Vehicle end trajectory data desensitization method and device, storage medium and electronic device

Abstract

The application provides a vehicle end trajectory data desensitization method and device, a storage medium and an electronic device, wherein the method comprises: acquiring a driving trajectory sequence collected by a positioning module of a vehicle; identifying the privacy sensitivity of the driving trajectory sequence; constructing a feature vector sequence using the driving trajectory sequence and the privacy sensitivity; generating a perturbation offset sequence in a state space according to the feature vector sequence, wherein the sequence length of the perturbation offset sequence is the same as the sequence length of the driving trajectory sequence; desensitizing the driving trajectory sequence using the perturbation offset sequence to obtain a target trajectory sequence after desensitization, and replacing the driving trajectory sequence with the target trajectory sequence. Through this embodiment, the technical problem of discontinuous trajectory semantics caused by trajectory data desensitization in the prior art is solved, and a more intelligent, controllable and safe trajectory desensitization solution is provided for intelligent networked vehicle trajectory collection, data uploading and privacy compliance tasks.

Description

Technical Field

[0001] This invention relates to the field of data security technology, and more specifically, to a method and apparatus for desensitizing vehicle trajectory data, a storage medium, and an electronic device. Background Technology

[0002] In related technologies, vehicle trajectory data serves as a crucial spatiotemporal information resource in typical application scenarios such as intelligent connected vehicles, autonomous driving, and high-precision map updates, undertaking important functions such as navigation and positioning, crowdsourced data collection, and traffic analysis. Vehicles continuously collect their own position, speed, timestamps, and other information through onboard positioning modules, forming a continuous sequence of trajectory points, which is then used for tasks such as path reconstruction, driving behavior modeling, and map update verification.

[0003] With the evolution of vehicle communication systems and data platform architectures, an increasing amount of trajectory data is being uploaded to the cloud in real time, either locally within the vehicle or during vehicle-to-infrastructure (V2I) communication. However, as a high-frequency, time-series, and spatial information carrier, trajectory data naturally contains sensitive information such as the user's activity range, location, and travel patterns. Direct transmission without processing poses serious privacy risks. Therefore, effectively and dynamically anonymizing trajectory data on the vehicle side has become one of the key technical challenges and critical requirements in the current data governance of intelligent connected vehicles.

[0004] Currently, research on trajectory privacy protection mainly focuses on three types of methods: perturbation, generalization, and replacement. Perturbation methods typically add noise to the spatial coordinates of trajectory points to achieve positional ambiguity; generalization methods use grid mapping and other techniques to blur the original trajectory to a coarser spatial resolution; and replacement methods replace the original data entirely by matching similar or template trajectories. While these methods offer protection to varying degrees, their applicability is relatively limited and they cannot yet meet the comprehensive requirements of real-time processing, precision, and deployment feasibility for trajectory desensitization in intelligent connected vehicles.

[0005] Most existing methods rely on fixed rules and lack the ability to deeply model trajectory structure and contextual features, failing to flexibly adjust the desensitization intensity according to the actual environment and user behavior patterns. For example, there are significant differences in privacy sensitivity between urban main roads and residential alleys. If a fixed-intensity perturbation strategy is used uniformly, excessive perturbation in low-sensitivity areas may cause data distortion, while insufficient desensitization in high-sensitivity areas may leave privacy risks. Furthermore, some methods model trajectory desensitization as a point-level operation, ignoring the continuity and semantic consistency of the trajectory, affecting downstream applications such as map matching and path reconstruction.

[0006] No efficient and accurate solution has yet been found to address the aforementioned issues in the relevant technologies. Summary of the Invention

[0007] This invention provides a method, apparatus, storage medium, and electronic device for desensitizing vehicle trajectory data, in order to solve technical problems in related technologies.

[0008] According to an embodiment of the present invention, a method for desensitizing vehicle trajectory data is provided, comprising: acquiring a driving trajectory sequence collected by a vehicle's positioning module; identifying the privacy sensitivity of the driving trajectory sequence; constructing a feature vector sequence using the driving trajectory sequence and the privacy sensitivity; generating a perturbation offset sequence in the state space based on the feature vector sequence, wherein the sequence length of the perturbation offset sequence is the same as the sequence length of the driving trajectory sequence; performing desensitization processing on the driving trajectory sequence using the perturbation offset sequence to obtain a desensitized target trajectory sequence, and replacing the driving trajectory sequence with the target trajectory sequence.

[0009] Optionally, obtaining the driving trajectory sequence collected by the vehicle's positioning module includes: periodically obtaining a set of trajectory point data collected by the vehicle's positioning module at each moment, wherein each trajectory point data in the trajectory point data set includes the coordinate data of the trajectory point, the vehicle speed, and a timestamp; sorting the trajectory point data set according to the timestamp to obtain the driving trajectory sequence.

[0010] Optionally, identifying the privacy sensitivity of the driving trajectory sequence includes: for each trajectory point in the driving trajectory sequence, obtaining the location sensitivity of the location to which the trajectory point belongs, obtaining the region sensitivity of the area where the trajectory point is located, and obtaining the state sensitivity of the vehicle state of the trajectory point; and calculating the privacy sensitivity of the trajectory point by weighting the location sensitivity, the region sensitivity, and the state sensitivity.

[0011] Optionally, obtaining the location sensitivity of the trajectory point includes: locating the geographical location of the trajectory point in map data; if the geographical location is an external road, determining the location sensitivity of the trajectory point to a first privacy level; if the geographical location is an internal road, determining the location sensitivity of the trajectory point to a second privacy level; if the geographical location is a building, determining the location sensitivity of the trajectory point to a third privacy level, wherein the location sensitivity of the first privacy level is less than the location sensitivity of the second privacy level, and the location sensitivity of the second privacy level is less than the location sensitivity of the third privacy level.

