Roadside holographic sensing platform data synchronization method and system based on big data

Abstract

The invention discloses a big data-based roadside holographic sensing platform data synchronization method and system, relates to the technical field of smart traffic, and aims to improve the accuracy of global traffic situation sensing in heterogeneous data synchronization and fusion processes of an existing roadside holographic sensing platform. The method comprises the following steps: acquiring heterogeneous data with time stamps from a plurality of pieces of edge sensing equipment, and respectively collecting the plurality of pieces of heterogeneous data into respective corresponding sliding time windows according to the time stamps carried by each piece of heterogeneous data; according to the data integrity score and the data credibility score, one fusion strategy is selected from multiple preset data fusion strategies, and the multiple data fusion strategies at least comprise a standard fusion strategy and a degradation fusion strategy; and performing fusion processing on the heterogeneous data in the sliding time window according to the selected fusion strategy to generate a global traffic situation view with a data quality identifier.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation technology, and in particular to a data synchronization method and system for a roadside holographic perception platform based on big data. Background Technology

[0002] In the field of intelligent transportation, roadside holographic perception platforms aim to build a comprehensive and real-time digital view of the traffic environment. This requires efficient and accurate synchronization and fusion of heterogeneous data from various edge sensing devices. However, in actual operation, a slight frequency offset from the upstream clock source leads to accumulated timestamp errors between distributed nodes, which in turn triggers a series of chain reactions such as misjudgments in data distribution strategies, loss of data packets in network switches, and even the central platform's security verification module mistakenly rejecting retransmission of data packets. Ultimately, this causes the data fusion mechanism to fail, seriously affecting the accuracy of global traffic situational awareness. Summary of the Invention

[0003] This application provides a data synchronization method and system for a roadside holographic perception platform based on big data. It aims to solve the problem that in the process of heterogeneous data synchronization and fusion of existing roadside holographic perception platforms, the cumulative timestamp error between distributed nodes caused by the slight frequency offset of the upstream clock source leads to a series of chain reactions, such as misjudgment of data distribution strategy, loss of data packets in network switches, and false rejection of retransmission of data packets by the central platform security verification module. Ultimately, this causes the data fusion mechanism to fail and seriously affects the accuracy of global traffic situation perception.

[0004] Firstly, to address the aforementioned technical problems, this invention provides a data synchronization method for a roadside holographic perception platform based on big data. This method includes: acquiring time-stamped heterogeneous data from multiple edge sensing devices, and aggregating the heterogeneous data into their respective corresponding sliding time windows based on the time stamp carried by each piece of data; for each sliding time window, determining a data integrity score to characterize the data coverage within the sliding time window based on a preset list of expected data sources and the actual amount of data received within the sliding time window; and determining a data reliability score to characterize the reliability of the data within the window based on the logical relationship between the time stamp of the data packet constituting the heterogeneous data within the sliding time window and the sequence number of the data packet itself, the historical time stamp quality archive of each edge sensing device, and the consistency of the sensing content of the sensors contained in different edge sensing devices for the same traffic target; selecting a fusion strategy from a preset set of multiple data fusion strategies based on the data integrity score and the data reliability score, wherein the multiple data fusion strategies include at least a standard fusion strategy and a degraded fusion strategy; and performing fusion processing on the heterogeneous data within the sliding time window according to the selected fusion strategy to generate a global traffic situation view with data quality identifiers.

[0005] Secondly, this application provides a data synchronization system for a roadside holographic perception platform based on big data, comprising: an acquisition unit, configured to acquire heterogeneous data with time stamps from multiple edge sensing devices, and to aggregate the multiple heterogeneous data into their respective corresponding sliding time windows according to the time stamps carried by each heterogeneous data; a determination unit, configured to, for each sliding time window, determine a data integrity score to characterize the data coverage within the sliding time window based on a preset expected data source list and the actual amount of data received within the sliding time window, and determine a data reliability score to characterize the data reliability within the window based on the logical relationship between the time stamps of the data packets constituting the heterogeneous data within the sliding time window and the sequence numbers of the data packets themselves, the historical time stamp quality archives of each edge sensing device, and the consistency of the sensing content of the sensors contained in different edge sensing devices for the same traffic target; a selection unit, configured to select a fusion strategy from a preset plurality of data fusion strategies based on the data integrity score and the data reliability score, wherein the plurality of data fusion strategies include at least a standard fusion strategy and a degraded fusion strategy; and a generation unit, configured to perform fusion processing on the heterogeneous data within the sliding time window according to the selected fusion strategy to generate a global traffic situation view with data quality identifiers.

[0006] This application offers at least the following advantages: It provides a data synchronization method for a roadside holographic perception platform based on big data. By acquiring heterogeneous data with time stamps and aggregating it into a sliding time window, it effectively addresses the problem of accumulated timestamp errors between distributed nodes, laying a precise temporal foundation for subsequent data processing. By introducing data integrity scores and data reliability scores, this application can comprehensively and objectively evaluate the coverage and reliability of data within the sliding time window, thereby overcoming the shortcomings of existing technologies where fusion mechanisms fail due to data quality issues. Furthermore, this application dynamically selects appropriate fusion strategies based on data integrity and reliability scores, including standard fusion strategies and degraded fusion strategies. This avoids blindly fusion of low-quality data and effectively solves a series of chain reactions such as misjudgment of data distribution strategies, loss of network switch data packets, and erroneous rejection of retransmission of data packets by the central platform's security verification module. Finally, the heterogeneous data is fused using the selected fusion strategy to generate a global traffic situation view with data quality indicators. This enables the platform to output coherent, accurate traffic information with clear quality indicators, significantly improving the roadside holographic perception platform's real-time perception capability and accuracy of traffic situation, providing a reliable basis for intelligent transportation decision-making, and overcoming the shortcomings of existing technologies. Attached Figure Description

[0007] Figure 1This is a flowchart illustrating a data synchronization method for a roadside holographic sensing platform based on big data, as provided in this application. Detailed Implementation

[0008] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0009] It should be noted that similar reference numerals 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. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0010] Traditional roadside holographic sensing platforms, when synchronizing and fusing heterogeneous data from various edge sensing devices, suffer from accumulated timestamp errors between distributed nodes due to minute frequency offsets in the upstream clock source. This leads to a series of chain reactions, including misjudgments in data distribution strategies, packet loss in network switches, and even erroneous rejection of retransmissions by the central platform's security verification module. Ultimately, this causes the data fusion mechanism to fail, severely impacting the accuracy of global traffic situation awareness. Without addressing these issues, the platform will be unable to construct a coherent and accurate global traffic view, fundamentally undermining its real-time traffic situation awareness and accuracy. The output traffic information will be riddled with severe and unpredictable spatial and temporal biases, failing to provide a reliable basis for intelligent transportation decision-making.

[0011] In view of the above problems, this application provides a data synchronization method for a roadside holographic perception platform based on big data. By introducing data integrity score and data credibility score to comprehensively evaluate data quality, and dynamically selecting data fusion strategy based on this, it effectively addresses problems such as distributed clock deviation, data loss and verification misjudgment, and ensures the accuracy and reliability of the global traffic situation view.

[0012] The following specific embodiments will provide a detailed introduction and explanation of the data synchronization method and system for the roadside holographic perception platform based on big data provided in this application.

[0013] Reference Figure 1This application provides a data synchronization method for a roadside holographic sensing platform based on big data. The method may include the following steps: S1. Obtain heterogeneous data with time stamps from multiple edge sensing devices, and according to the time stamp carried by each heterogeneous data, collect the multiple heterogeneous data into their respective sliding time windows.

[0014] It should be noted that edge sensing devices refer to various sensor devices, such as cameras, radar, and lidar, deployed along roads to collect real-time traffic environment data. These devices can sense information such as vehicles, pedestrians, and traffic events, and generate raw data with time stamps.

[0015] Heterogeneous data refers to data generated from edge sensing devices of different types and manufacturers, which may have different formats, data structures, and time stamp precision.

[0016] A timestamp is a timestamp contained in a data packet and generated by the device's internal clock, used to indicate the time of data acquisition or generation. In this application, the precision of the timestamp can reach the nanosecond or microsecond level to meet the requirements of high-precision data synchronization.

[0017] A sliding time window is a data processing mechanism that defines a continuous time interval to aggregate heterogeneous data whose timestamps fall within that interval for processing. The window slides forward by a preset time length, ensuring the real-time nature and continuity of data processing.

