Privacy trajectory fuzzy matching method and device for unmanned aerial vehicle collision early warning

Abstract

The invention discloses a privacy track fuzzy matching method for collision early warning of an unmanned aerial vehicle. The method comprises the following steps that S1, an initiator and a responder set and share public parameters related to collision early warning; s2, the initiator and the responder respectively execute a super-dimensional cube division algorithm based on the common parameters, and track point coordinate information of the initiator and the responder is converted into a tetrad identifier based on a division result; s3, the initiator carries out OKVS coding on the three-dimensional positions of the own track points and the neighborhood points and the corresponding timestamps to obtain a message of an OKVS structure and sends the message to the responder; step S4, the responder performs linear decoding on the request message by using a Gaussian elimination method to generate a DDH re-randomized tuple, and further constructs a mask message and feeds back the mask message to the initiator; and S5, the initiator carries out decryption and comparison based on the feedback message to obtain a collision early warning judgment result. The method has the advantages of being high in early warning real-time performance, high in precision and high in safety.

Description

Technical Field

[0001] This invention relates to the field of drone collision warning technology, and in particular to a privacy trajectory fuzzy matching method and device for surface drone collision warning. Background Technology

[0002] In recent years, the rapid development of drones has gradually made them an important engine for promoting the low-altitude economy. As an emerging cutting-edge technology, drones, with their advantages of flexibility, convenient deployment, and low cost, have been widely used in key areas such as parcel delivery, traffic management, environmental monitoring, agricultural spraying, and emergency rescue, becoming a core driving force for the low-altitude economy.

[0003] However, while the large-scale application of drones brings convenience and benefits, it also raises new security challenges. With the widespread deployment of drone systems in densely populated urban areas, collaborative operations in complex urban airspace and dynamic environments are becoming increasingly frequent, leading to a surge in the risk of trajectory intersections and collisions, seriously threatening flight equipment, cargo, and public safety. To effectively detect nearby targets and provide early warnings in low-altitude networks, most drones currently need to share their real-time location coordinates during flight, exchanging a large amount of high-precision trajectory data during mission execution. However, sharing real-time location data can be maliciously eavesdropped on and exploited, leading to equipment hijacking and causing significant economic and security risks. Furthermore, the transmission and sharing of such data carries privacy risks: on the one hand, real-time location information is easily eavesdropped on and tampered with, potentially triggering security incidents such as equipment hijacking and causing significant economic losses; on the other hand, the shared trajectory data includes flight position, altitude, timestamps, and takeoff and landing locations, often involving sensitive facilities such as users and warehousing centers, potentially exposing logistics routes, customer identities, mission intentions, and even business strategies. Compared to traditional mobile terminal data, drone trajectory data has characteristics such as high spatiotemporal resolution, large spatial coverage, and strong inferability. Even after anonymization, it can still be reconstructed through trajectory re-identification and pattern analysis, thereby revealing the real identity or sensitive locations. This means that traditional privacy protection technologies are insufficient to meet the practical application needs in low-altitude collision avoidance scenarios for drones.

[0004] The core idea of ​​collision warning mechanisms is to find potential collision points through trajectory matching. Early research focused on optimizing the efficiency and accuracy of trajectory matching and collision avoidance, with insufficient attention paid to protecting location privacy. In recent years, some studies have begun to focus on location privacy protection in UAV trajectory matching. The key to current research is how to improve efficiency and reduce energy consumption while using fuzzy matching of privacy-preserving trajectories. Existing research has significant shortcomings: on the one hand, while warning schemes relying on airborne sensors do not depend on communication, their effective range is limited, and their performance is prone to degradation due to recognition delays and computational loads in high-speed, high-density environments, posing security risks. On the other hand, while cryptographic schemes for protecting location privacy can theoretically prevent information leakage, they generally incur high computational and communication overhead, making it difficult to meet real-time requirements. On the other hand, using lightweight encryption to improve speed sacrifices security, leading to a high false alarm rate.