[0012] Optionally, obtaining the regional sensitivity of the area where the trajectory point is located includes: finding the number of points of interest (POIs) within a fixed radius area in the map data with the trajectory point as the center; calculating the POI density based on the number of POIs and the area, wherein the POI density is positively correlated with the regional sensitivity.

[0013] Optionally, obtaining the state sensitivity of the vehicle state of the trajectory point includes: obtaining the vehicle speed of the trajectory point; determining whether the vehicle speed is less than a preset speed; if the vehicle speed is less than the preset speed and remains so for a preset duration, determining that the vehicle state of the trajectory point is a stationary state, and assigning a first state sensitivity to the trajectory point; if the vehicle speed is greater than or equal to the preset speed, determining that the vehicle state of the trajectory point is a moving state, and assigning a second state sensitivity to the trajectory point, wherein the first state sensitivity is greater than the second state sensitivity.

[0014] Optionally, constructing a feature vector sequence using the driving trajectory sequence and the privacy sensitivity includes: constructing the following trajectory feature vectors for each time step in the driving trajectory sequence. : ,in, These represent the horizontal axis data, vertical axis data, and vehicle speed at the time step, respectively. The sensitivity values ​​for the time step are position sensitivity, region sensitivity, state sensitivity, and privacy sensitivity, respectively. The trajectory feature vectors of all time steps are sorted by time to obtain a feature vector sequence.

[0015] Optionally, generating a perturbation offset sequence in the state space based on the feature vector sequence includes: calculating the perturbation offset at time step t in the perturbation offset sequence using the following formula in the Mamba model. : ;in, It is a sequence of feature vectors. , Let be the state vectors of the Mamba model at time step t and time step t-1, respectively. Let be the state transition matrix of the Mamba model, representing the state transition relationship of the state vector in the time dimension. Let be the control matrix of the Mamba model, representing the influence of the feature vector sequence on the state vector update. The observation matrix of the Mamba model describes the linear mapping relationship between the state vector and the perturbation offset.

[0016] Optionally, the perturbation offset at time step t in the perturbation offset sequence is calculated in the Mamba model using the following formula. Previously, the method also included: constructing sample data using original trajectory samples and reference perturbation labels; and employing the following loss function. The model parameters for training the Mamba model are:

[0017] ;

[0018] in The reference perturbation amount of the reference perturbation tag at time step t. , These are the feature vectors of the original trajectory sample at time step t and time step t-1, respectively. The first balance parameter is used to constrain the disturbance intensity error, and the second balance parameter is used to constrain the disturbance smoothness. The model parameters include the state transition matrix, the control matrix, and the observation matrix.

[0019] Optionally, desensitizing the driving trajectory sequence using the disturbance offset sequence includes: desensitizing the driving trajectory sequence using the following formula to obtain the desensitized target trajectory sequence. : ;in, The coordinate data of the trajectory point at time t in the driving trajectory sequence. It is the offset at time t in the perturbation offset sequence.

[0020] According to another embodiment of the present invention, a desensitization device for vehicle trajectory data is provided, comprising: an acquisition module for acquiring a driving trajectory sequence collected by a vehicle positioning module; an identification module for identifying the privacy sensitivity of the driving trajectory sequence; a construction module for constructing a feature vector sequence using the driving trajectory sequence and the privacy sensitivity; a generation module for generating a perturbation offset sequence in a state space based on the feature vector sequence, wherein the sequence length of the perturbation offset sequence is the same as the sequence length of the driving trajectory sequence; and a desensitization module for performing desensitization processing on the driving trajectory sequence using the perturbation offset sequence to obtain a desensitized target trajectory sequence, and replacing the driving trajectory sequence with the target trajectory sequence.

[0021] Optionally, the acquisition module includes: an acquisition unit, configured to acquire a set of trajectory point data collected by the vehicle's positioning module at each time point according to a period, wherein each trajectory point data in the trajectory point data set includes the coordinate data of the trajectory point, the vehicle speed, and a timestamp; and a sorting unit, configured to sort the trajectory point data set according to the timestamp to obtain a driving trajectory sequence.

[0022] Optionally, the identification module includes: an acquisition unit, configured to acquire, for each trajectory point in the driving trajectory sequence, the location sensitivity of the location to which the trajectory point belongs, the region sensitivity of the region where the trajectory point is located, and the state sensitivity of the vehicle state of the trajectory point; and a calculation unit, configured to calculate the privacy sensitivity of the trajectory point by weighting the location sensitivity, the region sensitivity, and the state sensitivity.

[0023] Optionally, the acquisition unit includes: a positioning subunit, configured to locate the geographical location of the trajectory point in map data; and a determination subunit, configured to determine the location sensitivity of the trajectory point to a first privacy level if the geographical location is an external road; determine the location sensitivity of the trajectory point to a second privacy level if the geographical location is an internal road; and determine the location sensitivity of the trajectory point to a third privacy level if the geographical location is a building, wherein the location sensitivity of the first privacy level is less than the location sensitivity of the second privacy level, and the location sensitivity of the second privacy level is less than the location sensitivity of the third privacy level.

[0024] Optionally, the acquisition unit includes: a search subunit, used to search for the number of points of interest within a fixed radius area in the map data, with the trajectory point as the center; and a calculation subunit, used to calculate the point of interest density based on the number of points of interest and the area, wherein the point of interest density is positively correlated with the area sensitivity.

[0025] Optionally, the acquisition unit includes: an acquisition subunit for acquiring the vehicle speed of the trajectory point; a judgment subunit for judging whether the vehicle speed is less than a preset speed; and an allocation subunit for determining that if the vehicle speed is less than the preset speed and remains so for a preset duration, the vehicle state of the trajectory point is a stationary state, and a first state sensitivity is allocated to the trajectory point; if the vehicle speed is greater than or equal to the preset speed, the vehicle state of the trajectory point is a moving state, and a second state sensitivity is allocated to the trajectory point, wherein the first state sensitivity is greater than the second state sensitivity.