[0018] In one implementation, a data acquisition module can be used to receive data packets from the edge sensing device. This module can be a high-performance network interface card configured with dedicated hardware acceleration capabilities, enabling continuous data stream reception with extremely high throughput. Each data packet is time-stamped by an internal clock within the edge sensing device upon generation. This time-stamp can be at the nanosecond or microsecond level to ensure sufficient time accuracy. Received data packets are treated as heterogeneous data because they may originate from different types of sensors and have different data formats. Subsequently, based on the time-stamp of each data packet, it is assigned to a sliding time window of a preset length. For example, a 50-millisecond sliding time window can be set, and all data packets whose time-stamps fall within this 50-millisecond interval will be aggregated into this window.

[0019] In another implementation, the data acquisition module can be a software-defined networking (SDN) controller that programmatically controls network traffic, routing data packets from edge sensing devices to specific processing queues. The timestamp of each data packet is generated by the edge sensing devices after synchronization via PTP (Precise Time Protocol). The SDN controller dynamically allocates the data packets to different sliding time windows based on their timestamps. For example, data packets can be categorized into corresponding sliding time windows based on the second and millisecond portions of their timestamps.

[0020] S2. For each sliding time window, based on the preset expected data source list and the actual amount of data received within the sliding time window, determine the data integrity score to characterize the data coverage within the sliding time window. Based on the logical relationship between the timestamp of the data packet constituting the heterogeneous data within the sliding time window and the sequence number of the data packet itself, the historical timestamp quality profile of each edge sensing device, and the consistency of the sensing content of the sensors contained in different edge sensing devices for the same traffic target, determine the data reliability score to characterize the data reliability within the sliding time window.

[0021] Among them, the data integrity score is an indicator used to characterize the degree of data coverage within the sliding time window, reflecting the degree of matching between the expected data and the actual received data.

[0022] The data credibility score is an indicator used to characterize the reliability of data within a sliding time window, taking into account the consistency of time stamps, the reliability of data sources, and the consistency of perceived content.

[0023] The data integrity score can be determined based on a pre-defined list of expected data sources and the actual amount of data received within a sliding time window. For example, the expected data source list could be a configuration file recording all edge sensing devices that should report data within a specific road segment and time period, as well as the expected data reporting frequency or amount for each device. Within each sliding time window, the system counts the actual number or amount of data packets sent by each edge sensing device. Then, it calculates the ratio of the actual data amount to the expected data amount for each edge sensing device. For example, if a device is expected to report 100 data packets per second but only reports 80 in the current window, its ratio is 0.8. Finally, summing these ratios for all edge sensing devices and dividing by the total number of devices yields the data integrity score for that sliding time window.

[0024] The determination of data credibility score is more complex. It takes into account the logical relationship between the timestamp of the data packet that constitutes the heterogeneous data within the sliding time window and the sequence number of the data packet itself, the historical timestamp quality profile of each edge sensing device, and the consistency of the sensing content of the sensors contained in different edge sensing devices for the same traffic target.

[0025] For example, regarding the logical relationship between packet timestamps and sequence numbers, one can check whether the timestamp of each packet is consistent with the logical order indicated by its sequence number. If a packet with a later sequence number carries a timestamp that is significantly earlier than a packet with an earlier sequence number, it indicates a timestamp anomaly. A timestamp consistency score can be assigned to each packet, and then the scores can be aggregated to obtain the timestamp consistency score for the window.

[0026] For each edge sensing device, a historical time-stamp quality profile is provided, recording the device's average time-stamp deviation and jitter over a past period. This historical data can be used to assess the stability of the time-stamps generated by the device. For example, a source reliability score can be assigned to each edge sensing device based on this historical data, and then the scores can be aggregated to obtain a sliding time window source reliability score.

[0027] To ensure consistency in the perceived content of the same traffic target by sensors within different edge sensing devices, perception reports from at least two different sensors can be obtained for each traffic target detected within a sliding time window. Each perception report includes the traffic target's spatial location, speed, and target type attribute. Then, the deviation values ​​for the same attribute in the perception reports from different sensors, such as position deviation and speed deviation, are calculated. Based on the extent to which these deviation values ​​exceed a preset allowable threshold, a content verification score is assigned to the traffic target, and these scores are aggregated to obtain the content verification score for the sliding time window.

[0028] Finally, the time stamp consistency score, source reliability score, and content verification score of the sliding time window are weighted and summed, and the sum is used as the data credibility score.

[0029] S3. Based on the data integrity score and data credibility score, select one data fusion strategy from a variety of preset data fusion strategies.

[0030] Among them, multiple data fusion strategies include at least standard fusion strategies and degraded fusion strategies.

[0031] It should be noted that data fusion strategy refers to the method of fusion of heterogeneous data by selecting different algorithms and processing procedures based on the data quality assessment results, such as standard fusion strategy and degradation fusion strategy.

[0032] The preset data fusion strategies include at least a standard fusion strategy and a degraded fusion strategy. For example, two thresholds can be set: a first threshold and a second threshold. When both the data completeness score and the data credibility score are higher than the preset first threshold, it indicates that the data quality is extremely high, and the standard fusion strategy is selected. The standard fusion strategy instructs the use of high-precision fusion algorithms to fuse all heterogeneous data within the current sliding time window, such as using advanced algorithms like Kalman filtering or particle filtering, to generate the most accurate traffic situation view.

[0033] When the data integrity score or data reliability score is below a first threshold but above a preset second threshold, it indicates a certain degree of data quality degradation, and a downgraded fusion strategy is selected. This strategy prioritizes fusion of data reported by edge sensing devices whose data integrity and reliability scores are both above the predetermined standards. For sensing areas corresponding to edge sensing devices where the reported data volume is below expectations or whose reported data reliability score is below the predetermined standard, a partial holographic view is generated with an added data incompleteness warning. Furthermore, for critical data points with missing data in a continuous time series, the downgraded fusion strategy performs predictive data filling based on the movement trend of the same traffic target within the previous sliding time window, and marks the filling result as predicted data.

[0034] When both the data integrity score and the data credibility score fall below the second threshold, it indicates severely inadequate data quality, and a severe degradation strategy is selected. The severe degradation strategy instructs the generation of a simplified traffic situation view or the direct issuance of a data quality alert, avoiding the use of low-quality data for complex fusion and thus preventing the generation of misleading information.

[0035] S4. Based on the selected fusion strategy, perform fusion processing on the heterogeneous data within the sliding time window to generate a global traffic situation view with data quality labels.

[0036] The global traffic situation view is a comprehensive and real-time digital representation of the current traffic environment generated after fusion processing, with data quality indicators. It can be a three-dimensional digital sand table or a two-dimensional map interface.

[0037] Based on the selected fusion strategy, heterogeneous data within the sliding time window are fused to generate a real-time 3D digital sandbox or 2D map interface as the base view. For example, if the standard fusion strategy is selected, a high-precision algorithm will be used to fuse all data, generating a richly detailed and highly accurate base view. If the degraded fusion strategy is selected, some data will be fused first, and missing data will be predicted and filled in, generating a base view with some incomplete or predicted data.

[0038] In the base view, based on the data integrity and reliability scores within the sensing areas corresponding to each edge sensing device, and the processing results of the selected fusion strategy, differentiated visual markers are used to render different sensing areas or different traffic targets to identify their data quality, ultimately resulting in a global traffic situation view. For example, sensing areas processed using a standard fusion strategy and whose corresponding data integrity and reliability scores are both higher than a preset first threshold can be rendered using bright colors and solid lines, marking them as high-reliability areas. Sensing areas processed using a downgraded fusion strategy can be rendered using semi-transparent colors and dashed lines, with overlaid text prompts to explain the incompleteness of the data or the presence of predicted data attributes within the area. Sensing areas whose data integrity and reliability scores are both lower than a preset second threshold are highlighted with specific colors as warning markers, marking them as areas with severely missing data.

[0039] The data synchronization method for a roadside holographic perception platform based on big data proposed in this application comprehensively and in real-time evaluates the quality of heterogeneous data by introducing data integrity scores and data reliability scores. This evaluation mechanism can effectively identify data quality degradation caused by problems such as upstream clock source frequency offset, misjudgment of data distribution strategies, loss of network switch data packets, and false rejection by security verification modules. By dynamically selecting standard fusion strategies, degraded fusion strategies, or severely degraded strategies, this application can flexibly adjust the fusion method according to the actual data quality, avoiding the generation of erroneous or unreliable global traffic situation views when the data quality is poor by traditional fixed fusion strategies.