[0005] The limitations of existing solutions are mainly reflected in the following aspects: First, traditional collision warning schemes do not take privacy protection into account, while airborne sensing or third-party dependent privacy schemes are not practical enough. Traditional schemes achieve collision warning by sharing trajectory data in real time, which poses risks of sensitive information leakage and device hijacking; airborne sensing autonomous avoidance schemes that do not share trajectory data are limited by detection range, recognition latency, and computational load, resulting in poor collision avoidance performance in high-density airspace; privacy schemes that rely on trusted third parties are unavailable in remote areas, and airborne network connections are prone to introducing latency, affecting the real-time performance of warnings. Second, privacy-preserving cryptographic schemes and lightweight distributed schemes cannot simultaneously meet the requirements of low overhead, high security, and high-precision monitoring. Schemes based on secure multi-party computation and fully homomorphic encryption technologies have high computational and communication overhead, making them difficult to adapt to resource-constrained UAVs; lightweight encryption or simplified spatiotemporal coordinate dimensions, while reducing overhead, can lead to privacy leaks, false alarms and missed alarms in collision warnings, or face the risk of dictionary attacks, failing to meet the actual needs of UAVs for real-time and accurate collision avoidance.

[0006] In summary, existing methods struggle to balance efficiency, accuracy, and security. There is an urgent need to propose a privacy-preserving trajectory matching protocol that can efficiently and accurately calculate the spatiotemporal intersection points of UAV trajectories in order to address this critical challenge. Summary of the Invention

[0007] In view of this, it is necessary to provide a privacy trajectory fuzzy matching method and device for surface-based UAV collision warning, so as to effectively solve the technical problem that it is difficult to balance real-time performance, accuracy and security when matching trajectories.

[0008] This invention provides a privacy trajectory fuzzy matching method for surface-based UAV collision warning, comprising the following steps: Step S1: The initiator and the responder set common parameters related to the collision warning, establish communication between the two parties, and share the common parameters; Step S2: Based on the common parameters, the initiator and the responder respectively execute the hypercube partitioning algorithm to discretize the continuous three-dimensional geographic space into multiple hypercubes; based on the partitioning results, the coordinate information of the trajectory points of both parties is converted into indexed quadruple identifiers. Step S3: The initiator performs OKVS encoding on the three-dimensional position and corresponding timestamp of its own trajectory point and its neighboring points to obtain a message with an OKVS structure, and sends it as a request message to the responder. Step S4: The responder uses Gaussian elimination to linearly decode the request message to generate a DDH rerandomized tuple, constructs a mask message based on the DDH rerandomized tuple, and sends it back to the initiator. Step S5: The initiator decrypts and compares the feedback message to obtain the collision warning judgment result.

[0009] Preferably, step S1 specifically includes: Setting the threshold radius: The initiator sets the calculation formula for the threshold radius based on environmental parameters, calculates the real-time radius value, and broadcasts it to the responder; Establish a unified coordinate system: The initiator selects a coordinate origin to construct a unified coordinate system, adds a random perturbation vector to perturb the origin, and synchronizes the offset and updated origin information to the responder; Setting common parameters for the cyclic group: The common parameters for the cyclic group are established through negotiation between both parties; Establish communication between the two parties and share the threshold radius, the origin information, and the common parameters of the cyclic group.

[0010] Preferably, step S2 specifically comprises: The three-dimensional space is divided into cubes with side lengths twice the threshold radius according to the grid. Map the trajectory points of both parties onto the grid, construct a position sphere with the trajectory point as the center and the radius as the threshold radius, merge multiple cubes that intersect with the position sphere into a block, and obtain the index of the block as the unique identifier of the corresponding trajectory point; The quadruple identifier is obtained by combining the index of the block and the corresponding timestamp index.