[0026] Optionally, the construction module includes: a construction unit, configured to construct the following trajectory feature vector for each time step in the driving trajectory sequence. : ,in, These represent the horizontal axis data, vertical axis data, and vehicle speed at the time step, respectively. These are respectively the position sensitivity, region sensitivity, state sensitivity, and privacy sensitivity of the time step; the sorting unit is used to sort the trajectory feature vectors of all time steps according to time to obtain a feature vector sequence.

[0027] Optionally, the generation module includes: a calculation unit, used to calculate the perturbation offset at time step t in the perturbation offset sequence in the Mamba model using the following formula. : ;in, It is a sequence of feature vectors. , Let be the state vectors of the Mamba model at time step t and time step t-1, respectively. Let be the state transition matrix of the Mamba model, representing the state transition relationship of the state vector in the time dimension. Let be the control matrix of the Mamba model, representing the influence of the feature vector sequence on the state vector update. The observation matrix of the Mamba model describes the linear mapping relationship between the state vector and the perturbation offset.

[0028] Optionally, the generation module further includes: a construction unit, configured to calculate the perturbation offset at time step t in the perturbation offset sequence in the Mamba model using the following formula in the calculation unit. Previously, sample data was constructed using original trajectory samples and reference perturbation labels; training units were used to employ the following loss function. The model parameters for training the Mamba model are: ;in The reference perturbation amount of the reference perturbation tag at time step t. , These are the feature vectors of the original trajectory sample at time step t and time step t-1, respectively. The first balance parameter is used to constrain the disturbance intensity error, and the second balance parameter is used to constrain the disturbance smoothness. The model parameters include the state transition matrix, the control matrix, and the observation matrix.

[0029] Optionally, the desensitization module includes: a desensitization unit, used to desensitize the driving trajectory sequence using the following formula to obtain the desensitized target trajectory sequence. : ;in, The coordinate data of the trajectory point at time t in the driving trajectory sequence. It is the offset at time t in the perturbation offset sequence.

[0030] According to another aspect of the embodiments of this application, a storage medium is also provided, the storage medium including a stored program that executes the above steps when the program is run.

[0031] According to another aspect of the embodiments of this application, an electronic device is also provided, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; wherein: the memory is used to store computer programs; and the processor is used to execute the steps in the above method by running the programs stored in the memory.

[0032] This application also provides a computer program product containing instructions that, when run on a computer, cause the computer to perform the steps in the above-described method.

[0033] The beneficial effects of this invention are:

[0034] 1. Applicable to dynamic privacy protection processing of trajectory data in actual road scenarios, meeting the regulatory requirements of data collection and desensitization and transmission compliance, solving the technical problem of low regulatory efficiency of geographic information data in existing technologies, improving the rationality and semantic consistency of trajectory disturbance, and providing a more intelligent, controllable and secure trajectory desensitization solution for intelligent connected vehicles in tasks such as trajectory collection, data uploading and privacy compliance;

[0035] 2. While ensuring the continuity and availability of the trajectory, it significantly enhances the ability to protect against disturbances to the trajectory in sensitive areas. Attached Figure Description

[0036] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0037] Figure 1 This is a hardware structure block diagram of a vehicle according to an embodiment of the present invention;

[0038] Figure 2 This is a flowchart of a method for desensitizing vehicle trajectory data according to an embodiment of the present invention;

[0039] Figure 3 This is a schematic diagram of the Mamba model in an embodiment of the present invention;

[0040] Figure 4 This is a flowchart of the vehicle trajectory dynamic desensitization method based on state space modeling in this embodiment of the invention;

[0041] Figure 5 This is a structural block diagram of a device for desensitizing vehicle trajectory data according to an embodiment of the present invention. Detailed Implementation

[0042] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, and not all of them. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present application. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of the present application can be combined with each other.

[0043] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0044] Example 1

[0045] The method embodiment provided in Embodiment 1 of this application can be executed in a vehicle, server, processor, computer, or similar processing device. Taking its operation in a vehicle as an example, Figure 1 This is a hardware structure block diagram of a vehicle according to an embodiment of the present invention. For example... Figure 1 As shown, a vehicle may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. Optionally, the vehicle may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the vehicle described above. For example, the vehicle may also include components that are larger than... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0046] The memory 104 can be used to store vehicle programs, such as application software programs and modules, like the vehicle program corresponding to a method for desensitizing vehicle trajectory data in an embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the vehicle program stored in the memory 104, thereby implementing the aforementioned method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the vehicle via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0047] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the vehicle's communication provider. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module used for wireless communication with the Internet.

[0048] This embodiment provides a method for de-identifying vehicle trajectory data. Figure 2 This is a flowchart of a method for desensitizing vehicle trajectory data according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps:

[0049] Step S201: Obtain the driving trajectory sequence collected by the vehicle's positioning module;

[0050] Optionally, during normal vehicle operation or map collection, trajectory points are collected periodically to form a driving trajectory sequence.

[0051] Step S202: Identify the privacy sensitivity of the driving trajectory sequence;

[0052] Step S203: Construct a feature vector sequence using the driving trajectory sequence and the privacy sensitivity;

[0053] The feature vector sequence includes multiple trajectory feature vectors, each containing trajectory point data and privacy sensitivity information for the corresponding trajectory point.

[0054] Step S204: Generate a disturbance offset sequence in the state space based on the feature vector sequence, wherein the sequence length of the disturbance offset sequence is the same as the sequence length of the driving trajectory sequence;

[0055] Step S205: The driving trajectory sequence is desensitized using the disturbance offset sequence to obtain the desensitized target trajectory sequence, and the driving trajectory sequence is replaced with the target trajectory sequence.

[0056] The target trajectory sequence can then be uploaded to a cloud server for updating the crowdsourced map, training autonomous driving models, and other purposes.

