Privacy protection multi-task allocation method for unmanned aerial vehicle crowdsourcing perception system

Abstract

The invention provides a privacy protection multi-task allocation method for an unmanned aerial vehicle crowdsourcing perception system, which combines addition secret sharing and homomorphic encryption to realize bilateral privacy protection of a task side and an unmanned aerial vehicle position. The task requester and the working party split a task position and a UAV departure position into two secret shares [.] 1 and [.] 2 by adopting an additive secret sharing mechanism respectively, and send the two secret shares [.] 1 and [.] 2 to the two independent service parties S1 and S2 respectively; the service party calculates an encrypted square Euclidean distance between the UAV and the task based on the shared share, and generates two distance matrix shares D1 and D2. On the basis, the two service parties cooperatively execute privacy comparison and a minimum index protocol under the optimized Paillier, and multi-task encryption shortest path distribution is realized in combination with a Hungary algorithm. According to the method, addition secret sharing is fused, Paillier encryption and task matching optimization are optimized, the total flight path of the UAV is remarkably reduced while position information privacy of the whole task allocation process is guaranteed, and good safety and deployability are achieved.

Description

Technical Field

[0001] This invention relates to the field of information security technology, and in particular to a privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles. Background Technology

[0002] With the rapid development of mobile internet, IoT, and UAV (Unmanned Aerial Vehicle) technologies, UAV-based mobile crowdsensing (MCS) has gradually emerged and is widely used in low-altitude data collection scenarios such as traffic management, environmental monitoring, and urban sensing. In a typical UAV-MCS system, the task requester submits a sensing task to the platform. The platform then assigns the task to a suitable UAV operator based on the flight distance between the UAV's takeoff point and the task location, optimizing system efficiency and energy consumption. Task allocation is the core issue of a UAV-MCS system, and its efficiency directly affects system operating costs. However, the geographical location of sensing tasks often carries sensitive information about the task requester (such as coordinates of homes, businesses, or medical facilities), and the operator's UAV departure location may also reveal their activity area, daily trajectory, and even social identity. Therefore, directly optimizing paths based on plaintext coordinates during task allocation faces a serious risk of location privacy leakage.

[0003] To alleviate these problems, researchers have proposed a series of privacy-preserving task allocation schemes based on differential privacy or cryptographic computation. However, these methods generally suffer from the following drawbacks: some schemes only protect the privacy of the worker or the task requester, failing to achieve bilateral location privacy protection; differential privacy methods, while efficient, sacrifice task allocation accuracy due to the introduction of noise perturbations, especially in multi-task scenarios, significantly lengthening flight paths; while homomorphic encryption can support precise computation, cryptographic schemes often face high overhead in the comparison and allocation stages, making them difficult to deploy in low-latency UAV systems; the privacy-preserving allocation problem in multi-task scenarios is more complex, involving minimum total path matching, and its solution requires multiple steps such as comparability, minimum value index extraction, and matrix iterative optimization, while most existing schemes cannot simultaneously achieve a balance between efficiency, privacy, and accuracy.

[0004] Therefore, there is an urgent need for a privacy protection scheme that can simultaneously protect the location information of the task requester and the worker, support accurate comparison operations, and efficiently perform multi-task allocation optimization, so as to meet the dual requirements of privacy and system performance in real-world UAV-MCS scenarios. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles (UAVs). The aim is to achieve the extraction of directional gradient histogram features without decrypting the original image data, thereby improving the data security and privacy protection capabilities of computer vision tasks.

