Privacy-preserving multi-task data aggregation method in blockchain-assisted crowd sensing

Abstract

The present invention discloses a multi-task aggregation method for protecting privacy in blockchain-assisted crowd intelligence perception. The method uses the Chinese remainder theorem to merge multiple task data involved in users into one piece of data and participate in data aggregation, that is, it can aggregate multiple concurrent tasks from multiple task requesters, which not only improves the task concurrency of privacy-protected data aggregation, but also reduces the computational overhead from privacy protection. The method generates a shared mask based on key negotiation to mask data and protect the privacy of user data and requester aggregation. The method also supports multiple data aggregation types, including sum, average, variance, weighted sum, and weighted average. In addition, by deploying smart contracts on the blockchain to assist in key negotiation and data verification, the reliability of aggregate calculations is guaranteed without the need for a trusted third party.

Description

Technical Field

[0001] The present invention relates to the technical field of information security and privacy protection, and in particular to a multi-task data aggregation method for protecting privacy in blockchain-assisted crowd intelligence perception. Background Art

[0002] In recent years, with the continuous upgrading of hardware, smartphone devices have become more popular and have become an indispensable part of people's work and life. In this context, mobile crowd sensing (MCS) as a data collection method performed by workers has attracted widespread attention. Compared with traditional wireless sensor networks (WSNs), MCS has the characteristics of low cost, scalability, flexibility and efficiency. These characteristics make MCS a key component of the emerging Internet of Things (IoT) and an important driving force for the development of smart cities, especially in applications such as smart transportation, environmental monitoring, smart medical care and smart grids. Data aggregation, as an important means of statistical analysis, is a key component of MCS. However, the data uploaded by workers often contains a lot of personal information, such as identity, location and preferences. Privacy issues reduce the enthusiasm of workers to participate in MCS tasks, which in turn hinders the application and deployment of MSC. For the task requester (TR), obtaining the aggregated results usually requires a certain cost. Therefore, TR regards these statistical results as their own private assets and is unwilling to share them with others. Therefore, data aggregation schemes should protect the privacy of workers and TRs.

[0003] In practical applications, as MCS is increasingly integrated into daily life, scenarios involving concurrent multitasking and multiple requesters have become very common. Therefore, researchers have also focused on multitasking scenarios, such as multitasking allocation, multitasking data aggregation, and multitasking incentive mechanisms. According to the number of requesters and tasks, existing data aggregation methods in MCS can be divided into three categories: single-task data aggregation (one task and one TR), multi-dimensional data aggregation (multiple tasks and one TR), and multi-task data aggregation (multiple tasks and multiple TRs). The present invention is committed to realizing privacy-preserving multitasking data aggregation in MCS. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings and deficiencies of existing data aggregation solutions in multi-task scenarios, and proposes a multi-task data aggregation method based on blockchain-assisted crowd intelligence perception to protect privacy. It can aggregate multiple concurrent tasks from multiple task requesters in one round of communication, and it supports multiple data aggregation types, including sum, average, variance, weighted sum, and weighted average. In addition, the present invention ensures that the privacy of the worker's data and the aggregation results of the task requester can be protected even under collusion attacks. The privacy protection of the perception data is achieved by using the shared mask generated by the key negotiation between the worker and the requester, and the multiple task data of the worker are merged into one through the Chinese remainder theorem.

[0005] To achieve the above purpose, the technical solution provided by the present invention is: a multi-task data aggregation method based on blockchain-assisted crowd intelligence perception to protect privacy, comprising the following steps:

[0006] 1) System initialization: The perception platform generates and publishes public parameters and deploys smart contracts on the blockchain. Workers and requesters register with the system and generate public and private keys.

[0007] 2) Task publishing: The requester sends the perception task to the perception platform, which leads the worker selection and task allocation.

[0008] 3) Mask calculation: The selected worker and the requester upload their public keys to the blockchain, calculate the shared key between the two parties by querying the other party’s public key, and further generate a shared mask.

[0009] 4) Multi-task data submission: Each worker collects the perception data of multiple tasks and performs data masking and multi-dimensional data fusion, generates a verification sequence and uploads it to the blockchain, and sends the multi-task data to the perception platform.

[0010] 5) Multi-task data aggregation: The perception platform batch verifies and aggregates the multi-task data submitted by workers. The aggregation results of multiple tasks with masks are extracted from the aggregation results and sent to the corresponding requesters.