[0057] Through the above steps, the driving trajectory sequence collected by the vehicle's positioning module is obtained; the privacy sensitivity of the driving trajectory sequence is identified; a feature vector sequence is constructed using the driving trajectory sequence and the privacy sensitivity; a disturbance offset sequence is generated in the state space based on the feature vector sequence, wherein the sequence length of the disturbance offset sequence is the same as the sequence length of the driving trajectory sequence; the driving trajectory sequence is desensitized using the disturbance offset sequence to obtain the desensitized target trajectory sequence, and the driving trajectory sequence is replaced with the target trajectory sequence. By introducing the privacy sensitivity of the driving trajectory sequence, the rationality and semantic consistency of the trajectory disturbance are improved, solving the technical problem of discontinuous trajectory semantics caused by the desensitization of trajectory data in existing technologies. This method is suitable for dynamic privacy protection processing of trajectory data in actual road scenarios, meeting the regulatory requirements of data collection and desensitization, and transmission and compliance. It provides a more intelligent, controllable, and secure trajectory desensitization solution for intelligent connected vehicles in tasks such as trajectory collection, data uploading, and privacy compliance.

[0058] In one embodiment of this example, obtaining the driving trajectory sequence collected by the vehicle's positioning module includes: periodically acquiring a set of trajectory point data collected by the vehicle's positioning module at each moment, wherein each trajectory point data in the trajectory point data set includes the coordinate data of the trajectory point, the vehicle speed, and a timestamp; sorting the trajectory point data set according to the timestamp to obtain the driving trajectory sequence.

[0059] A trajectory acquisition module can be deployed on the vehicle to continuously collect Global Navigation Satellite System (GNSS) trajectory data, forming a structured trajectory sequence. During vehicle operation, GNSS trajectory points are collected at a frequency of 10Hz to form the raw trajectory sequence. The trajectory points at each time point t contain coordinates, velocity, and timestamp information, organized into a structured data format. The trajectory points contain coordinates timestamp ,speed These basic fields constitute the original trajectory point dataset: .

[0060] In this embodiment, identifying the privacy sensitivity of the driving trajectory sequence includes: for each trajectory point in the driving trajectory sequence, obtaining the location sensitivity of the location to which the trajectory point belongs, obtaining the region sensitivity of the area where the trajectory point is located, and obtaining the state sensitivity of the vehicle state of the trajectory point; and calculating the privacy sensitivity of the trajectory point by weighting the location sensitivity, the region sensitivity, and the state sensitivity.

[0061] Optionally, the location sensitivity, the region sensitivity, and the state sensitivity are respectively One value is used, and through weighted calculation, the final privacy sensitivity is obtained.

[0062] In one example, obtaining the location sensitivity of the trajectory point includes: locating the geographical location of the trajectory point in map data; if the geographical location is an external road, determining the location sensitivity of the trajectory point to a first privacy level; if the geographical location is an internal road, determining the location sensitivity of the trajectory point to a second privacy level; if the geographical location is a building, determining the location sensitivity of the trajectory point to a third privacy level, wherein the location sensitivity of the first privacy level is less than the location sensitivity of the second privacy level, and the location sensitivity of the second privacy level is less than the location sensitivity of the third privacy level.

[0063] In this example, location sensitivity is obtained based on road level: coordinates are queried using the map API or a local high-precision map. The road's classification level is adjusted based on whether it's a residential area, government building area, or similar building zone. Optionally, the location sensitivity is 0 for the first privacy level, 0.5 for the second privacy level, and 0.8 for the third privacy level.

[0064] In one example, obtaining the regional sensitivity of the area where the trajectory point is located includes: finding the number of points of interest (POIs) within a fixed radius area in the map data with the trajectory point as the center; calculating the POI density based on the number of POIs and the area, wherein the POI density is positively correlated with the regional sensitivity.

[0065] In this example, the sensitivity of a region is obtained based on the density of Points of Interest (POI). A neighborhood with a radius of r is constructed with the coordinates as the center, the number of points of interest in the neighborhood is counted, and the density is calculated. The more POIs are clustered together, the higher the risk of privacy exposure.

[0066] In one example, the state sensitivity of obtaining the vehicle state of the trajectory point includes: obtaining the vehicle speed of the trajectory point; determining whether the vehicle speed is less than a preset speed; if the vehicle speed is less than the preset speed and remains so for a preset duration, determining that the vehicle state of the trajectory point is a stationary state, and assigning a first state sensitivity to the trajectory point; if the vehicle speed is greater than or equal to the preset speed, determining that the vehicle state of the trajectory point is a moving state, and assigning a second state sensitivity to the trajectory point, wherein the first state sensitivity is greater than the second state sensitivity.

[0067] In this example, state sensitivity is calculated based on the StayFlag and driving status: when the vehicle's speed is less than 0.3 m / s for several consecutive moments, it is identified as a suspected stop point, and the vehicle's state is "stayed." These points are typically associated with residences or workplaces, and stay points are more likely to expose information such as residence, so they should be given higher weight. Optionally, the first state sensitivity is 0.8, and the second state sensitivity is 0.3.

[0068] Finally, a sensitivity function is constructed using the above three indicators (location sensitivity, region sensitivity, and state sensitivity): ;

[0069] Among them, the function This is implemented using a weighted normalized linear combination, outputting the sensitivity score at each time point. This serves as one of the decision inputs for subsequent perturbation decisions.

[0070] In one embodiment of this example, constructing a feature vector sequence using the driving trajectory sequence and the privacy sensitivity includes: constructing the following trajectory feature vector for each time step in the driving trajectory sequence. : ,in, These represent the horizontal axis data, vertical axis data, and vehicle speed at the time step, respectively. The sensitivity values ​​for the time step are position sensitivity, region sensitivity, state sensitivity, and privacy sensitivity, respectively. The trajectory feature vectors of all time steps are sorted by time to obtain a feature vector sequence.