[0006] The technical solution of this invention is implemented as follows:

[0007] A privacy-preserving multi-task allocation method for a drone crowdsourced perception system includes the following steps:

[0008] Step 1, System Initialization: The system deploys two independent and non-collusive service providers. and Both have storage and computing capabilities, and under the security assumption, they do not collude with each other;

[0009] Step 2, Encrypted Data Upload: Each task requester uploads the task location. With the task content Two shares are generated through addition and secret sharing. and in, , And respectively Send to , Send to the service provider Each drone operator also displays their drone's current location information. Perform addition and secret sharing to generate a location ciphertext share: , and will Send to , Send to ;

[0010] Step 3, Distance Matrix Construction and Standardization: In a multi-task scenario, there is a task set. Each task is performed by a drone operated by a worker, who is a group of workers. ,in Service Provider and Calculate the position of each pair of tasks separately. and the drone's starting position The encrypted squared Euclidean distance between them: , And generate two distance matrix shares and Then the non-square matrix was standardized to The square matrix is ​​adapted to the input requirements of the Hungarian algorithm;

[0011] Step 4: Privacy-preserving Hungarian algorithm execution: Service provider and Based on their respective distance shares and The Hungarian allocation process, which coordinates and protects privacy, is implemented through the following steps;

[0012] Step 41: Execute row rules, jointly determine the column index corresponding to the minimum encrypted squared distance of each row in the distance matrix through the privacy protection minimum value index protocol, and then subtract the minimum value from all elements in that row;

[0013] Step 42: Execute column rules, determine the row index of the minimum value in each column through the privacy-preserving minimum index protocol, and then subtract the minimum value from all elements in that column;

[0014] Step 43: Selecting Independent Zero Elements. Attempt to find a set of "independent zero elements" in the current row and column normalized matrix to construct a feasible allocation scheme. If the final number of marked independent zero elements is... If the matrix order is specified, it indicates that the optimal allocation has been found, and each marked position... That is to say, the task Assigned to workers If there are fewer than 10 independent zero elements Then matrix adjustment is required;

[0015] Step 5, Task Distribution and Resumption: After task distribution is completed, the service provider... and The secret share of the perception task is sent to the selected worker. The worker then reconstructs the complete task information through additive secret sharing and executes the perception task.

[0016] As a further optimization of the above scheme, in step 2, the secret sharing mechanism is additive secret sharing, that is, for any location coordinate value or task content data... The task requester or the worker selects a module within the domain. Uniform random numbers And generate two secret shares as follows: , ,in Large prime numbers that are publicly known;

[0017] As a further optimization of the above solution, in step 2, the task content... It consists of character text, where each character is converted into an 8-bit binary number according to ASCII encoding and then represented as an integer.

[0018] As a further optimization of the above scheme, in step 3, if the number of tasks m is not equal to the number of drones n, the original m×n distance matrix needs to be standardized: by introducing virtual tasks to fill in the gaps to n×n, the requirements of the Hungarian algorithm for the input matrix structure are met, and the corresponding element values ​​of the newly added rows or columns are zero.

[0019] As a further optimization of the above scheme, step 4 includes the following steps in the privacy-preserving minimum index protocol:

[0020] Step 51, Service Provider and Each holds a share of the distance matrix. and First, we agree to set index 1 as the position of the current minimum value. ;

[0021] Step 52, starting from position index 2 to Perform the following operations;

[0022] Step 53, Service Provider and Invoking a privacy-preserving comparison protocol to determine the encrypted distance share of the current candidate position , Share with the current minimum position , As input, the two sizes are compared. If the privacy-preserving comparison protocol returns "new position smaller", then both parties will... Updated to Otherwise, remain unchanged;

[0023] Step 54: After all positions have been compared, This refers to the index pointing to the smallest encryption distance in that row (or column). (Service provider) Output the index directly and inform the service provider. .

[0024] As a further optimization of the above solutions, the privacy protection comparison protocol steps are as follows:

[0025] Step 61, Service Provider From the difference between its two distance shares In the middle, randomly add a large module Obtain the mask value Subsequently, Using Paillier public key right Paillier homomorphic encryption is performed to obtain the ciphertext. and the mask value ciphertext and Send to ;

[0026] Step 62 After receiving the data, add the difference in its own shareholding to the mask to obtain... ,Then, Then, the value and a new random number were used. Perform Paillier homomorphic encryption on each side to obtain the ciphertext. and .

[0027] Step 63 Constructing ciphertext using existing numbers Return it to ;

[0028] Step 64 Using Paillier's private key Decryption To obtain a value .if Greater than key parameters Explain the original difference It is positive, therefore the judgment is... If the condition is met, the output result is 1; otherwise, the output result is 0.