[0011] Preferably, step S3 specifically comprises: Step S31: The initiator queries the responder's trajectory data and initializes the cyclic group parameters to generate a private key and a public key; Step S32: The initiator obtains all neighboring points within a radius of a threshold radius centered on its own trajectory point, and performs hash calculation based on the coordinate information of the trajectory point and its neighboring points and the corresponding block code to generate a list of key-value pairs; Step S33: The initiator maps the key-value pairs in the key-value pair list to a set of four-dimensional group elements in batches; Step S34: Perform OKVS encoding on the four-dimensional group element set to obtain an OKVS structure message; package the public key information and the OKVS structure message and send them to the responder.

[0012] Preferably, step S32 specifically includes: For its own trajectory point, the initiator determines the corresponding block number, then enumerates all neighboring points within a sphere centered on the trajectory point and with a threshold radius as the radius, and performs hash calculation to generate a key based on the block number and the coordinate information of its neighboring points; the initiator randomly samples from the cyclic group, generates a corresponding random value for each key, and constructs a list of key-value pairs to be encoded.

[0013] Preferably, step S33 specifically includes: The initiator first expands the storage space according to the expansion rate, and then maps the key-value pairs in batches to the four-dimensional group element set.

[0014] Preferably, step S4 specifically comprises: The responder parses the request message and extracts the public parameters, public key, and OKVS structure. The responder obtains a set of spatial block labels with a side length of four times the threshold radius for its own trajectory points, as well as the timestamp information of the cube containing the corresponding trajectory points; Each spatial block label is combined with its corresponding timestamp to generate a query key. Gaussian elimination is used to perform linear decoding on the messages of the OKVS structure to obtain the group elements. The responder selects a random number as a blinding factor, performs a modular exponentiation operation on the group elements and the initiator's public key to generate a rerandomized ciphertext tuple, and aggregates the ciphertext tuple to generate the DDH rerandomized tuple; The responder calculates the hash value of its own block index and performs an XOR mask operation with the DDH rerandomized tuple to obtain the mask message, and then feeds the mask message back to the initiator.

[0015] Preferably, step S5 specifically comprises: The initiator parses the masked message, extracts the DDH rerandomized tuple and the masked ciphertext, and verifies it using its own private key. It performs exponentiation and XOR operations on the group element part of the DDH rerandomized tuple, and verifies whether the two parties satisfy the DDH tuple relationship at the current time and space based on the operation results. If they do, it is determined that there is a collision risk; otherwise, it is determined that there is no collision risk.

[0016] The present invention also provides a privacy trajectory fuzzy matching device for surface-mounted UAV collision warning, including a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, it implements the privacy trajectory fuzzy matching method for surface-mounted UAV collision warning.

[0017] Compared with existing technologies, the advantages of this invention are as follows: First, this invention employs spatial hashing technology to discretize the continuous three-dimensional geographic space, mapping the UAV trajectory into a quadruple containing a spatial unit index and a timestamp. This process transforms the complex trajectory matching problem into the intersection calculation of a discrete point set and an extended sphere set, avoiding direct interaction with precise GPS coordinates. Second, this invention utilizes the OKVS data structure to construct query requests, batch-encoding all trajectory neighborhood points of the initiator into a compressed encrypted data structure. Specifically, this invention ensures the solvability of linear equations by expanding storage space and achieves fast decoding by combining Gaussian elimination. This mechanism transforms the originally cumbersome point-to-point encrypted comparison into efficient batch computation, significantly reducing communication rounds and local computing load, thereby meeting the requirements of low latency and real-time performance for UAV obstacle avoidance. Finally, a re-randomization technique based on the DDH assumption is introduced in the data interaction stage. The responder, without decrypting the initiator's data, uses random numbers to blind the ciphertext and generates a feedback message. This process ensures the computational indistinguishability of the data, allowing the initiator to identify only non-empty intersections and preventing the acquisition of any location information in non-collision regions, thus achieving bidirectional privacy protection under a semi-honest model. This invention overcomes the limitation of traditional privacy set intersection protocols, which can only achieve precise intersections. It enables fuzzy detection of intersection points in 3D trajectory data without exposing real, precise coordinate data, thereby revealing potential trajectory collision risks. It effectively addresses the privacy leakage defects of existing traditional trajectory sharing schemes, avoiding the leakage of sensitive information caused by directly sharing precise coordinates. Even if an attacker eavesdrops on the communication channel and attempts to deduce coordinate information from intermediate data, this invention still ensures that no trajectory privacy information is leaked. Attached Figure Description