[0071] In this embodiment, the feature modeling module constructs a feature vector sequence by combining the driving trajectory sequence with the privacy sensitivity of the sensitivity score, which is used for perturbation generation model modeling and training. This module uses time steps... Construct eigenvectors for each unit:

[0072] ;

[0073] The complete sequence of feature vectors is: This module completes the structured fusion and normalization of trajectory and semantic features, and is the core input interface for the training and inference of the desensitized model.

[0074] In one embodiment of this example, generating a perturbation offset sequence in the state space based on the feature vector sequence includes: calculating the perturbation offset at time step t in the perturbation offset sequence using the following formula in the Mamba model. : ;in, It is a sequence of feature vectors. , Let be the state vectors of the Mamba model at time step t and time step t-1, respectively. Let be the state transition matrix of the Mamba model, representing the state transition relationship of the state vector in the time dimension. Let be the control matrix of the Mamba model, representing the influence of the feature vector sequence on the state vector update. Let A, B, and C be the observation matrix of the Mamba model, describing the linear mapping relationship between the state vector and the perturbation offset. A, B, and C are parameter matrices in the Mamba model, which are linear mapping matrices matching the hidden state dimension of the state vector and the input dimension of the feature vector sequence.

[0075] In one example, generating a perturbation offset sequence in the state space based on the feature vector sequence includes: for each trajectory feature vector in the feature vector sequence, reading the privacy sensitivity of the trajectory feature vector, assigning a random range to the value of the privacy sensitivity, wherein the random range is positively correlated with the value of the privacy sensitivity, randomly selecting a coordinate offset within the random range, and outputting the coordinate offset as the perturbation offset of the coordinate data of the trajectory point at the corresponding time in the trajectory feature vector.

[0076] The perturbation generation module is based on state-space modeling technology. To achieve efficient modeling and real-time generation of trajectory perturbation sequences, this module introduces a deep neural network structure based on state-space modeling—the Mamba model. The Mamba model treats trajectory features as temporal inputs and establishes an evolution mechanism for trajectory states, significantly improving the model's ability to express trajectory change patterns. This module uses feature vector sequences... As input, the output is a perturbation offset sequence of equal length: Each disturbance offset is , representing the magnitude of the perturbation on the original coordinates. While maintaining lightweight design and causal modeling capabilities, the Mamba structure can perceive long-term dependencies and sensitivity change trends, achieving efficient generation of temporal perturbations. Figure 3 This is a schematic diagram of the Mamba model in this embodiment of the invention. A represents the state transition matrix, B represents the control matrix, C represents the observation matrix, ※ represents matrix multiplication, the input is the trajectory feature vector of the driving trajectory sequence, the output is the disturbance offset, and the solid line is the data flow channel.

[0077] The Mamba model in this embodiment is based on a deep learning architecture designed for state-space models, used to output high-quality perturbation trajectory points. The perturbation generation module is implemented using the Mamba structure, which is an efficient state-space modeling method that combines global dependency modeling capabilities with deployment efficiency on edge devices.

[0078] Optionally, the perturbation offset at time step t in the perturbation offset sequence is calculated in the Mamba model using the following formula. Previously, it also included: constructing sample data using original trajectory samples and reference perturbation labels; and using the following loss function. The model parameters for training the Mamba model are: ;in The reference perturbation amount of the reference perturbation tag at time step t. , These are the feature vectors of the original trajectory sample at time step t and time step t-1, respectively. The first balance parameter is used to constrain the disturbance intensity error, and the second balance parameter is used to constrain the disturbance smoothness. The model parameters include the state transition matrix, the control matrix, and the observation matrix.

[0079] In the process of generating sample feature vectors, it is similar to the above embodiment, except that the input data source is the pre-acquired original trajectory sample, while in the real-time inference process, the input source is the driving trajectory sequence collected by the vehicle in real time.

[0080] By introducing a gating mechanism and parameter sharing strategy, the Mamba model achieves linear complexity modeling, making it suitable for high-frequency trajectory perturbation generation tasks. The module ultimately outputs a perturbation offset sequence, used to adjust the original coordinates and achieve position perturbation desensitization. To ensure the practicality and controllability of the perturbation strategy, the Mamba model is trained using an end-to-end supervised learning approach. Training data is collected and labeled by the automaker under the guidance of compliance support units, forming paired samples of the original trajectory and the reference desensitized trajectory. The training objective is to fit the reference perturbation while maintaining the smoothness of the perturbation sequence.

[0081] In one example of this embodiment, desensitizing the driving trajectory sequence using the disturbance offset sequence includes: desensitizing the driving trajectory sequence using the following formula to obtain the desensitized target trajectory sequence. : ;in, The coordinate data of the trajectory point at time t in the driving trajectory sequence. It is the offset at time t in the perturbation offset sequence.

[0082] The trajectory output module applies the perturbation results to the original trajectory to generate desensitized trajectory points: and output the complete target trajectory sequence. This module is used for compliance reporting, map updates, or other downstream applications. It supports optional output of the original trajectory and the disturbed trajectory, meeting the customized requirements of different systems for original data retention strategies. The disturbance generation process is executed entirely locally on the vehicle. The anonymized trajectory data does not contain original trajectory and disturbance parameter information and does not support reverse recovery, thus possessing irreversible anonymization capabilities. It is suitable for application scenarios with high requirements for data privacy protection and compliance.

[0083] After model training is completed, the automaker solidifies the model parameters and manages the version, then deploys the model to the vehicle's edge computing unit through a secure channel. In actual operation, disturbance prediction is completed immediately after trajectory acquisition, and the post-disturbance coordinates are output. The entire processing is completed independently on the vehicle side, avoiding the uploading of raw sensitive data and achieving compliance and anonymization from the source.