[0040] Specifically, the steps described above—acquiring time-stamped heterogeneous data from multiple edge sensing devices and grouping the heterogeneous data into their respective sliding time windows based on the time stamp carried by each piece of heterogeneous data—can be implemented in the following ways.

[0041] The process of acquiring heterogeneous data with time stamps from multiple edge sensing devices and aggregating the heterogeneous data into their respective sliding time windows based on the time stamps carried by each heterogeneous data includes: continuously receiving data packets reported by roadside edge sensing devices through a data acquisition server based on a high-performance network interface card, each data packet containing a nanosecond or microsecond-level time stamp generated by the device's internal clock; using the received data packets from different edge sensing devices as the heterogeneous data, and allocating each data packet to a sliding time window of a preset time length based on the time stamp of each data packet, wherein each sliding time window is used to aggregate data packets from different edge sensing devices whose time stamps fall within the same continuous time interval.

[0042] Specifically, the data acquisition server is configured to continuously receive data packets reported by roadside edge sensing devices via a high-performance network interface card. The high-performance network interface card provides high-throughput and low-latency data transmission capabilities to handle the influx of large amounts of real-time sensing data. Roadside edge sensing devices typically include, but are not limited to, various sensors such as cameras, radar, and lidar. They continuously monitor the road traffic environment and generate raw data containing information such as the location, speed, and type of traffic targets (e.g., vehicles, pedestrians, non-motorized vehicles). Each data packet is assigned a high-precision timestamp by the device's internal clock upon generation; this timestamp can be at the nanosecond or microsecond level, ensuring accurate time synchronization of the data.

[0043] Data packets received from different edge sensing devices are considered heterogeneous data because they may originate from different types and manufacturers of devices, and have different data formats and content. To effectively manage the timing of this heterogeneous data and perform initial aggregation, the system assigns each data packet to a sliding time window of a preset length based on the timestamp carried by that packet. The sliding time window is a dynamic time interval whose length can be configured according to actual application requirements. The core function of each sliding time window is to aggregate data packets from different edge sensing devices whose timestamps fall within the same continuous time interval, thereby forming a time-aligned and content-complementary data set, laying the foundation for subsequent data fusion processing.

[0044] This application's solution introduces a data acquisition server based on a high-performance network interface card, ensuring real-time and efficient reception of massive amounts of data from roadside edge sensing devices and avoiding data transmission bottlenecks. Nanosecond or microsecond-level time stamps are generated by the device's internal clock, guaranteeing high-precision alignment of data in the time dimension. This is crucial for accurately determining the logical order of data packets and performing multi-source data fusion. By allocating these heterogeneous data packets with precise time stamps to a sliding time window, preliminary temporal aggregation of data from different sources and of different types is achieved. This allows all relevant sensing data to be centrally processed within a specific time period, providing a unified data view for subsequent data integrity and reliability assessments and fusion strategy selection.

[0045] The above technical solutions effectively address issues such as low efficiency, insufficient time synchronization accuracy, and chaotic management of heterogeneous data that may exist in traditional data collection and preliminary aggregation processes. The use of high-performance network interface cards significantly improves data throughput, ensuring lossless reception of real-time data; the introduction of high-precision time stamps provides a solid time benchmark for subsequent data quality assessment and fusion; and the sliding time window mechanism enables automated and refined time-series management of massive amounts of heterogeneous data, greatly improving the efficiency and accuracy of data preprocessing and laying a solid foundation for building a high-quality global traffic situation view.

[0046] Specifically, in the above-mentioned data synchronization method for the roadside holographic perception platform based on big data, for each of the multiple sliding time windows, a data integrity score is determined to characterize the data coverage within the sliding time window based on a preset list of expected data sources and the actual amount of data received within the sliding time window.

[0047] The expected data source list records all edge sensing devices that should report data within the current road segment and time period, as well as the expected data reporting frequency or data volume. For each of the multiple sliding time windows, based on the preset expected data source list and the actual data volume received within the sliding time window, a data integrity score is determined to characterize the data coverage within the sliding time window. This includes: counting the number of data packets or data volume actually sent by each edge sensing device within the current sliding time window; calculating the ratio of the actual data volume to the expected data volume for each edge sensing device; summing the ratios for all edge sensing devices and dividing by the total number of devices to obtain the data integrity score.

[0048] The expected data source list can be understood as a configuration list or database that records detailed information about all edge sensing devices participating in data reporting within a specific time period and geographical area. This information includes not only the unique identifier of each device but also the data frequency (e.g., the number of data packets sent per second) or the expected total data volume that each device should report under normal operating conditions. This list serves as a benchmark for assessing data integrity, ensuring that the system can identify any missing or insufficiently reported data.

[0049] Specifically, counting the number or amount of data packets actually sent by each edge sensing device within the current sliding time window refers to the system counting or quantifying the data received from each edge sensing device within that window at the end of each preset sliding time window. For example, it can count the total number of data packets sent by each device within the window, or the total number of bytes in these data packets. This step aims to obtain the actual contribution of each data source.

[0050] Furthermore, the ratio of the actual data volume to the expected data volume for each edge sensing device is calculated. The sum of these ratios for all edge sensing devices is then divided by the total number of devices to obtain the data integrity score. This ratio reflects the degree to which a single device's data reporting meets the target. For example, if a device is expected to report 100 data packets per second but only reports 80, its ratio is 0.8. Summing these ratios for all devices and dividing by the total number of devices participating in the reporting yields a comprehensive data integrity score. This score is a value between 0 and 1; the closer it is to 1, the higher the data integrity, meaning the closer the actual received data volume is to the expected amount.

[0051] This application's solution, by pre-setting a list of expected data sources, clarifies all edge sensing devices that should report data within a specific spatiotemporal range and their expected data volumes, thus providing an objective benchmark for data integrity assessment. By statistically analyzing the actual data volume of each edge sensing device within each sliding time window and comparing it with the expected data volume, the contribution of each data source can be quantified. Furthermore, by summarizing and averaging the ratio of actual to expected data volumes for all devices, this solution can comprehensively reflect the data coverage within the entire sliding time window, thereby effectively assessing the data integrity within the current time window.

[0052] The aforementioned technical solution enables a quantitative and objective assessment of the data integrity of a roadside holographic perception platform within a specific sliding time window. This method not only identifies data reporting anomalies from individual edge sensing devices but also provides a macroscopic view of the data coverage across the entire sensing area through comprehensive calculations. This precise data integrity score provides a crucial basis for subsequent data fusion strategy selection, helping the system to take appropriate degradation measures when data is incomplete, thereby avoiding decision-making errors due to data gaps and improving the platform's accuracy and robustness in traffic situation perception.

[0053] Specifically, in the above method, the determination process of the data confidence score, which is used to characterize the reliability of data within the sliding time window, can be further refined.

[0054] In response, this application further proposes a method for determining the data reliability score used to characterize the reliability of data within the window, comprising: assigning a time stamp consistency score to each data packet based on the logical consistency between the timestamp of each data packet and the sequence number of the data packet itself within the sliding time window, and determining the time stamp consistency score of the window based on the time stamp consistency scores of all data packets within the window; assigning a source reliability score to each edge sensing device based on the time stamp deviation and jitter recorded in the historical time stamp quality archive of each edge sensing device, and determining the source reliability score of the sliding time window based on the source reliability scores of all edge sensing devices contributing data to the sliding time window; assigning a content verification score to each traffic target within the sliding time window based on the matching degree between the perceived content of the traffic target from different sensors, and determining the content verification score of the sliding time window based on the content verification scores of all traffic targets within the sliding time window; and weighted summing the time stamp consistency score, the source reliability score, and the content verification score of the sliding time window, and using the summation result as the data reliability score.