[0018] 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: Figure 1 This is a flowchart of an embodiment of the privacy trajectory fuzzy matching method for surface-based UAV collision warning provided by the present invention. Detailed Implementation

[0019] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0020] Example 1 This embodiment aims to solve the core challenge of balancing privacy protection, detection accuracy, and real-time efficiency in existing drone trajectory matching and collision avoidance technologies, ensuring the safety and privacy of drones in low-altitude collaborative operations, and meeting the needs of large-scale commercial applications of drones in low-altitude economic scenarios.

[0021] Please see Figure 1 The privacy trajectory fuzzy matching method for surface-based UAV collision warning in this embodiment specifically includes the following steps: Step S1: The initiator and the responder set common parameters related to the collision warning, establish communication between the two parties, and share the common parameters; Step S2: Based on the common parameters, the initiator and the responder respectively execute the hypercube partitioning algorithm to discretize the continuous three-dimensional geographic space into multiple hypercubes; based on the partitioning results, the coordinate information of the trajectory points of both parties is converted into indexed quadruple identifiers. Step S3: The initiator performs OKVS encoding on the three-dimensional position and corresponding timestamp of its own trajectory point and its neighboring points to obtain a message with an OKVS structure, and sends it as a request message to the responder. Step S4: The responder uses Gaussian elimination to linearly decode the request message to generate a DDH rerandomized tuple, constructs a mask message based on the DDH rerandomized tuple, and sends it back to the initiator. Step S5: The initiator decrypts and compares the feedback message to obtain the collision warning judgment result.

[0022] This embodiment specifically detects potential collision risks while protecting the privacy of the trajectories of two drones. Since directly exchanging precise coordinates would expose sensitive flight paths, this embodiment utilizes hyperdimensional cube partitioning and spatial hashing techniques to convert the drone's spatiotemporal trajectory points into fuzzy quadruple identifiers. During data exchange, this embodiment introduces an OKVS batch encoding structure to batch process and encode the trajectory quadruples to reduce computational overhead, and combines this with a Diffie-Hellman decision mechanism (DDH re-randomization) to blind the encrypted data. This method effectively meets the performance requirements of drones for real-time collision detection while ensuring data privacy and security under a semi-honest model.

[0023] This method integrates a fuzzy privacy set intersection protocol with spatial hashing technology, overcoming the limitation of traditional privacy set intersection protocols that can only achieve precise intersection. It enables fuzzy detection of intersection points in 3D trajectory data without exposing real, precise coordinate data, thereby revealing potential trajectory collision risks. It effectively addresses the privacy leakage defects of existing traditional trajectory sharing schemes, avoiding the leakage of sensitive information caused by directly sharing precise coordinates. Even if an attacker eavesdrops on the communication channel and attempts to deduce coordinate information from intermediate data, this invention can still ensure that no trajectory privacy information is leaked.

[0024] This embodiment specifically includes four modules: parameter setting and initialization, spatial partitioning and trajectory mapping, OKVS data structure construction, and DDH rerandomization and verification.

[0025] First module: Parameter setting and initialization. (Initiator) First, a radius threshold is set by comprehensively considering factors such as flight safety distance and local GPS error. And broadcast, this threshold determines the granularity of subsequent spatial division and the scope of collision warnings. (Initiator) Randomly select an origin point Furthermore, a random perturbation vector is generated to offset and update the origin, preventing attackers from reverse-engineering the coordinates. Based on this, the participating parties... and Negotiate to establish common parameters for the cyclic group Establish communication between the two parties and share parameters such as radius threshold and origin.