[0084] This embodiment provides a method and apparatus for desensitizing vehicle trajectory data, a storage medium, and electronic devices. The aim is to achieve real-time sensitivity perception and desensitization perturbation generation of vehicle trajectory data at the vehicle end, ensuring location privacy and security while meeting data compliance requirements. The method consists of a trajectory acquisition module, a sensitivity scoring module, a feature modeling module, a perturbation generation module, and a trajectory output module. These modules work collaboratively to form a deployable, dynamically adaptive closed-loop trajectory desensitization process at the vehicle end. Specifically, the trajectory acquisition module continuously collects GNSS trajectory data at the vehicle end, forming a structured driving trajectory sequence. The sensitivity scoring module identifies the privacy sensitivity of each trajectory point. The feature modeling module constructs a feature vector sequence by combining the trajectory sequence and the sensitivity score. The perturbation generation module, based on state-space modeling technology, efficiently models and generates the trajectory perturbation sequence in real time. The trajectory output module applies the perturbation results to the original trajectory to generate desensitized trajectory points. Figure 4 This is a flowchart of a vehicle trajectory dynamic desensitization method based on state-space modeling in this invention, including: acquiring original vehicle trajectory data and performing format standardization processing to ensure temporal continuity and field integrity; extracting road level, POI density, and dwell state features for each trajectory point to construct a privacy sensitivity score, used to characterize the desensitization priority of the trajectory point; fusing basic trajectory attributes and sensitivity scores to generate a temporal feature vector sequence containing location information, speed information, and privacy features; inputting the feature sequence into a pre-trained Mamba network within the enterprise to generate a perturbation offset sequence; and performing dynamic perturbation operations on the original trajectory coordinates based on the model output to generate a desensitized trajectory that meets compliance requirements.

[0085] This embodiment fully integrates a semantically driven sensitivity assessment mechanism with state-space modeling sequence generation capabilities in its architecture design, constructing a trajectory perturbation generation scheme for real-time desensitization needs. It achieves efficient perturbation sequence prediction through the Mamba model, completing trajectory privacy processing and real-time output with low computational overhead on the vehicle side. Compared to traditional desensitization methods based on randomized perturbations or static fuzziness, this method, through feature-driven and model integration, not only improves the rationality and semantic consistency of trajectory perturbations but also possesses controllable perturbation strength and deployment adaptability. This provides a more intelligent, controllable, and secure trajectory desensitization solution for intelligent connected vehicles in tasks such as trajectory acquisition, data uploading, and privacy compliance.

[0086] The proposed solution has been validated in a simulation environment. During the test, a complete dynamic trajectory desensitization module was deployed and continuously run for approximately 30 minutes in a typical urban road scenario, collecting and processing over 18,000 desensitized trajectory points. Experimental results show that this embodiment significantly enhances the ability to protect trajectories from disturbances in sensitive areas (such as residential entrances and exits, and areas surrounding medical institutions) while ensuring trajectory continuity and availability.

[0087] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0088] Example 2

[0089] This embodiment also provides a device for desensitizing vehicle trajectory data. This device is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0090] Figure 5 This is a structural block diagram of a vehicle trajectory data desensitization device according to an embodiment of the present invention, such as... Figure 5 As shown, the device includes:

[0091] The acquisition module 51 is used to acquire the driving trajectory sequence collected by the vehicle's positioning module;

[0092] The identification module 52 is used to identify the privacy sensitivity of the driving trajectory sequence;

[0093] Construction module 53 is used to construct a feature vector sequence using the driving trajectory sequence and the privacy sensitivity;

[0094] The generation module 54 is used to generate a disturbance offset sequence in the state space based on the feature vector sequence, wherein the sequence length of the disturbance offset sequence is the same as the sequence length of the driving trajectory sequence;

[0095] The desensitization module 55 is used to desensitize the driving trajectory sequence using the disturbance offset sequence to obtain the desensitized target trajectory sequence, and replace the driving trajectory sequence with the target trajectory sequence.

[0096] Optionally, the acquisition module includes: an acquisition unit, configured to acquire a set of trajectory point data collected by the vehicle's positioning module at each time point according to a period, wherein each trajectory point data in the trajectory point data set includes the coordinate data of the trajectory point, the vehicle speed, and a timestamp; and a sorting unit, configured to sort the trajectory point data set according to the timestamp to obtain a driving trajectory sequence.

[0097] Optionally, the identification module includes: an acquisition unit, configured to acquire, for each trajectory point in the driving trajectory sequence, the location sensitivity of the location to which the trajectory point belongs, the region sensitivity of the region where the trajectory point is located, and the state sensitivity of the vehicle state of the trajectory point; and a calculation unit, configured to calculate the privacy sensitivity of the trajectory point by weighting the location sensitivity, the region sensitivity, and the state sensitivity.

[0098] Optionally, the acquisition unit includes: a positioning subunit, configured to locate the geographical location of the trajectory point in map data; and a determination subunit, configured to determine the location sensitivity of the trajectory point to a first privacy level if the geographical location is an external road; determine the location sensitivity of the trajectory point to a second privacy level if the geographical location is an internal road; and determine the location sensitivity of the trajectory point to a third privacy level if the geographical location is a building, wherein the location sensitivity of the first privacy level is less than the location sensitivity of the second privacy level, and the location sensitivity of the second privacy level is less than the location sensitivity of the third privacy level.

[0099] Optionally, the acquisition unit includes: a search subunit, used to search for the number of points of interest within a fixed radius area in the map data, with the trajectory point as the center; and a calculation subunit, used to calculate the point of interest density based on the number of points of interest and the area, wherein the point of interest density is positively correlated with the area sensitivity.

[0100] Optionally, the acquisition unit includes: an acquisition subunit for acquiring the vehicle speed of the trajectory point; a judgment subunit for judging whether the vehicle speed is less than a preset speed; and an allocation subunit for determining that if the vehicle speed is less than the preset speed and remains so for a preset duration, the vehicle state of the trajectory point is a stationary state, and a first state sensitivity is allocated to the trajectory point; if the vehicle speed is greater than or equal to the preset speed, the vehicle state of the trajectory point is a moving state, and a second state sensitivity is allocated to the trajectory point, wherein the first state sensitivity is greater than the second state sensitivity.