[0029] As a further optimization of the above scheme, the matrix adjustment steps are as follows:

[0030] Step 71: Mark all rows that do not contain independent zero elements (indicated by a "check mark");

[0031] Step 72: In each marked row, find the columns containing all the crossed-out zero elements and mark these columns;

[0032] Step 73: In each marked column, find the rows containing all the crossed-out individual zeros and mark these rows;

[0033] Step 74: Repeat steps 72 and 73 until no more markers can be added;

[0034] Step 75: Underline all elements except unmarked rows and marked columns, and then extract the minimum value from these uncovered elements using a privacy-preserving minimum indexing protocol;

[0035] Step 76: Subtract the minimum value from all elements of each marked row, and add the minimum value to all elements of each marked column;

[0036] Step 77: Return to the "Independent Zero Element Selection" step and continue iterating.

[0037] As a further optimization of the above scheme, the random number mentioned in the privacy protection comparison protocol... Taken from the addition secret shared module domain random numbers , ,and ,in For safety parameters .

[0038] As a further optimization of the above scheme, the Paillier key generation, encryption, and decryption algorithm is implemented in the following manner:

[0039] Key generation: Randomly select two pairs of coprime odd prime numbers and And construct prime numbers respectively. , Calculate the modulus and private key parameters. , Select Calculate the auxiliary public key factor , Generate public key and private key .

[0040] Encryption algorithm: for plaintext Randomly select mask factor Calculate ciphertext .

[0041] Decryption algorithm: For ciphertext Calculate using private key ,in .

[0042] Compared with the prior art, the present invention achieves the following beneficial effects:

[0043] (1) This invention proposes a privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles. For the first time, it realizes bilateral privacy protection of the location of the task requester and the worker. Through the additive secret sharing mechanism, the real location information is split into two independent shares, which are stored and processed by two non-cooperative service parties respectively. This avoids the risk of leakage of location data during transmission or calculation. Compared with the method that only protects the privacy of one party, it is more comprehensive and reliable.

[0044] (2) This invention designs a privacy-preserving comparison protocol that supports distance comparison in the encrypted domain. Combining additive secret sharing and optimized Paillier encryption system, it can efficiently and accurately compare the square of the flight distance between two without plaintext involvement. This overcomes the problems of low efficiency in traditional homomorphic encryption schemes or precision loss in differential privacy schemes, and ensures the feasibility of optimal task allocation.

[0045] (3) This invention constructs a privacy-preserving Hungarian algorithm. Based on the encrypted distance matrix, it supports minimum path optimization in UAV multi-task scenarios by designing a secure minimum value index extraction protocol and matrix adjustment process. It ensures that the task allocation result achieves the shortest total path that is similar to the non-privacy scheme while maintaining privacy, which is significantly better than the existing encryption methods that only support single task or approximate allocation.

[0046] (4) This invention protects privacy while taking into account system overhead. The proposed encryption protocol and allocation mechanism have low communication and computational complexity, light load on the task requester and worker devices, support parallel deployment on the platform side service provider, are suitable for UAV-MCS task scheduling environment with high real-time requirements, and have good scalability and practical deployment potential. Attached Figure Description

[0047] Figure 1 This is an algorithm schematic diagram of a privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles provided in an embodiment of the present invention;

[0048] Figure 2 This is an interactive diagram of a privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles, provided by an embodiment of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions in the embodiments of this invention 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 this invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0050] like Figure 1 , Figure 2 As shown, this embodiment provides a privacy-preserving multi-task allocation method for a drone crowdsourced perception system, including the following steps:

[0051] Step 1, System Initialization: The data holder generates a public modulus for secret sharing. =4294967311; The data holder recruited two unrelated first cloud servers. Second cloud server Send the module to all cloud servers Cloud servers possess powerful computing capabilities and can store encrypted data.