[0026] The second module: Spatial segmentation and trajectory mapping. During the offline or local preprocessing stage, the initiator... and responders Based on the origin determined through negotiation and distance threshold The hyperdimensional cube partitioning algorithm is executed separately. This algorithm discretizes the continuous three-dimensional geographic space into several cubes with side lengths of [missing information]. The relevant hypercube mesh. For each spatiotemporal data point in their respective trajectories, both parties convert it into a four-tuple identifier, which consists of the cell index corresponding to the 3D geographic coordinates and the corresponding timestamp index. Through this module, both parties generate unique identifiers for their respective trajectory points, transforming physical location matching into mesh index matching.

[0027] The third module: OKVS data structure construction. (Initiator) In order to query without revealing one's own trajectory information trajectory data, First, initialize the cyclic group parameters and generate the private key. With public key Specifically, Not only for the spatiotemporal quadruple on its own trajectory, but also for each trajectory point as the center and radius For all neighboring points within the range, generate a list of key-value pairs by hashing the coordinates of these points with their corresponding block numbers. This is to ensure the high probability solvability of the OKVS linear equation system. Based on expansion rate The storage space is expanded, and then the key-value pairs mentioned above are mapped in batches to a set of four-dimensional group elements. After the build is complete, The message will contain the public key parameter and the OKVS structure. Send as a request to the participants .

[0028] The fourth module: DDH rerandomization and validation. Participants Receive request message Then, for each spatiotemporal point in its own trajectory, determine the location containing that point. The side length is The neighboring block labels are used, and the keys generated by the block number and timestamp are used to perform linear decoding of OKVS using Gaussian elimination to generate rerandomized tuples. Then calculate the hash value of the block tag and match it with... Perform an XOR operation to construct a mask message to feed back. . After receiving the message, use the private key right Perform exponentiation and hashing, and XOR the result with the received ciphertext for verification. If the verification result is equal to the hash value of its own block label, it indicates that the two parties satisfy the DDH tuple relationship at this time and space point, that is, it is determined that there is a potential collision point and an alarm is triggered; otherwise, the current flight is determined to be safe.

[0029] Each step will be explained in detail below.

[0030] This embodiment describes an urban low-altitude logistics delivery network, involving two logistics drones performing a mission, defined as the initiating drone. and responder drones All drones are equipped with GNSS modules for self-localization. To mitigate positioning errors, a certain threshold distance is set for fuzzy proximity detection. They also feature wireless communication modules, employing WiFi-Direct technology to establish secure point-to-point links without the need for trusted third parties, and using the TLS protocol to ensure data transmission security. Given the limited computing resources of the onboard processor and the lack of online public key infrastructure, the drones are pre-installed with public-private key pairs and certificates for offline authentication. During mission execution, The system preprocesses a set of four-dimensional tuple trajectories composed of three-dimensional spatial coordinates and timestamps, and periodically broadcasts encrypted trajectory information. Upon receiving the information, the system responds to the protocol and performs real-time collision detection under privacy protection on a resource-constrained local onboard device.

[0031] The general process of this method is as follows: Figure 1 As shown, it specifically includes the following five steps: Step S1: Protocol Parameter Settings. This step mainly completes the calculation of the threshold radius and the establishment of a unified coordinate system, exchanging common parameters such as the threshold radius, coordinate origin, and elliptic curve parameters, providing the basic settings for the subsequent fuzzy matching protocol.

[0032] Step S11: Calculate the radius threshold Initiator Based on the current environment, key parameters are set to determine the threshold radius. Settings, such as setting GNSS positioning error Meters, the minimum safe distance between trajectory points meters, the maximum flight speed of the drone meters per second, estimated execution time of the protocol Second. Calculate the radius threshold according to the formula : After substituting in specific values, set Meters, this threshold and others subsequently broadcast to .

[0033] Step S12: Origin selection and perturbation update. As the initiator of the agreement, choose ,For example Establish a unified coordinate reference system.

[0034] Step S13 An origin perturbation mechanism was introduced during the initialization phase, namely, adding a perturbation vector. ,For example And update the origin: Step S14: Synchronize the updated origin information to This ensures that both parties perform fuzzy matching operations within the same reference frame.