[0055] Specifically, the determination of the data reliability score aims to comprehensively evaluate the intrinsic quality and external reliability of heterogeneous data within the sliding time window. The time stamp consistency score is calculated by analyzing the degree of matching between the time stamps of data packets and the logical order indicated by the packet sequence numbers. For example, if the time stamps of data packets deviate significantly from the expected time order, their consistency score will decrease accordingly. The allocation of the source reliability score is based on the historical performance of each edge sensing device, i.e., the average deviation and jitter of its time stamps. Devices with poorer historical performance receive lower data source reliability scores. The determination of the content verification score focuses on the consistency of the perceived content of the same traffic target by different sensors. For example, when multiple sensors report significant differences in the position, speed, or type of the same vehicle, the content verification score of that traffic target will be affected. Finally, by weighted summing the scores of these three dimensions, a comprehensive data reliability score can be obtained, fully reflecting the reliability level of the data within the current sliding time window. The weighting coefficients of the summation can be preset and adjusted according to the actual application scenario and the degree of emphasis on different reliability factors.

[0056] This application's solution refines the evaluation of data credibility scores into three independent and complementary dimensions: timestamp consistency, source reliability, and content corroboration. This allows for a more comprehensive and accurate capture of potential data issues at the time, source, and content levels. The introduction of a timestamp consistency score helps identify inaccurate timestamps caused by device clock drift, network latency, or abnormal data transmission. The consideration of source reliability scores enables the system to assign different levels of trust to data from different sources based on the historical performance and stability of edge sensing devices, thereby preventing a decline in overall data quality due to a single faulty device. The calculation of content corroboration scores leverages the redundancy and complementarity of multi-sensor fusion. By comparing the perception results of different sensors on the same target, it effectively identifies and quantifies inconsistencies or errors in the perceived data. Therefore, through a multi-dimensional and refined evaluation mechanism, this application can generate a more accurate and robust data credibility score, providing a solid foundation for subsequent data fusion strategy selection.

[0057] Through the above technical solution, this application provides a more refined and comprehensive data credibility assessment mechanism. Compared to assessment methods that rely solely on a single indicator, this application significantly improves the accuracy and representativeness of data credibility scores by comprehensively considering time stamp consistency, the reliability of edge sensing device sources, and multi-sensor content verification. This multi-dimensional assessment method enables the system to more effectively identify and quantify potential errors, inconsistencies, or omissions in the data, thereby providing a more reliable quality basis for subsequent data fusion processing. Therefore, when selecting a data fusion strategy, decisions can be made based on a more accurate data credibility score, avoiding deviations in fusion results due to data quality issues. Ultimately, this generates a more valuable and practical global traffic situation view, enhancing the overall data processing capabilities and decision support level of the roadside holographic perception platform.

[0058] Specifically, the above-mentioned method of assigning a timestamp consistency score to each data packet based on the logical consistency between the timestamp of each data packet and its own sequence number within the sliding time window, and determining the timestamp consistency score of the window based on the timestamp consistency scores of all data packets within the window, can be further refined into the following steps: For each data packet within the sliding time window, obtain the logical order indicated by the timestamp of the data packet and its own sequence number; determine whether the deviation value exceeds a preset timestamp deviation threshold based on the deviation value between the timestamp of the data packet and the logical time expected by the logical order; if the deviation value does not exceed the timestamp deviation threshold, then the timestamp consistency score of the data packet is determined to be a full score; if the deviation value exceeds the timestamp deviation threshold, then the timestamp consistency score of the data packet is reduced proportionally according to the proportion of the deviation value relative to the timestamp deviation threshold; and the timestamp consistency scores of all data packets within the sliding time window are summarized to obtain the timestamp consistency score of the window.

[0059] The timestamp of a data packet refers to a timestamp generated by the internal clock of the edge sensing device, used to identify the generation or transmission time of the data packet. The sequence number of the data packet itself reflects the logical order of the data packet during transmission or generation. The logical time expected by this logical order can be understood as a theoretical time point calculated based on the data packet sequence number. For example, if the sequence numbers are consecutive and the transmission frequency of the data packets is known, the timestamp that each sequence number should correspond to can be calculated. The timestamp deviation threshold is a preset tolerance range used to measure the acceptable difference between the timestamp and the expected logical time. The maximum score is usually set to the highest score, such as 100 points, indicating a high degree of timestamp consistency. Proportional reduction means that the greater the deviation exceeds the threshold, the greater the score reduction, to accurately reflect the degree of degradation in timestamp quality. The timestamp consistency scores of all data packets can be summarized by calculating the average, weighted average, or median, to comprehensively evaluate the timestamp consistency level of the data within the entire sliding time window.

[0060] This application's solution meticulously analyzes the logical relationship between the timestamps and sequence numbers of each data packet within a sliding time window, enabling precise evaluation of the timestamp quality of individual data packets. Specifically, by comparing the actual timestamps with the expected logical timestamps derived from the sequence numbers, anomalies or inconsistencies in the timestamps can be identified. When the deviation is within an acceptable threshold, the timestamp quality is considered good, and a full score is awarded. Conversely, when the deviation exceeds the threshold, the score is reduced proportionally, thus quantifying the unreliability of the timestamps. Finally, by summarizing the scores of all data packets within the window, the consistency level of the timestamps across the entire sliding time window can be comprehensively and objectively reflected, providing accurate input for subsequent calculations of data reliability scores.

[0061] The aforementioned technical solution enables a refined and quantitative assessment of the time stamp consistency of heterogeneous data in a roadside holographic sensing platform. This assessment method not only considers the logical relationship between time stamps and data packet sequence numbers but also introduces quantifiable deviation thresholds and a proportional score reduction mechanism, making the judgment of time stamp quality more objective and accurate. Therefore, it can effectively identify data packets with significant time stamp deviations or jitter, providing a more reliable and detailed basis for subsequent data credibility score calculations, thereby improving the accuracy and robustness of the entire data synchronization method in data quality assessment.

[0062] Specifically, when determining the data reliability score used to characterize the reliability of data within a sliding time window, the process of assigning a source reliability score to each edge sensing device based on the time stamp deviation and jitter levels recorded in the historical time stamp quality archive of each edge sensing device, and determining the source reliability score of the sliding time window based on the source reliability scores of all edge sensing devices contributing data to the sliding time window, includes the following steps: For each edge sensing device, obtain the average time stamp deviation and time stamp jitter value of the device in the past time period from the historical time stamp quality archive; calculate the source reliability score of the device based on the average deviation value and the jitter value, according to a preset first weighting coefficient for characterizing the influence of deviation and a second weighting coefficient for characterizing the influence of jitter, wherein the larger the average deviation value or the larger the jitter value, the lower the calculated source reliability score; and summarize the source reliability scores of all edge sensing devices contributing data within the sliding time window to obtain the source reliability score of the sliding time window.

[0063] Specifically, the historical timestamp quality archive can be understood as a database or file system that continuously records and stores the timestamp quality performance of each edge sensing device during its historical operation. This archive details the average timestamp deviation and timestamp jitter values ​​for each device over different time periods. The average timestamp deviation refers to the average difference between the timestamp of the data reported by the device and the actual standard time or system reference time, reflecting the long-term accuracy of the device's clock. The timestamp jitter value characterizes the fluctuation or instability of the device's timestamp over a short period, reflecting the short-term stability of the device's clock. The average deviation and jitter values ​​are two key indicators when calculating the device's source reliability score. This application uses a preset first weighting coefficient to characterize the degree of deviation influence and a second weighting coefficient to characterize the degree of jitter influence to weight these two indicators. For example, linear weighting or nonlinear functions can be used to ensure that the larger the average deviation or the larger the jitter value, the lower the calculated source reliability score, thus accurately reflecting the negative impact of device timestamp quality on data reliability. Furthermore, after obtaining the source reliability scores of all contributing edge sensing devices within the sliding time window, these scores need to be aggregated. Aggregation methods may include, but are not limited to, averaging, weighted averaging, or weighted averaging based on the data contribution of each device within the current window, ultimately yielding the source reliability score for the sliding time window. This aggregated score comprehensively reflects the overall reliability level of all data source devices within the current window.

[0064] This application's solution introduces a historical time-stamp quality archive and calculates the source reliability score for each edge-sensing device based on the recorded average time-stamp deviation and time-stamp jitter values. This allows for a quantitative assessment of the inherent time synchronization performance of each data source device. This assessment considers the accuracy and stability of time stamps over long-term operation, avoiding the bias of judging solely based on single or short-term data performance. By summarizing the source reliability scores of these devices, the overall reliability level of all data source devices within the current sliding time window can be objectively reflected, providing a more comprehensive and accurate dimension for determining subsequent data credibility scores. This historical performance-based assessment mechanism enables data credibility scores to more accurately identify and distinguish heterogeneous data from devices with different quality levels, providing a solid foundation for subsequent data fusion strategy selection.