[0052] Step 2, Encrypted Upload of Location Data: The original assumption was that there were 2 sensing tasks and three drones (and their operators):

[0053] Task / Employee ID Original location (latitude, longitude) <![CDATA[t1]]> (30.0001, 120.0002) <![CDATA[t2]]> (30.0010, 120.0020) <![CDATA[w1]]> (29.9990, 119.9995) <![CDATA[w2]]> (30.0005, 120.0000) <![CDATA[w3]]> (30.0020, 120.0015)

[0054] Step 3, Secret Sharing of Task Locations: For the dimension x of task t1 t1 = 30.0001, perform fixed-point encoding:

[0055] Randomly select r = 1234567890 and generate: [x t1 1 = r mod p = 1234567890, [x t1 2 = (30000100 - r) mod p = 3062529421.

[0056] Similarly, generate shares for longitude and task content c i and send [t1]1 = <[c1]1, ([x t1 1, [y t1 1)> to S1, and [t1]2 = <[c1]2, ([x t1 2, [y t1 2)> to S2;

[0057] Step 4, Secret Sharing of Drone Locations: Perform the same processing on the location of worker w1 to generate: [x w1 1 = r mod p = 987654321, [x w1 2 = (29999000 - 987654321) mod p = 330751698.

[0058] Similarly, generate shares for longitude and task content c i and send [x w1 1, [y w1 1 to S1, and [x w1 2, [y w1 2 to S2.

[0059] Step 5, Construction and Standardization of Distance Matrix: Calculate the squared Euclidean distance to obtain the distance matrix , Obtain the distance matrix , since m = 2 < n = 3, append a row of virtual task rows at the end of the matrix, and set all elements to 0 to obtain a 3×3 square matrix share;

[0060] Step 6, Execute the Privacy-Preserving Hungarian Algorithm until 3 independent 0s are obtained in the matrix;

[0061] Step 7, Task Distribution and Recovery: and Summarize independent zero index pairs , issue the corresponding and For workers Worker share restored And perform the task;

[0062] Based on the disclosure and teachings of the above specification, those skilled in the art can make changes and modifications to the above embodiments. Therefore, the present invention is not limited to the specific embodiments disclosed and described above, and some modifications and changes to the present invention should also fall within the protection scope of the claims of the present invention. Furthermore, although some specific terms are used in this specification, these terms are only for convenience of explanation and do not constitute any limitation on the present invention.

Claims

1. A privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: Step 1, System Initialization: The system deploys two independent and non-collusive service providers. and Both have storage and computing capabilities, and under the security assumption, they do not collude with each other; Step 2, Encrypted Data Upload: Each task requester uploads the task location. With the task content Two shares are generated through addition and secret sharing. and in, , And respectively Send to , Send to the service provider Each drone operator also displays their drone's current location information. Perform addition and secret sharing to generate a location ciphertext share: , and will Send to , Send to ; Step 3, Distance Matrix Construction and Standardization: In a multi-task scenario, there is a task set. Each task is performed by a drone operated by a worker, who is a group of workers. ,in Service Provider and Calculate the position of each pair of tasks separately. and the drone's starting position The encrypted squared Euclidean distance between them: , And generate two distance matrix shares and Then the non-square matrix was standardized to The square matrix is ​​adapted to the input requirements of the Hungarian algorithm; Step 4: Privacy-preserving Hungarian algorithm execution: Service provider and Based on their respective distance shares and The Hungarian allocation process, which coordinates and protects privacy, is implemented through the following steps; Step 41: Execute row rules, jointly determine the column index corresponding to the minimum encrypted squared distance of each row in the distance matrix through the privacy protection minimum value index protocol, and then subtract the minimum value from all elements in that row; Step 42: Execute column rules, determine the row index of the minimum value in each column through the privacy-preserving minimum index protocol, and then subtract the minimum value from all elements in that column; Step 43: Selecting Independent Zero Elements. Attempt to find a set of "independent zero elements" in the current row and column normalized matrix to construct a feasible allocation scheme. If the final number of marked independent zero elements is... If the matrix order is specified, it indicates that the optimal allocation has been found, and each marked position... That is to say, the task Assigned to workers If there are fewer than 10 independent zero elements Then matrix adjustment is required; Step 5, Task Distribution and Resumption: After task distribution is completed, the service provider... and The secret share of the perception task is sent to the selected worker. Subsequently, the worker recovers the complete task information through additive secret sharing and executes the perception task.