[0035] Step S1, by introducing a random origin offset and a dynamically calculated radius threshold, effectively prevents attackers from reverse-engineering the actual flight path through the grid while achieving fuzzy matching.

[0036] Step S2: Hyperdimensional cube partitioning and trajectory block mapping. Based on the exchanged common parameters, trajectory points are block-mapped into quadruple indices, providing the basic setup for subsequent fuzzy matching protocols.

[0037] The three-dimensional space is divided into grids with side lengths of... The cubes, each considered a unit. The drone at the timestamp... Location It will be mapped to the cell it belongs to, a cell with a radius of of Most balls and The units intersect. The system will... The units are merged into a side length of The hyperdimensional cube block is used as the unique identifier for the trajectory point: Assuming a drone Coordinates are Based on the relative coordinates of the new origin: The block index obtained after partitioning is .

[0038] Step S3: Generate the OKVS data structure. In this step, the initiator performs a local computation according to a preset protocol. The three-dimensional location points of its own trajectory and the corresponding timestamps are encoded into OKVS to achieve efficient encrypted storage and transmission of batch data.

[0039] Step S31 Choose a prime-order cyclic group and its generators And generate a private key by randomly sampling locally. The corresponding public key is calculated based on the private key. .

[0040] Step S32: For each spatiotemporal point in the planned trajectory, the initiator first determines its corresponding block number, and then enumerates all neighboring points within a preset radius centered on that point. The key is obtained by hashing the block number and the tuples in its neighborhood using a hash function to calculate the cyclic group element. ,in .

[0041] Enumeration refers to neighborhood points. In order to eliminate positioning errors and ensure the effectiveness of fuzzy matching, neighboring points in geospatial space are uniformly included in the coding range, ensuring that matching can be triggered as long as the responder is within this range.

[0042] The initiator randomly samples from the cyclic group and generates a corresponding random value for each generated "key", thereby constructing a list of key-value pairs to be encoded. From the group Random sampling and generate the corresponding value. .

[0043] Step S33, Initiator Using the OKVS encoding algorithm, the key-value pair list generated in step S23 is batch encoded into an efficient query data structure, that is, key-value pairs... Encoded as OKVS structure .

[0044] Step S34 Package the public key information generated in step S21 and the OKVS data structure generated in step S24 into a message. Send to .

[0045] Step S4: DDH rerandomization. This step is performed by the responder. In encrypted mode, local trajectory data is used in the calculation to generate intermediate query results containing potential intersection information.

[0046] Step S41, upon receiving from After receiving the detection request, message tuples The data is parsed to extract the public parameters, the initiator's public key, and the OKVS data structure.

[0047] Step S42 For each point in its own trajectory , Get a set of side lengths The spatial block label, and the timestamp information corresponding to the cell containing that point.

[0048] The aforementioned set of spatial block labels is designed to cover the possible range of positioning errors, ensuring that if the physical locations of both parties are close, there will be at least one identical block label.

[0049] Step S43: For each spatial block label, Combine it with the timestamp to generate a query key: The OKVS structure extracted in step S31 is then linearly decoded using Gaussian elimination.

[0050] The linear decoding algorithm treats OKVS as a system of linear equations and solves it using locally generated keys as coefficient vectors. If both sides have the same block labels (i.e., there is an intersection of trajectories), this decoding step will reconstruct the group elements with specific algebraic relationships.

[0051] Step S44: The responder selects a random number as the blinding factor and performs a modular exponentiation operation on the group elements decoded in step S33 and the initiator's public key to generate a re-randomized ciphertext tuple. Specifically, Rerandomized tuples Aggregate and generate the final tuple. ,in , .

[0052] Step S45: To obfuscate the results, Calculate the hash value of its own block index and perform an XOR mask operation. . Will include Feedback to .

[0053] Step S5: DDH message verification and collision detection. This step is initiated by the party initiating the process. The process involves using the local private key to decrypt and compare the ciphertext, ultimately determining the collision risk.