[0101] Optionally, the construction module includes: a construction unit, configured to construct the following trajectory feature vector for each time step in the driving trajectory sequence. : ,in, These represent the horizontal axis data, vertical axis data, and vehicle speed at the time step, respectively. These are respectively the position sensitivity, region sensitivity, state sensitivity, and privacy sensitivity of the time step; the sorting unit is used to sort the trajectory feature vectors of all time steps according to time to obtain a feature vector sequence.

[0102] Optionally, the generation module includes: a calculation unit, used to calculate the perturbation offset at time step t in the perturbation offset sequence in the Mamba model using the following formula. : ;in, It is a sequence of feature vectors. , Let be the state vectors of the Mamba model at time step t and time step t-1, respectively. Let be the state transition matrix of the Mamba model, representing the state transition relationship of the state vector in the time dimension. Let be the control matrix of the Mamba model, representing the influence of the feature vector sequence on the state vector update. The observation matrix of the Mamba model describes the linear mapping relationship between the state vector and the perturbation offset.

[0103] Optionally, the generation module further includes: a construction unit, configured to calculate the perturbation offset at time step t in the perturbation offset sequence in the Mamba model using the following formula in the calculation unit. Previously, sample data was constructed using original trajectory samples and reference perturbation labels; training units were used to employ the following loss function. The model parameters for training the Mamba model are: ;in The reference perturbation amount of the reference perturbation tag at time step t. , These are the feature vectors of the original trajectory sample at time step t and time step t-1, respectively. The first balance parameter is used to constrain the disturbance intensity error, and the second balance parameter is used to constrain the disturbance smoothness. The model parameters include the state transition matrix, the control matrix, and the observation matrix.

[0104] Optionally, the desensitization module includes: a desensitization unit, used to desensitize the driving trajectory sequence using the following formula to obtain the desensitized target trajectory sequence. : ;in, The coordinate data of the trajectory point at time t in the driving trajectory sequence. It is the offset at time t in the perturbation offset sequence.

[0105] It should be noted that the above modules can be implemented by software or hardware. For the latter, they can be implemented in the following ways, but are not limited to: all the above modules are located in the same processor; or, the above modules are located in different processors in any combination.

[0106] Example 3

[0107] Embodiments of the present invention also provide a storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above method embodiments when running.

[0108] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0109] S1, Obtain the driving trajectory sequence collected by the vehicle's positioning module;

[0110] S2, Identify the privacy sensitivity of the driving trajectory sequence;

[0111] S3, construct a feature vector sequence using the driving trajectory sequence and the privacy sensitivity;

[0112] S4, generate a disturbance offset sequence in the state space based on the feature vector sequence, wherein the sequence length of the disturbance offset sequence is the same as the sequence length of the driving trajectory sequence;

[0113] S5, the driving trajectory sequence is desensitized using the disturbance offset sequence to obtain the desensitized target trajectory sequence, and the driving trajectory sequence is replaced with the target trajectory sequence.

[0114] Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0115] Embodiments of the present invention also provide an electronic device including a memory and a processor, the memory storing a computer program and the processor being configured to run the computer program to perform the steps in any of the above method embodiments.

[0116] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.

[0117] Optionally, in this embodiment, the processor can be configured to perform the following steps via a computer program:

[0118] S1, Obtain the driving trajectory sequence collected by the vehicle's positioning module;

[0119] S2, Identify the privacy sensitivity of the driving trajectory sequence;

[0120] S3, construct a feature vector sequence using the driving trajectory sequence and the privacy sensitivity;

[0121] S4, generate a disturbance offset sequence in the state space based on the feature vector sequence, wherein the sequence length of the disturbance offset sequence is the same as the sequence length of the driving trajectory sequence;

[0122] S5, the driving trajectory sequence is desensitized using the disturbance offset sequence to obtain the desensitized target trajectory sequence, and the driving trajectory sequence is replaced with the target trajectory sequence.

[0123] Optionally, specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementations, and will not be repeated here.

[0124] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0125] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0126] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also include the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0127] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for desensitizing vehicle trajectory data, characterized in that, include: Obtain the driving trajectory sequence collected by the vehicle's positioning module; Identify the privacy sensitivity of the driving trajectory sequence; A feature vector sequence is constructed using the driving trajectory sequence and the privacy sensitivity. A disturbance offset sequence is generated in the state space based on the feature vector sequence, wherein the sequence length of the disturbance offset sequence is the same as the sequence length of the driving trajectory sequence; The driving trajectory sequence is desensitized using the disturbance offset sequence to obtain the desensitized target trajectory sequence, and the driving trajectory sequence is replaced with the target trajectory sequence. The generation of the perturbation offset sequence in the state space based on the feature vector sequence includes: calculating the perturbation offset at time step t in the perturbation offset sequence in the Mamba model using the following formula. : ;in, It is a sequence of feature vectors. , Let be the state vectors of the Mamba model at time step t and time step t-1, respectively. Let be the state transition matrix of the Mamba model, representing the state transition relationship of the state vector in the time dimension. Let be the control matrix of the Mamba model, representing the influence of the feature vector sequence on the state vector update. The observation matrix of the Mamba model describes the linear mapping relationship between the state vector and the perturbation offset.

2. The method according to claim 1, characterized in that, The vehicle's trajectory sequence acquired by the positioning module includes: The system acquires a set of trajectory point data collected by the vehicle's positioning module at each time point according to a periodic schedule. Each trajectory point data in the trajectory point data set includes the coordinate data of the trajectory point, the vehicle speed, and a timestamp. The trajectory point data set is sorted according to the timestamp to obtain the driving trajectory sequence.

