves as the core scheduling unit of the multi-dimensional collaborative detection system, characterized in that, The multi-dimensional collaborative detection system also includes: The core equipment cluster is bidirectionally connected to the central control module and uses a high-precision time synchronization method to achieve nanosecond-level time alignment of multi-source data across the entire system, providing a unified time base for the three-layer detection units. The three-layer detection unit is interconnected with the central control module via a high-speed industrial bus. It synchronously collects multi-source data based on the unified time base provided by the core equipment cluster and uploads it to the central control module. The three-layer detection unit includes a laboratory static verification subsystem, a road dynamic measurement subsystem, and a virtual simulation reproduction subsystem. The three subsystems are interconnected through the central control module to achieve scene switching, bidirectional data mapping, and parameter calibration. The four-dimensional collaborative evaluation unit, embedded in the central control module, performs parallel collaborative calculations on four dimensions—link performance, QoS / QoE, positioning and timing, and security and anti-interference—based on multi-source data aligned at the same time. It also establishes a correlation analysis model of positioning error, communication delay, beam switching status, and QoE experience. The data fusion processing unit is connected to the central control module. It receives multi-source data from the three-layer detection unit, completes data integration and cross-scenario consistency processing, and then inputs it into the four-dimensional collaborative evaluation unit to complete the quantitative evaluation. It outputs a unified quantitative evaluation result and feeds the result back to the central control module. The central control module adaptively adjusts the test parameters of the three-layer detection unit based on the result, or generates terminal design iteration suggestions, forming a closed loop of the entire process of "collection-evaluation-analysis-optimization-re-verification". Visualization and Feedback Unit: Interacts bidirectionally with the central control module to present detection data, evaluation indicators, correlation analysis results, and anomaly diagnosis reports; The three-layer detection unit includes: The laboratory static verification subsystem, deployed in the OTA anechoic chamber, simulates multi-constellation signals, dynamic channel characteristics, and complex electromagnetic interference through multi-device linkage, conducts basic index tests, and collects core parameters to provide benchmark data for subsequent tests. The road dynamic measurement subsystem is used to conduct dynamic tests on a preset test route and simultaneously collect satellite signal strength, positioning data, beam switching status and communication interruption event parameters to verify the communication robustness in a real road environment. The virtual simulation reproduction subsystem is used to reproduce extreme working conditions and abnormal scenarios, conduct system redundancy and self-recovery capability tests, and support unlimited reproduction of test scenarios and iterative optimization of parameters. The three subsystems of the three-layer detection unit adopt a unified time base for synchronous acquisition, with an acquisition time deviation of ≤10ns and a sampling frequency of ≥1kHz. The three subsystems achieve bidirectional data mapping and calibration through a central control module: the benchmark parameters output by the laboratory static verification subsystem are automatically transmitted to the road dynamic measurement subsystem as a reference benchmark. The abnormal working condition data collected by the road dynamic measurement subsystem triggers the virtual simulation reproduction subsystem to perform 1:1 scene reproduction and parameter iteration. The optimized parameters of the virtual simulation reproduction subsystem are transmitted back to the laboratory static verification subsystem for verification, forming a cross-scene closed loop. The road dynamic measurement subsystem also includes a scene tag association module. The scene tag association module encodes the real-time collected road scene type information into scene tags, associates and models them with the synchronously collected link performance, positioning accuracy, and safety anti-interference indicators, outputs scene-level diagnostic conclusions, and synchronously feeds back the scene tags and abnormal event data to the virtual simulation reproduction subsystem.

2. The multi-dimensional collaborative detection system for whole vehicle satellite communication as described in claim 1, characterized in that, The four-dimensional collaborative evaluation unit includes: The link performance evaluation subsystem is used to evaluate signal-to-noise ratio, bit error rate, short message transmission and reception success rate, beam switching success rate, and link interruption duration. The QoS / QoE collaborative evaluation subsystem adopts a QoS / QoE fusion evaluation model to achieve accurate quantitative mapping from objective QoS indicators to subjective experience quality. The positioning and timing evaluation subsystem is used to evaluate the positioning accuracy, first acquisition time, and time deviation of CEP95. The security anti-interference assessment subsystem is used to assess anti-interference capability thresholds, data transmission integrity, and tamper resistance.

3. The multi-dimensional collaborative detection system for whole vehicle satellite communication as described in claim 1, characterized in that, The data fusion processing unit includes: The high-precision data synchronization subsystem adopts any one of the following protocols: PTP v2, GNSSDO, IEEE1588, or hard-triggered synchronization, to achieve time alignment of multi-source data. The intelligent data processing subsystem integrates an improved Kalman filter algorithm, an LSTM neural network anomaly detection model, and a Gaussian process regression deviation calibration model to achieve noise filtering, anomaly detection, and deviation correction for cross-scene data. The automated control subsystem automatically adjusts test parameters or generates terminal optimization suggestions based on the evaluation results, realizing a closed-loop iteration of test-analysis-optimization. The LSTM neural network anomaly recognition model of the intelligent data processing subsystem has a 64-dimensional input layer. The input features include time-series data of signal power, delay, packet loss rate, SNR, BER, and beam switching status, with a time step of 30 sampling periods. The hidden layer consists of 3 LSTM layers, each with 64 neurons, and a dropout rate of 0.