[0054] Step S51 Analysis from News The rerandomized tuples and masked ciphertext are extracted. Then, the initiator uses its own private key. To verify the message, the group element portion of the tuple is exponentially and hashed to reconstruct the intermediate calculated value used for verification. .

[0055] Step S52 The initiator performs an XOR operation between the intermediate value calculated in step S41 and the mask portion in the feedback message to verify the equation. The system checks whether the condition is met. If it is met, a collision risk is identified, an alarm is triggered, and collision avoidance is performed. If it is not met, no collision risk is identified, and the flight mission continues.

[0056] This method combines the batch unintentional key-value store (OKVS) structure with the decision Diffie-Hellman (DDH) re-randomization encryption technology. It aims to resolve the contradiction between high security, high detection accuracy and low computational overhead and high operating efficiency in existing UAV privacy protection trajectory matching schemes. It overcomes the shortcomings of traditional privacy protection schemes, which are unable to adapt to the resource-constrained characteristics of UAVs due to excessive computational and communication overhead, or sacrifice detection accuracy and security in pursuit of efficiency.

[0057] This method achieves high security and privacy protection for sensitive trajectory data. It applies fuzzy privacy set intersection technology to ensure that during the collision detection interaction process, the initiator can only identify the non-empty intersection of its own index and the spatiotemporal point of the responder, providing provable security under the semi-honest attack model.

[0058] This method improves the robustness and accuracy of UAV collision detection, overcomes positioning error interference to a certain extent, and is more practical. By batch encoding all points within the threshold neighborhood, it not only avoids the risk of missed detections at boundary points after block division but also effectively tolerates positioning noise from UAVs in real-world environments. This invention can achieve high-accuracy prediction of potential collision risks while ensuring a safe distance.

[0059] This method significantly reduces the computational and energy costs in privacy-preserving computations, making it suitable for resource-constrained drone scenarios. By introducing a batch encoding structure, all four tuples of the corresponding trajectory are batch encoded and combined with re-randomization. The receiver can complete the comparison in a single batch call during decoding, reducing local computing power consumption and improving the execution efficiency of the protocol.

[0060] Example 2 This embodiment provides a privacy trajectory fuzzy matching device for surface-based UAV collision warning, including a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, it implements the privacy trajectory fuzzy matching method for surface-based UAV collision warning described in Embodiment 1.

[0061] The privacy trajectory fuzzy matching device for surface drone collision warning provided in this embodiment is used to implement the privacy trajectory fuzzy matching method for surface drone collision warning. Therefore, the privacy trajectory fuzzy matching device for surface drone collision warning also has the same technical effects as the method for surface drone collision warning, and will not be described again here.

[0062] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of the present invention.

Claims

1. A privacy trajectory fuzzy matching method for surface-based UAV collision warning, characterized in that, Includes the following steps: Step S1: The initiator and the responder set common parameters related to the collision warning, establish communication between the two parties, and share the common parameters; Step S2: Based on the common parameters, the initiator and the responder respectively execute the hypercube partitioning algorithm to discretize the continuous three-dimensional geographic space into multiple hypercubes; based on the partitioning results, the coordinate information of the trajectory points of both parties is converted into indexed quadruple identifiers. Step S3: The initiator performs OKVS encoding on the three-dimensional position and corresponding timestamp of its own trajectory point and its neighboring points to obtain a message with an OKVS structure, and sends it as a request message to the responder. Step S4: The responder uses Gaussian elimination to linearly decode the request message to generate a DDH rerandomized tuple, constructs a mask message based on the DDH rerandomized tuple, and sends it back to the initiator. Step S5: The initiator decrypts and compares the feedback message to obtain the collision warning judgment result.

2. The privacy trajectory fuzzy matching method for surface-based UAV collision warning according to claim 1, characterized in that, Step S1 specifically includes: Setting the threshold radius: The initiator sets the calculation formula for the threshold radius based on environmental parameters, calculates the real-time radius value, and broadcasts it to the responder; Establish a unified coordinate system: The initiator selects a coordinate origin to construct a unified coordinate system, adds a random perturbation vector to perturb the origin, and synchronizes the offset and updated origin information to the responder; Setting common parameters for the cyclic group: The common parameters for the cyclic group are established through negotiation between both parties; Establish communication between the two parties and share the threshold radius, the origin information, and the common parameters of the cyclic group.