[0065] Through the above technical solutions, this application can more precisely assess the reliability of data sources for each edge sensing device. By utilizing historical time-stamped quality archives, misjudgments of data reliability caused by short-term device anomalies or random errors can be avoided, making the assessment of data source reliability more stable and accurate. Furthermore, by introducing weighted coefficients to comprehensively consider average deviation and jitter values, the impact of different time-stamped quality issues on the reliability score can be flexibly adjusted. This allows the final determined sliding time window's source reliability score to more realistically and comprehensively reflect the overall reliability of heterogeneous data within the current window, thereby improving the accuracy and guiding significance of data quality labeling in the global traffic situation view.

[0066] In some embodiments described above in this application, a data reliability score is proposed to be determined based on the logical relationship between the timestamp of the data packet constituting the heterogeneous data within the sliding time window and the sequence number of the data packet itself, the historical timestamp quality profile of each edge sensing device, and the consistency of the sensing content of the sensors contained in different edge sensing devices for the same traffic target. Specifically, in the process of determining the data credibility score, for each traffic target within the sliding time window, a content verification score is assigned to the traffic target based on the matching degree between the perceived content of the traffic target from different sensors. The content verification score of the sliding time window is determined based on the content verification scores of all traffic targets within the sliding time window. The specific steps are as follows: For each traffic target detected within the sliding time window, perception reports of the traffic target from at least two different sensors are obtained. Each perception report contains the spatial location, speed, and target type attribute of the traffic target. The deviation value of the same attribute in the perception reports of different sensors is calculated. The deviation value includes at least a position deviation value and a speed deviation value. It is determined whether the position deviation value exceeds a preset allowable position deviation threshold and whether the speed deviation value exceeds a preset allowable speed deviation threshold. The content verification score of the traffic target is determined based on the degree to which the deviation value exceeds the corresponding allowable deviation threshold. When any attribute deviation value exceeds the allowable threshold, the content verification score of the traffic target is reduced accordingly. The content verification scores of all traffic targets within the sliding time window are summarized to obtain the content verification score of the sliding time window.

[0067] Specifically, when acquiring perception reports, the roadside holographic perception platform receives perception data about the same traffic target from different types of edge perception devices, such as millimeter-wave radar, lidar, and high-definition cameras. After preliminary processing, this perception data forms a standardized perception report containing the traffic target's spatial location (e.g., X, Y, Z coordinates), movement speed (e.g., velocity components or resultant velocity in the X, Y, and Z directions), and target type attributes (e.g., vehicle, pedestrian, non-motorized vehicle, etc.). To ensure the validity of the content verification, perception reports from at least two different sensors are required.

[0068] Calculating the deviation value of the same attribute in perception reports from different sensors refers to comparing the same attribute regarding the same traffic target from perception reports from different sensors. For example, if two sensors report the position of the same vehicle, the Euclidean distance between the coordinates of these two positions is calculated as the position deviation value; if both report speed, the difference between the speed vectors is calculated as the speed deviation value. These deviation values ​​are key indicators for measuring the consistency of perception results from different sensors.

[0069] Furthermore, determining whether the position deviation value exceeds a preset allowable position deviation threshold and whether the speed deviation value exceeds a preset allowable speed deviation threshold is to set an acceptable error range. The allowable position deviation threshold and allowable speed deviation threshold can be preset according to the actual application scenario, sensor accuracy, and requirements for data reliability. For example, in a highway scenario, the accuracy requirements for position and speed may be higher, and the thresholds would be set smaller.

[0070] Therefore, the content verification score of the traffic target is determined based on the degree to which the deviation value exceeds the corresponding allowable deviation threshold. This means that if the deviation value is within the allowable threshold, the content verification score of the traffic target is high, or even full; if the deviation value exceeds the allowable threshold, the content verification score of the traffic target is reduced accordingly based on the excess ratio or a preset scoring function. When the deviation value of any attribute (position or speed) exceeds the allowable threshold, the content verification score of the traffic target will decrease, reflecting the inconsistency of perceived content.

[0071] Finally, the content verification scores of all traffic targets within the sliding time window are summarized to obtain the content verification score for the sliding time window. The summarization method can be an average, a weighted average, or the lowest score, to comprehensively reflect the consistency level of the perceived content of all traffic targets within the current window.

[0072] This application's solution quantifies the consistency of sensor perception by acquiring perception reports of the same traffic target from at least two different sensors and cross-comparing key attributes (such as spatial location, speed, and target type) in these reports. By setting an allowable deviation threshold and determining the content verification score based on the relationship between the actual deviation value and the threshold, the reliability of individual traffic target perception data can be objectively assessed. Finally, by aggregating the content verification scores of all traffic targets, a dimension based on multi-source data cross-validation is provided for the data credibility score across the entire sliding time window, effectively avoiding potential errors or anomalies in single-sensor data and improving the accuracy of data reliability assessment.

[0073] Through the aforementioned technical solution, this application can more precisely evaluate the reliability of traffic target perception content in heterogeneous data received by the roadside holographic perception platform. Especially in multi-sensor fusion scenarios, this solution can effectively identify and quantify the differences in perception results between different sensors, thereby providing early warning and handling of potential perception errors or data anomalies before data fusion. This not only improves the accuracy of data credibility scores but also provides a more solid foundation for subsequent data fusion strategy selection, ensuring that the final generated global traffic situation view has higher accuracy and reliability, and avoiding the risk of overall situation judgment errors due to single sensor failure or data deviation.

[0074] In some embodiments described above, this application proposes selecting a fusion strategy based on data integrity scores and data reliability scores. However, during implementation, without careful judgment and strategy adjustment for dynamic changes in data quality, a high-precision fusion algorithm may still be used when data quality is poor, leading to the introduction of erroneous or inaccurate traffic situation information. Alternatively, excessive degradation processing may occur when data quality is acceptable, failing to fully utilize effective data. Therefore, this application further proposes a more refined data fusion strategy selection mechanism to ensure that the most suitable fusion method is adopted under different data quality conditions, thereby improving the accuracy and reliability of the global traffic situation view.

[0075] To address this, this application further proposes the above-mentioned method of selecting a fusion strategy from a variety of preset data fusion strategies based on data integrity score and data credibility score, including: when both the data integrity score and data credibility score are higher than a preset first threshold, a standard fusion strategy is selected, wherein the standard fusion strategy indicates that a high-precision fusion algorithm is used to fuse all heterogeneous data within the current sliding time window; when either the data integrity score or the data credibility score is lower than the first threshold but higher than a preset second threshold, a degraded fusion strategy is selected; the degraded fusion strategy includes: prioritizing the fusion of data from edge detection systems where both the data integrity score and data credibility score are higher than a predetermined standard. For edge sensing devices whose reported data volume does not meet expectations or whose reported data credibility score is lower than a predetermined standard, a partial holographic view is generated and a data incompleteness warning is added. For key data points with missing data in a continuous time sequence, predictive data filling is performed based on the movement trend of the same traffic target in the previous sliding time window, and the filling result is marked as predicted data. When the data completeness score is lower than the second threshold and the data credibility score is lower than the second threshold, a severe degradation strategy is selected. The severe degradation strategy indicates the generation of a simplified traffic situation view or the issuance of a data quality alarm.

[0076] Specifically, the first and second thresholds are key parameters used to distinguish different data quality levels. The first threshold is typically set to a higher value, such as 0.8 or 0.9, to define scenarios with excellent data quality that can be fused with high precision. The second threshold is set to a lower value, such as 0.5 or 0.6, to define scenarios with extremely poor data quality that require severe degradation or alarm measures. These two thresholds can be flexibly configured and adjusted according to the needs of the actual application scenario, the deployment density of roadside sensing devices, and the accuracy requirements of the traffic situation view.

[0077] The standard fusion strategy refers to the fusion method adopted when the data quality reaches a high standard. In this case, the system assumes that all or most data sources are reliable and complete, so it can use complex, computationally intensive, high-precision fusion algorithms, such as multi-sensor fusion algorithms based on Kalman filtering, particle filtering, or deep learning, to extract the most effective information from the data and generate the most detailed and accurate global traffic situation view.