2. The output layer outputs the anomaly type and confidence level. The anomaly judgment rules are as follows: power drift refers to a signal power change ≥ 0.5dB and a duration ≥ 5s; delay abrupt change refers to a delay change ≥ 50ms and a duration ≥ 3s; and link interruption refers to a signal power ≤ -140dBm and a duration ≥ 10ms.

4. The multi-dimensional collaborative detection system for whole vehicle satellite communication as described in claim 1, characterized in that, It also includes an anomaly root cause inference module, which performs correlation analysis on power drift, time delay abrupt change, beam switching failure anomaly events with vehicle attitude, signal obstruction, and electromagnetic environment parameters, and outputs anomaly root cause diagnosis results.

5. A multi-dimensional collaborative detection system for whole vehicle satellite communication as described in any one of claims 1 to 4, characterized in that, The multi-dimensional collaborative detection system can be deployed in any of the following ways: Form 1: Vehicle-side data acquisition module + edge computing unit + cloud analysis platform. The vehicle-side data acquisition module collects multi-source data and uploads it to the edge computing unit for preprocessing. The edge computing unit then uploads the preprocessed data to the cloud analysis platform for fusion evaluation and closed-loop optimization. Form 2: Vehicle-side data acquisition module + factory server. The vehicle-side data acquisition module collects multi-source data and stores it locally, while simultaneously uploading it to the factory server for data fusion analysis, evaluation, and optimization suggestion generation. In Form 1, the vehicle-mounted data acquisition module includes an integrated vehicle-mounted data acquisition terminal equipped with an FPGA chip, a GNSS high-precision positioning module, and an omnidirectional polarized satellite signal receiving antenna; the edge computing unit preprocesses the acquired data and then uploads it to the cloud.

6. A multi-dimensional collaborative detection method for whole vehicle satellite communication, implemented based on the multi-dimensional collaborative detection system for whole vehicle satellite communication as described in any one of claims 1 to 4, characterized in that, Includes the following steps: S1: Parameter configuration phase, configure the appropriate standard system according to test requirements, and set the satellite constellation type, communication standard, test scenario parameters and evaluation threshold range; S2: Multi-scenario data acquisition phase, the three-layer detection unit is started simultaneously to collect multi-source data, including: link and radio frequency parameters in the controlled environment, dynamic communication data, and system response data in abnormal scenarios; S3: Multi-dimensional collaborative evaluation stage, which uses a four-dimensional collaborative evaluation unit to process the collected data in parallel, including: calculating the signal-to-noise ratio and beam switching success rate, analyzing positioning accuracy and time deviation, detecting anti-interference capability and data integrity, and outputting latency, packet loss rate and experience index. S4: Data fusion and analysis stage, using AI algorithms to identify abnormal data, calibrate cross-scenario test deviations, and generate visual analysis results; S5: Result output and optimization stage. The test report is displayed through the visualization and feedback unit. Feedback is given to the central control module for non-compliant indicators, and test parameters are adjusted adaptively or terminal optimization suggestions are generated. The test scenario parameters include vehicle trajectory, satellite orbit parameters, signal power, elevation angle distribution, and interference source type, and support multi-constellation signal combination configuration of GPS / GLONASS / Galileo / BeiDou / LEO.

7. The multi-dimensional collaborative detection method for whole vehicle satellite communication as described in claim 6, characterized in that, The AI ​​algorithm includes an anomaly detection model and a deviation calibration model. The anomaly detection model automatically detects power drift and delay mutation problems based on an LSTM neural network. The anomaly judgment rules are as follows: power drift refers to a signal power change ≥ 0.5dB and a duration ≥ 5s; delay mutation refers to a delay change ≥ 50ms and a duration ≥ 3s; link interruption refers to a signal power ≤ -140dBm and a duration ≥ 10ms. The deviation calibration model is based on Gaussian process regression to correct for environmental differences between laboratory and road testing.

8. The multi-dimensional collaborative detection method for whole vehicle satellite communication as described in claim 6, characterized in that, The analysis of positioning accuracy and time deviation specifically employs a three-level process to verify the accuracy of positioning and timing, including: QoS metric extraction is based on the PTP v2 timestamp, calculating the difference between the message sending time and the receiving time using the following formula: D = Trecv - Tsend, where: D is the end-to-end delay; Tsend is the timestamp of the message being sent from the test device; Trecv is the timestamp of the vehicle communication module receiving and acknowledging the message. The QoE model is used to calculate QoE experience quality, employing an exponential function combination model. The formula is as follows: , where α, β, γ, and δ are regression weights calibrated based on multi-scenario test data; For packet loss / frame loss rate, MOS is the voice quality score; The rating system is based on the QoE (Quality of Experience) output rating. The rating rules are as follows: QoE ≥ 85 is excellent; 70-84 is good; and < 70 is substandard.

9. The multi-dimensional collaborative detection method for whole vehicle satellite communication as described in claim 6, characterized in that, Step S4 includes the following sub-steps: Data preprocessing: Time synchronization alignment, noise filtering, and feature parameter alignment are performed on the data collected by the three-layer detection unit; An LSTM neural network model is used to achieve automatic anomaly detection; Gaussian process regression algorithm is used to correct cross-scenario testing bias; Generate visualizations of signal coverage heatmaps, link reliability curves, and QoE distribution maps.

Images