2. The privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, In step 2, the secret sharing mechanism is additive secret sharing, that is, for any location coordinate value or task content data... The task requester or worker selects a module in the domain. Uniform random numbers Two secret shares are generated: [α]1 = r, [α]2 = α - r mod p, where p = 2 ξ , is a publicly known large prime number.

3. The privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, In step 2, the task content It consists of character text, where each character is converted into an 8-bit binary number according to ASCII encoding and then represented as an integer.

4. The privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, In step 3, if the number of tasks Not equal to the number of drones Then it is necessary to examine the original The distance matrix is ​​standardized by introducing virtual tasks to complete it. This satisfies the Hungarian algorithm's requirements for the input matrix structure, with newly added rows or columns having corresponding element values ​​of zero.

5. A privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, In step 4, the privacy-preserving minimum index protocol includes the following steps: Step 51, Service Provider and Each holds a share of the distance matrix. and First, we agree to set index 1 as the position of the current minimum value. ; Step 52, starting from position index 2 to Perform the following operations; Step 53, Service Provider and Invoking a privacy-preserving comparison protocol to determine the encrypted distance share of the current candidate position , Share with the current minimum position , As input, the two sizes are compared. If the privacy-preserving comparison protocol returns "new position smaller", then both parties will... Updated to Otherwise, remain unchanged; Step 54: After all positions have been compared, This refers to the index pointing to the smallest encryption distance in that row (or column). (Service provider) Output the index directly and inform the service provider. .

6. A privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles (UAVs) according to claim 5, characterized in that, The privacy protection comparison agreement steps are as follows: Step 61, Service Provider From the difference between its two distance shares In the middle, randomly add a large module Obtain the mask value Subsequently, Using Paillier public key right Paillier homomorphic encryption is performed to obtain the ciphertext. and the mask value ciphertext and Send to ; Step 62 After receiving the data, add the difference in its own shareholding to the mask to obtain... ,Then, Then, the value and a new random number were used. Perform Paillier homomorphic encryption on each side to obtain the ciphertext. and .

7. Step 63 Constructing ciphertext using existing numbers Return it to ; Step 64 Using Paillier's private key Decryption To obtain a value .if Greater than key parameters Explain the original difference It is positive, therefore the judgment is... If the condition is met, the output result is 1; otherwise, the output result is 0.

8. A privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The matrix adjustment steps are as follows: Step 71: Mark all rows that do not contain independent zero elements (indicated by a "check mark"); Step 72: In each marked row, find the columns containing all the crossed-out zero elements and mark these columns; Step 73: In each marked column, find the rows containing all the crossed-out individual zeros and mark these rows; Step 74: Repeat steps 72 and 73 until no more markers can be added; Step 75: Underline all elements except unmarked rows and marked columns, and then extract the minimum value from these uncovered elements using a privacy-preserving minimum indexing protocol; Step 76: Subtract the minimum value from all elements of each marked row, and add the minimum value to all elements of each marked column; Step 77: Return to the "Independent Zero Element Selection" step and continue iterating.

9. A privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles (UAVs) according to claim 5, characterized in that, The random number described in the privacy protection comparison protocol Taken from the addition secret shared module domain random numbers , ,and ,in For safety parameters .

10. A privacy-preserving multi-task allocation method for a crowdsourced perception system for unmanned aerial vehicles (UAVs) according to claim 6, characterized in that, The Paillier key generation, encryption, and decryption algorithms are implemented as follows: Key generation: Randomly select two pairs of coprime odd prime numbers and And construct prime numbers respectively. , Calculate the modulus and private key parameters. , Select Calculate the auxiliary public key factor Generate public key and private key .

11. Encryption Algorithm: For plaintext Randomly select mask factor Calculate ciphertext .

12. Decryption Algorithm: For ciphertext Calculate using private key ,in .

Images