3. The method according to claim 1, characterized in that, The privacy sensitivity of identifying the driving trajectory sequence includes: For each trajectory point in the driving trajectory sequence, obtain the position sensitivity of the location to which the trajectory point belongs, obtain the region sensitivity of the area where the trajectory point is located, and obtain the state sensitivity of the vehicle state of the trajectory point. The privacy sensitivity of the trajectory point is calculated by weighting the location sensitivity, the region sensitivity, and the state sensitivity.

4. The method according to claim 3, characterized in that, Obtaining the position sensitivity of the location to which the trajectory point belongs includes: Locate the geographical location of the trajectory point in the map data; If the geographical location is an external road, the location sensitivity of the trajectory point is determined to be a first privacy level; if the geographical location is an internal road, the location sensitivity of the trajectory point is determined to be a second privacy level; if the geographical location is a building, the location sensitivity of the trajectory point is determined to be a third privacy level, wherein the location sensitivity of the first privacy level is less than the location sensitivity of the second privacy level, and the location sensitivity of the second privacy level is less than the location sensitivity of the third privacy level.

5. The method according to claim 3, characterized in that, Obtaining the region sensitivity of the area where the trajectory point is located includes: Using the trajectory points as the center, find the number of points of interest within a fixed radius area in the map data; The interest point density is calculated based on the number of interest points and the area of ​​the region, wherein the interest point density is positively correlated with the region sensitivity.

6. The method according to claim 3, characterized in that, The state sensitivity for obtaining the vehicle state of the trajectory points includes: Obtain the vehicle speed at the trajectory points; Determine whether the vehicle speed is less than a preset speed; If the vehicle speed is less than a preset speed and remains so for a preset duration, the vehicle state of the trajectory point is determined to be stationary, and a first state sensitivity is assigned to the trajectory point; if the vehicle speed is greater than or equal to the preset speed, the vehicle state of the trajectory point is determined to be moving, and a second state sensitivity is assigned to the trajectory point, wherein the first state sensitivity is greater than the second state sensitivity.

7. The method according to claim 1, characterized in that, Constructing a feature vector sequence using the driving trajectory sequence and the privacy sensitivity includes: For each time step in the driving trajectory sequence, the following trajectory feature vector is constructed. : ,in, These represent the horizontal axis data, vertical axis data, and vehicle speed at the time step, respectively. These are the location sensitivity, region sensitivity, state sensitivity, and privacy sensitivity of the time step, respectively. The trajectory feature vectors of all time steps are sorted according to time to obtain the feature vector sequence.

8. The method according to claim 1, characterized in that, The perturbation offset at time step t in the perturbation offset sequence is calculated in the Mamba model using the following formula. Previously, the method also included: Sample data is constructed using original trajectory samples and reference perturbation labels; The following loss function is used. The model parameters for training the Mamba model are: ； in, The reference perturbation amount of the reference perturbation tag at time step t. , These are the feature vectors of the original trajectory sample at time step t and time step t-1, respectively. The first balance parameter is used to constrain the disturbance intensity error, and the second balance parameter is used to constrain the disturbance smoothness. The model parameters include the state transition matrix, the control matrix, and the observation matrix.

9. The method according to claim 1, characterized in that, Desensitizing the driving trajectory sequence using the disturbance offset sequence includes: The driving trajectory sequence is desensitized using the following formula to obtain the desensitized target trajectory sequence. : ； in, The coordinate data of the trajectory point at time t in the driving trajectory sequence. It is the offset at time t in the perturbation offset sequence.

10. A device for desensitizing vehicle trajectory data, characterized in that, include: The acquisition module is used to acquire the driving trajectory sequence collected by the vehicle's positioning module; An identification module is used to identify the privacy sensitivity of the driving trajectory sequence; A construction module is used to construct a feature vector sequence using the driving trajectory sequence and the privacy sensitivity. A generation module is used to generate a disturbance offset sequence in the state space based on the feature vector sequence, wherein the sequence length of the disturbance offset sequence is the same as the sequence length of the driving trajectory sequence; The desensitization module is used to desensitize the driving trajectory sequence using the disturbance offset sequence to obtain the desensitized target trajectory sequence, and to replace the driving trajectory sequence with the target trajectory sequence. The generation module includes a calculation unit, used to calculate the perturbation offset at time step t in the perturbation offset sequence in the Mamba model using the following formula. : ;in, It is a sequence of feature vectors. , Let be the state vectors of the Mamba model at time step t and time step t-1, respectively. Let be the state transition matrix of the Mamba model, representing the state transition relationship of the state vector in the time dimension. Let be the control matrix of the Mamba model, representing the influence of the feature vector sequence on the state vector update. The observation matrix of the Mamba model describes the linear mapping relationship between the state vector and the perturbation offset.

11. The apparatus according to claim 10, characterized in that, The acquisition module includes: The acquisition unit is used to acquire the trajectory point data set collected by the vehicle's positioning module at each time point according to a period, wherein each trajectory point data in the trajectory point data set includes the coordinate data of the trajectory point, the vehicle speed, and the timestamp; the sorting unit is used to sort the trajectory point data set according to the timestamp to obtain the driving trajectory sequence.

12. The apparatus according to claim 10, characterized in that, The identification module includes: The acquisition unit is configured to acquire, for each trajectory point in the driving trajectory sequence, the location sensitivity of the location to which the trajectory point belongs, the region sensitivity of the area where the trajectory point is located, and the state sensitivity of the vehicle state of the trajectory point; the calculation unit is configured to calculate the privacy sensitivity of the trajectory point by weighting the location sensitivity, the region sensitivity, and the state sensitivity.

13. A storage medium, characterized in that, The storage medium stores a computer program, wherein the computer program is configured to execute the method described in any one of claims 1 to 9 when it is run.

14. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to perform the method as described in any one of claims 1 to 9.

Images