[0078] In practical applications, the degradation fusion strategy is a flexible processing mechanism for data quality that is moderate to low. When the data completeness score or data reliability score fails to reach the first threshold but is still higher than the second threshold, it indicates that the data has a certain degree of missing or unreliability, but is not completely unusable. In this case, the system will prioritize processing the data reported by edge sensing devices with higher quality to ensure the accuracy of core information. For areas with lower quality, a partial view is generated and prompts are provided to avoid misleading users. In addition, to compensate for the lack of key data points, the system will use historical data for predictive filling. For example, based on the vehicle's speed and direction in the previous sliding time window, its position in the current window is predicted, and such predicted data is explicitly marked to distinguish it from actual sensing data.

[0079] Furthermore, the severe degradation strategy serves as the last line of defense against extremely poor data quality. When both the data completeness score and the data reliability score fall below the second threshold, it indicates that the data within the current sliding time window is severely missing and / or highly unreliable. Forcing data fusion at this point could generate a large amount of erroneous information. Therefore, the system will choose to generate a simplified traffic situation view, such as displaying only the congestion level of major roads, or directly issue a data quality alarm to remind operators that the current data is unreliable and should not be used as a basis for decision-making.

[0080] This application's solution addresses the problem of poor fusion performance in traditional methods when data quality fluctuates by introducing multi-level thresholds and a differentiated fusion strategy. Specifically, when both data completeness and data reliability scores are high, it indicates that the roadside perception system is operating well, with comprehensive data coverage and high reliability. At this point, employing a standard fusion strategy can fully utilize all high-quality data and generate the most detailed and accurate global traffic situation view through a high-precision fusion algorithm. This ensures that, under ideal conditions, the system can provide optimal perception results.

[0081] When data quality declines to a certain extent—that is, when the data completeness score or data reliability score falls below the first threshold but rises above the second threshold—the system no longer blindly applies high-precision fusion. Instead, it intelligently switches to a degraded fusion strategy. This strategy effectively avoids contaminating the overall fusion result with low-quality data by prioritizing high-quality data sources, generating partial views and providing hints for low-quality areas, and predictively filling and marking key missing data. Simultaneously, it preserves as much valid information as possible and clearly identifies uncertainties. This tiered processing mechanism allows the system to maintain a certain level of awareness when facing localized data problems, preventing the entire system from crashing or outputting completely unusable results due to partial data issues.

[0082] When data quality deteriorates significantly—that is, when both data completeness and data reliability scores fall below the second threshold—the system switches to a severe degradation strategy. At this point, due to the high degree of data incompleteness and unreliability, any complex data fusion could be counterproductive. Therefore, the system chooses to generate a simplified traffic situation view or issue a direct alert. This not only avoids generating misleading information but also provides users with timely feedback on the true quality of the current data, thus preventing decisions based on erroneous data.

[0083] Through the aforementioned technical solution, this application can dynamically adjust the data fusion strategy based on real-time data integrity and data reliability scores, thereby significantly improving the adaptability and robustness of the roadside holographic perception platform in complex and ever-changing environments. This solution avoids the errors that may result from forcibly performing high-precision fusion when data quality is poor, and also avoids overly conservatively discarding valid data when data quality is acceptable. Through refined strategy selection, the system can always achieve the optimal balance between data quality and fusion effect, ensuring that the generated global traffic situation view has the highest availability and reliability under all circumstances, and clearly identifying the data quality status, providing a more accurate and reliable basis for subsequent traffic management and decision-making.

[0084] In some preferred embodiments, a specific example is given below. Suppose that multiple edge sensing devices, including millimeter-wave radar, high-definition cameras, and lidar, are deployed at a city intersection.

[0085] Scenario 1: On a clear day, all equipment is operating normally, and data reporting is timely and accurate. At this time, the system calculates a data integrity score and a data reliability score of 0.95, far exceeding the preset first threshold of 0.8. The system will select a standard fusion strategy and utilize a deep learning-based multimodal fusion algorithm to perform high-precision fusion of the perceived data from all devices, generating a real-time 3D digital sandbox containing precise locations, speeds, and types of vehicles, pedestrians, and non-motorized vehicles. This sandbox possesses extremely high accuracy and integrity.

[0086] Scenario 2: During a short-term rainfall event in a localized area, the field of view of an edge sensing device (e.g., a camera) is affected, resulting in a decrease in the amount of data it reports and a reduction in the accuracy of some target recognition. In this case, the data completeness score calculated by the system may be 0.75 (below the first threshold of 0.8), while the data reliability score may be 0.85 (above the first threshold of 0.8), or vice versa, or both may be below the first threshold but above the second threshold of 0.5. The system will choose a degraded fusion strategy. Specifically, the system will prioritize fusing high-quality data from other normally operating millimeter-wave radar and lidar. For the affected camera area, the system will generate a partial holographic view and overlay a text prompt indicating "data incomplete" or "predicted data exists" in that area. For example, for a vehicle trajectory that is continuously missing in the camera area, the system will predictively fill in the location based on the vehicle's movement trend (such as speed and direction) within a previous sliding time window, and mark the filled trajectory as predicted data to ensure the continuity of traffic flow display, while clearly informing the user of the data's source attributes.

[0087] Scenario 3: In a severe network failure or large-scale equipment damage event, data reporting from multiple edge sensing devices is interrupted or of extremely poor quality. In this case, the system's calculated data integrity score might be 0.3, and the data reliability score might be 0.4, both below the preset second threshold of 0.5. The system will select a severe degradation strategy. In this situation, the system will not attempt complex fusion but may instead generate a simplified 2D map interface displaying the traffic congestion level of major roads, or directly issue a prominent "Severe Data Quality Warning" message on the interface, advising users to carefully refer to the current view to avoid making incorrect judgments based on highly unreliable data.

[0088] In some of the embodiments described above in this application, although it is proposed to fuse heterogeneous data within a sliding time window according to the selected fusion strategy and generate a global traffic situation view with data quality labels, in practical applications, if the presentation of data quality labels is not intuitive or lacks a unified standard, it may be difficult for users to quickly and accurately understand the reliability and integrity of data in different areas or different traffic targets, thereby affecting decision-making efficiency and the practicality of the system.

[0089] To address this, this application further proposes the above-mentioned method of fusing heterogeneous data within a sliding time window according to a selected fusion strategy to generate a global traffic situation view with data quality indicators. This includes: fusing heterogeneous data within a sliding time window according to a selected fusion strategy to generate a real-time 3D digital sandbox or 2D map interface as a base view; in the base view, based on the data integrity score and data reliability score within the sensing area corresponding to each edge sensing device, and the processing result of the selected fusion strategy, rendering different sensing areas or different traffic targets using differentiated visual markers to indicate their data quality, thereby obtaining the global traffic situation view. Specifically, for perception areas processed using the standard fusion strategy and whose corresponding data integrity score and data confidence score are both higher than the preset first threshold, they are rendered with bright colors and solid lines and marked as high confidence areas; for perception areas processed using the downgraded fusion strategy, they are rendered with semi-transparent colors and dashed lines, and text prompts are superimposed to explain the incompleteness of the data in the area or the presence of predicted data attributes; for perception areas whose data integrity score and data confidence score are both lower than the preset second threshold, they are highlighted with specific colors and marked as severely missing data areas.

[0090] Specifically, generating the base view refers to the system first constructing a real-time visualization interface reflecting the current traffic situation based on the fusion processing results. This interface can be in the form of a 3D digital sandbox, providing a three-dimensional scene display, or a 2D map interface, providing a planar traffic overview. This base view carries the fused traffic data and serves as the carrier for subsequent data quality labeling and rendering. Differentiated visual markers refer to the use of various visual elements such as color, line style, transparency, and text prompts to distinguish and identify different perceptual areas or traffic targets in the base view. Rendering refers to applying these visual markers to the base view, enabling it to present data quality information in an intuitive way.