3. The privacy trajectory fuzzy matching method for surface-based UAV collision warning according to claim 1, characterized in that, Step S2 specifically involves: The three-dimensional space is divided into cubes with side lengths twice the threshold radius according to the grid. Map the trajectory points of both parties onto the grid, construct a position sphere with the trajectory point as the center and the radius as the threshold radius, merge multiple cubes that intersect with the position sphere into a block, and obtain the index of the block as the unique identifier of the corresponding trajectory point; The quadruple identifier is obtained by combining the index of the block and the corresponding timestamp index.

4. The privacy trajectory fuzzy matching method for surface-based UAV collision warning according to claim 1, characterized in that, Step S3 specifically involves: Step S31: The initiator queries the responder's trajectory data and initializes the cyclic group parameters to generate a private key and a public key; Step S32: The initiator obtains all neighboring points within a radius of a threshold radius centered on its own trajectory point, and performs hash calculation based on the coordinate information of the trajectory point and its neighboring points and the corresponding block code to generate a list of key-value pairs; Step S33: The initiator maps the key-value pairs in the key-value pair list to a set of four-dimensional group elements in batches; Step S34: Perform OKVS encoding on the four-dimensional group element set to obtain an OKVS structure message; package the public key information and the OKVS structure message and send them to the responder.

5. The privacy trajectory fuzzy matching method for surface-based UAV collision warning according to claim 4, characterized in that, Step S32 specifically involves: For its own trajectory point, the initiator determines the corresponding block number, then enumerates all neighboring points within a sphere centered on the trajectory point and with a threshold radius as the radius, and performs hash calculation to generate a key based on the block number and the coordinate information of its neighboring points; the initiator randomly samples from the cyclic group, generates a corresponding random value for each key, and constructs a list of key-value pairs to be encoded.

6. The privacy trajectory fuzzy matching method for surface-based UAV collision warning according to claim 4, characterized in that, Step S33 specifically includes: The initiator first expands the storage space according to the expansion rate, and then maps the key-value pairs in batches to the four-dimensional group element set.

7. The privacy trajectory fuzzy matching method for surface-based UAV collision warning according to claim 1, characterized in that, Step S4 specifically involves: The responder parses the request message and extracts the public parameters, public key, and OKVS structure. The responder obtains a set of spatial block labels with a side length of four times the threshold radius for its own trajectory points, as well as the timestamp information of the cube containing the corresponding trajectory points; Each spatial block label is combined with its corresponding timestamp to generate a query key. Gaussian elimination is used to perform linear decoding on the messages of the OKVS structure to obtain the group elements. The responder selects a random number as a blinding factor, performs a modular exponentiation operation on the group elements and the initiator's public key to generate a rerandomized ciphertext tuple, and aggregates the ciphertext tuple to generate the DDH rerandomized tuple; The responder calculates the hash value of its own block index and performs an XOR mask operation with the DDH rerandomized tuple to obtain the mask message, and then feeds the mask message back to the initiator.

8. The privacy trajectory fuzzy matching method for surface-based UAV collision warning according to claim 1, characterized in that, Step S5 specifically involves: The initiator parses the masked message, extracts the DDH rerandomized tuple and the masked ciphertext, and verifies it using its own private key. It performs exponentiation and XOR operations on the group element part of the DDH rerandomized tuple, and verifies whether the two parties satisfy the DDH tuple relationship at the current time and space based on the operation results. If they do, it is determined that there is a collision risk; otherwise, it is determined that there is no collision risk.

9. A privacy trajectory fuzzy matching device for surface-based UAV collision warning, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, it implements the privacy trajectory fuzzy matching method for surface-based UAV collision warning as described in any one of claims 1-8.

Images