[0091] Specifically, when both the data completeness score and data reliability score of a certain sensing area are higher than a preset first threshold, and the system processes it using a standard fusion strategy, that area will be rendered as a high-reliability area. For example, a bright green or blue color can be used, and the boundaries of the area or traffic targets within it can be outlined with solid lines to clearly indicate that the data quality in that area is high and the information is reliable. Further, when the data completeness score or data reliability score of a certain sensing area is lower than the first threshold but higher than a preset second threshold, and the system processes it using a downgraded fusion strategy, that area will be rendered as a warning area. For example, a semi-transparent yellow or orange color can be used, and the boundaries of the area or traffic targets within it can be represented with dashed lines. Simultaneously, to provide more detailed contextual information, text prompts such as "Data Incomplete" or "Contains Predictive Data" can be overlaid in this area to inform the user that the data in this area may be partially missing or that some information is based on prediction rather than real-time perception. Additionally, when both the data completeness score and data reliability score of a certain sensing area are lower than a preset second threshold, it indicates that the data quality in that area is severely insufficient. At this point, the system will select a severe degradation strategy and highlight the warning area with a specific color (such as a striking red), marking it as an area with severely missing data. This strong visual cue is designed to immediately draw the user's attention, indicating that traffic situation information in that area may be extremely unreliable or completely missing.

[0092] This application's solution effectively addresses the limitations of traditional traffic situation views in presenting data quality by combining data quality labeling with intuitive visual rendering. Specifically, the system first generates a basic traffic situation view based on the fusion processing results, which includes the fused traffic data. Building upon this, the system no longer simply adds an abstract "data quality label," but dynamically selects and applies differentiated visual markers for rendering based on the data completeness score, data reliability score, and the specific fusion strategy applied (standard fusion strategy, degraded fusion strategy, or severely degraded strategy) for each sensing area or traffic target. For example, high-reliability areas are presented with bright colors and solid lines, allowing users to quickly identify areas with high data quality and reliability; areas with degraded data quality are rendered with semi-transparent colors, dashed lines, and text prompts, clearly informing users that the data may be incomplete or contain predictive data, thus guiding users to consider this in their decision-making; and areas with severely deficient data quality are highlighted with specific colors as warning markers, forcibly reminding users that the data in that area is unreliable. This hierarchical and intuitive visual presentation method transforms data quality information from abstract numerical values ​​into visual elements directly integrated into the traffic situation view, greatly improving users' efficiency in understanding complex traffic data and the accuracy of their decisions.

[0093] Through the aforementioned technical solution, this application significantly enhances the intuitiveness and operability of the global traffic situation view. Users can quickly and accurately determine the data quality level and the fusion strategy employed by visually marking different areas or targets in the view without additional queries or analysis. This not only makes data quality information readily apparent, avoiding misjudgments or inefficient decisions due to data quality issues, but also effectively guides users to adopt different levels of trust and processing methods for traffic information in different areas through differentiated visual cues. For example, users can confidently rely on the information of high-trust areas for decision-making; for downgraded fusion areas, users will realize the need for caution or to seek supplementary information; and for areas with severely missing data, users will immediately identify potential risks and take avoidance measures. This refined data quality visualization management greatly enhances the practical value and user experience of the roadside holographic perception platform in complex and ever-changing traffic environments.

[0094] In some preferred embodiments, it is assumed that in an urban road network, a roadside holographic sensing platform is monitoring the traffic conditions of multiple road segments in real time.

[0095] Specifically, at a key intersection on a major road, due to the deployment of numerous high-performance edge sensing devices and their stable operation, the reported data integrity and reliability scores consistently exceed a preset first threshold. In this case, the system selects a standard fusion strategy to fuse the heterogeneous data in this area. In the generated global traffic situation view, traffic flow, vehicle location, and speed information for this key intersection area will be rendered with a bright green solid line, clearly identifying it as a high-reliability area, indicating extremely high data quality and its reliability as a basis for decision-making.

[0096] Furthermore, for a secondary arterial road undergoing construction, some edge sensing devices may experience intermittent data reporting interruptions due to construction, leading to a slight decrease in the data integrity score for that area, although it remains above the preset second threshold. Simultaneously, the data reliability score of some sensors may also decrease slightly due to potential interference. In this situation, the system will select a downgraded fusion strategy. In the traffic situation view, the traffic information for this construction section will be rendered as a semi-transparent yellow dashed line graphic, overlaid with the text prompt "Data incomplete, contains predicted data." This allows users to intuitively understand that the data in this area is somewhat missing or that some information is based on predictions, thus encouraging greater caution in traffic management or route planning.

[0097] Furthermore, for a remote road in the suburbs, equipment malfunctions or network outages may prevent edge sensing devices in that area from reporting data for an extended period, resulting in data integrity and reliability scores falling below the preset second threshold. In this case, the system will select a severe degradation strategy. In the traffic situation view, this remote road area will be highlighted in bright red as a warning, clearly identified as an area with severely missing data. This strong visual alert immediately alerts traffic managers that the traffic information in this area is extremely unreliable, potentially requiring on-site verification or other emergency measures.

[0098] Through the aforementioned differentiated visual rendering, users can clearly grasp the data quality status of different areas of the entire road network at a glance, thereby making more informed and timely decisions.

[0099] In some embodiments, this application proposes a data synchronization system for a roadside holographic perception platform based on big data, comprising: an acquisition unit, configured to acquire heterogeneous data with time stamps from multiple edge sensing devices, and to aggregate the multiple heterogeneous data into their respective corresponding sliding time windows according to the time stamps carried by each heterogeneous data; and a determination unit, configured to, for each sliding time window, determine a data integrity score characterizing the data coverage within the sliding time window based on a preset expected data source list and the actual amount of data received within the sliding time window, and to determine the data integrity score based on the time stamps of the data packets constituting the heterogeneous data within the sliding time window and the sequence numbers of the data packets themselves. The system determines a data reliability score to characterize the reliability of data within the window, based on the logical relationships between the data, the historical time stamp quality profiles of each edge sensing device, and the consistency of the sensing content of sensors contained in different edge sensing devices for the same traffic target. A selection unit selects a fusion strategy from a set of preset data fusion strategies based on the data integrity score and the data reliability score, wherein the multiple data fusion strategies include at least a standard fusion strategy and a degraded fusion strategy. A generation unit performs fusion processing on the heterogeneous data within the sliding time window according to the selected fusion strategy to generate a global traffic situation view with data quality indicators.

[0100] The system proposed in this application achieves real-time evaluation and adaptive fusion of heterogeneous data quality through the collaborative work of its acquisition unit, determination unit, selection unit, and generation unit. Specifically, the acquisition unit is responsible for efficiently collecting and initially organizing heterogeneous data from edge sensing devices; the determination unit performs multi-dimensional quality evaluation on this data, quantifying its completeness and reliability; the selection unit intelligently selects the most suitable fusion strategy based on these quality evaluation results; finally, the generation unit performs fusion processing on the data according to the selected strategy and generates a global traffic situation view with clear quality indicators. Therefore, this system can effectively address challenges such as distributed clock skew, data loss, and verification misjudgments, ensuring the accuracy and reliability of global traffic situation perception.

[0101] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A data synchronization method for a roadside holographic sensing platform based on big data, characterized in that, The method includes: Acquire time-stamped heterogeneous data from multiple edge sensing devices, and aggregate the multiple heterogeneous data into their respective sliding time windows according to the time stamp carried by each heterogeneous data. For each sliding time window, a data integrity score is determined based on the preset expected data source list and the actual amount of data received within the sliding time window. A data reliability score is determined based on the logical relationship between the timestamp of the data packet constituting the heterogeneous data within the sliding time window and the sequence number of the data packet itself, the historical timestamp quality profile of each edge sensing device, and the consistency of the sensing content of the sensors contained in different edge sensing devices for the same traffic target. Based on the data integrity score and the data credibility score, a fusion strategy is selected from a variety of preset data fusion strategies, wherein the variety of data fusion strategies includes at least a standard fusion strategy and a degraded fusion strategy; The heterogeneous data within the sliding time window are fused according to the selected fusion strategy to generate a global traffic situation view with data quality indicators.

2. The method according to claim 1, characterized in that, The step of acquiring time-stamped heterogeneous data from multiple edge sensing devices and, based on the time stamp carried by each heterogeneous data, grouping the multiple heterogeneous data into their respective corresponding sliding time windows includes: The data acquisition server, based on a high-performance network interface card, continuously receives data packets reported by roadside edge sensing devices. Each data packet contains a nanosecond or microsecond-level time stamp generated by the device's internal clock. The received data packets from different edge sensing devices are used as the heterogeneous data. Each data packet is assigned to a sliding time window with a preset time length according to its timestamp. Each sliding time window is used to aggregate data packets from different edge sensing devices whose timestamps fall within the same continuous time interval.

3. The method according to claim 1, characterized in that, The expected data source list records all edge sensing devices that should report data within the current road segment and time period, as well as the expected data reporting frequency or data volume; for each sliding time window, based on the preset expected data source list and the actual amount of data received within the sliding time window, a data integrity score is determined to characterize the data coverage within the sliding time window, including: Count the number of data packets or the amount of data actually sent by each edge sensing device within the current sliding time window; The ratio of the actual data volume to the expected data volume for each edge sensing device is calculated, and the sum of the ratios for all edge sensing devices is divided by the total number of devices to obtain the data integrity score.

4. The method according to claim 1, characterized in that, The determination of a data reliability score, characterizing the reliability of data within the window, based on the logical relationship between the timestamps of the data packets constituting the heterogeneous data within the sliding time window and the sequence numbers of the data packets themselves, the historical timestamp quality profiles of each edge sensing device, and the consistency of the sensing content of the sensors contained in different edge sensing devices for the same traffic target, includes: Based on the logical consistency between the timestamp of each data packet and its own sequence number within the sliding time window, a timestamp consistency score is assigned to each data packet, and the timestamp consistency score of the window is determined based on the timestamp consistency scores of all data packets within the window. Based on the time stamp deviation and jitter levels recorded in the historical time stamp quality archive of each edge sensing device, a source reliability score is assigned to each edge sensing device, and the source reliability score of the sliding time window is determined based on the source reliability scores of all edge sensing devices that contribute data to the sliding time window. For each traffic target within the sliding time window, a content verification score is assigned to the traffic target based on the degree of matching between the perceived content of the traffic target from different sensors, and the content verification score of the sliding time window is determined based on the content verification scores of all traffic targets within the sliding time window. The time stamp consistency score, source reliability score, and content verification score of the sliding time window are weighted and summed, and the sum is used as the data credibility score.

5. The method according to claim 4, characterized in that, The process of assigning a timestamp consistency score to each data packet based on the logical consistency between the timestamp of each data packet within the sliding time window and the sequence number of the data packet itself, and determining the timestamp consistency score of the window based on the timestamp consistency scores of all data packets within the window, includes: For each data packet within the sliding time window, obtain the logical order indicated by the data packet's timestamp and its own sequence number; Based on the deviation between the timestamp of the data packet and the logical time expected by the logical sequence, determine whether the deviation exceeds a preset timestamp deviation threshold. If the deviation value does not exceed the timestamp deviation threshold, the timestamp consistency score of the data packet is determined to be the full score. If the deviation value exceeds the timestamp deviation threshold, the timestamp consistency score of the data packet is reduced proportionally according to the proportion of the deviation value exceeding the timestamp deviation threshold. The time stamp consistency scores of all data packets within the sliding time window are summarized to obtain the time stamp consistency score of the window.

6. The method according to claim 4, characterized in that, The process involves assigning a source reliability score to each edge sensing device based on the time stamp deviation and jitter levels recorded in the historical time stamp quality archive of each device, and determining the source reliability score of the sliding time window based on the source reliability scores of all edge sensing devices contributing data to the sliding time window, including: For each edge sensing device, obtain the average time stamp deviation and time stamp jitter value of the device over the past time period from the historical time stamp quality archive; Based on the average deviation value of the time stamp and the jitter value of the time stamp, the source reliability score of the device is calculated according to the preset first weighting coefficient used to characterize the degree of influence of the deviation and the second weighting coefficient used to characterize the degree of influence of the jitter. The larger the average deviation value or the larger the jitter value, the lower the calculated source reliability score. The source reliability scores of all contributing edge sensing devices within the sliding time window are summarized to obtain the source reliability score of the sliding time window.

7. The method according to claim 4, characterized in that, For each traffic target within the sliding time window, a content verification score is assigned to the traffic target based on the matching degree between the perceived content of the traffic target from different sensors. The content verification score of the sliding time window is then determined based on the content verification scores of all traffic targets within the sliding time window, including: For each traffic target detected within the sliding time window, obtain perception reports of the traffic target from at least two different sensors. Each perception report includes the traffic target's spatial location, speed, and target type attributes. Calculate the deviation value of the same attribute in the perception reports of different sensors, wherein the deviation value includes at least the position deviation value and the velocity deviation value; Determine whether the position deviation value exceeds a preset allowable position deviation threshold and whether the speed deviation value exceeds a preset allowable speed deviation threshold, respectively. The content verification score of the traffic target is determined based on the degree to which the deviation value exceeds the corresponding allowable deviation threshold. When any attribute deviation value exceeds the allowable threshold, the content verification score of the traffic target is reduced accordingly. The content verification scores of all traffic targets within the sliding time window are summarized to obtain the content verification score of the sliding time window.

8. The method according to claim 1, characterized in that, The step of selecting a fusion strategy from a set of preset data fusion strategies based on the data integrity score and the data credibility score includes: When both the data integrity score and the data credibility score are higher than a preset first threshold, a standard fusion strategy is selected. The standard fusion strategy indicates that a high-precision fusion algorithm is used to fuse all heterogeneous data within the current sliding time window. When the data integrity score or the data credibility score is lower than the first threshold and higher than the preset second threshold, a downgrade fusion strategy is selected. The degradation fusion strategy includes: prioritizing the fusion of data reported by edge sensing devices whose data integrity score and data credibility score are both higher than the predetermined standard; generating a partial holographic view and attaching a data incompleteness prompt message for the sensing area corresponding to the edge sensing device whose data reporting volume does not meet the expectation or whose data credibility score is lower than the predetermined standard; and for key data points with missing data in a continuous time sequence, performing predictive data filling based on the movement trend of the same traffic target in the previous sliding time window and marking the filling result as predicted data. When the data integrity score is lower than the second threshold and the data credibility score is lower than the second threshold, a severe degradation strategy is selected, which indicates that a simplified traffic situation view is generated or a data quality alarm is issued.

9. The method according to claim 1, characterized in that, The step of fusing heterogeneous data within the sliding time window according to the selected fusion strategy to generate a global traffic situation view with data quality indicators includes: The heterogeneous data within the sliding time window are fused according to the selected fusion strategy to generate a real-time 3D digital sand table or 2D map interface as the base view. In the basic view, based on the data integrity score and data reliability score of the sensing area corresponding to each edge sensing device, as well as the processing result of the selected fusion strategy, different visual labels are used to render different sensing areas or different traffic targets to identify their data quality, thus obtaining the global traffic situation view. Among them, for the perception area processed by the standard fusion strategy and whose corresponding data integrity score and data confidence score are both higher than the preset first threshold, it is rendered with bright colors and solid line graphics and marked as a high confidence area. For the perception area processed using the downgrade fusion strategy, a semi-transparent color and dashed line graphic are used for rendering, and a text prompt is superimposed to explain the incompleteness of the data in the area or the existence of predicted data attributes. For perception areas where both the data integrity score and the data credibility score are lower than the preset second threshold, a specific color is used to highlight and alert the area, marking it as a severely missing data area.

10. A data synchronization system for a roadside holographic sensing platform based on big data, characterized in that, include: The acquisition unit is used to acquire time-stamped heterogeneous data from multiple edge sensing devices, and to aggregate the multiple heterogeneous data into their respective sliding time windows according to the time stamp carried by each heterogeneous data. The determining unit is used to determine, for each sliding time window, a data integrity score that characterizes the data coverage within the sliding time window based on a preset list of expected data sources and the actual amount of data received within the sliding time window; and a data reliability score that characterizes the data reliability within the sliding time window based on the logical relationship between the timestamps of the data packets constituting the heterogeneous data within the sliding time window and the sequence numbers of the data packets themselves, the historical timestamp quality profiles of each edge sensing device, and the consistency of the sensing content of the sensors contained in different edge sensing devices for the same traffic target. The selection unit is used to select a fusion strategy from a variety of preset data fusion strategies based on the data integrity score and the data credibility score, wherein the variety of data fusion strategies includes at least a standard fusion strategy and a degraded fusion strategy; The generation unit is used to perform fusion processing on heterogeneous data within the sliding time window according to the selected fusion strategy, and generate a global traffic situation view with data quality identifiers.

Images