[0011] 6) Aggregation result verification and demasking: Each requester verifies the correctness of the masked aggregation result received from the perception platform and removes the mask to obtain the task result.

[0012] 7) Implementation of multiple aggregation types: According to the aggregation requirements of the requester, the perception platform can adjust the calculations of the above steps accordingly to obtain multiple task results such as sum, mean, variance, weighted sum, weighted mean, etc.

[0013] In step 1), the detailed process of system initialization is as follows:

[0014] 1.1) Initialization of the perception platform: First, the perception platform specifies the security parameter λ and randomly generates a large prime number q that satisfies ||q|| = λ. From the finite field determined by q The generator g is randomly selected from the . Then the perception platform generates a set of prime numbers ,in is the upper limit of the number of crowd sensing tasks. .make ,calculate ,in Given three different secure hash functions And a secure elliptic curve Finally, the perception platform publishes public parameters , and deploy blockchain smart contract code.

[0015] 1.2) Worker and requester initialization: Each worker and requester registers a decentralized identity DID in the blockchain deployed by the perception platform. Assume that the DIDs of n workers and l requesters are , Each worker and requester follows the elliptic curve Generate their own public and private keys and upload their public keys to the blockchain. and requester The public and private keys are denoted as .

[0016] In step 2), the task publishing process is as follows:

[0017] Assume that all requesters A task request is sent to the perception platform, with DID as The task request of the requester is recorded as ,in They are task ID, task details, aggregation type (such as sum, variance) and expiration time. After the perception platform receives tasks from l requesters, it will Broadcast to a collection of workers .

[0018] In step 3), the detailed process of mask calculation is as follows:

[0019] 3.1) Workers and Each holds a unique elliptic curve key pair First, the worker and the requester query the public keys of all entities of the other party through the smart contract. Perform key negotiation with all requesters in R to obtain a list of shared keys .in . Similar to the worker operation, the requester Also calculated .in The public key transfer of the above key negotiation is completed through the blockchain smart contract code.

[0020] 3.2) Finally, all workers and requesters further calculate the shared mask with their respective shared key lists: .for and The resulting mask lists are . Similarly, , different lists of masks can be calculated in turn: The above processes are all interactive and can be performed offline, and can be executed asynchronously with subsequent steps.

[0021] In step 4), the process of multi-task data submission is as follows:

[0022] 4.1) Workers From the task list published by the perception platform Select the appropriate task and generate a decision vector based on the selection result , where each element takes a Boolean value, 1 represents participation in the corresponding task, and 0 represents non-participation in the corresponding task. Here we assume Participated in all tasks, then The perception data of all tasks are collected and the perception data vector is obtained .

[0023] 4.2) Workers According to the first two shared mask lists obtained in step 3), and To cover up, we get and According to the Chinese remainder theorem and the public parameter, workers Multidimensional data with mask , Fusion into single-dimensional multi-task data .

[0024] 4.3) To ensure data integrity, Generate a verification sequence for a message: .at last Will Send to the perception platform Upload to the blockchain. The rest of the workers do the same.

[0025] In step 5), the process of multi-task data aggregation is as follows:

[0026] 5.1) Eliminate exited users: Due to user device or network issues, users may exit during the crowd intelligence perception process, resulting in the failure to submit data within the scheduled time. After the deadline is reached, the perception platform counts the users who have not sent messages and excludes them from the user set. The remaining users are set as the set .

[0027] 5.2) Data aggregation and verification: Assume that the perception platform receives user data from workers , sum up to get multi-task aggregate data: . Afterwards To verify, if the equation If the equation is not true, then the result is correct. Verify one by one: , remove all invalid multi-task data and then aggregate them.

[0028] 5.3) Data extraction: The perception platform Extract the aggregation results corresponding to each requester in turn: Finally, the perception platform will Sent to the corresponding requester .

[0029] In step 6), the detailed process of aggregation result verification and masking is as follows:

[0030] 6.1) Data Verification: Requester Received Afterwards, according to Confirm that the user has not exited and call the smart contract to query the worker verification sequence set stored in the blockchain Then for the masked task results Verify separately, if the equation If it is established, it means the result is correct and you can proceed to the next step. Otherwise, it means the integrity of the task result is destroyed.

[0031] 6.2) Data masking: Query the first two masks in your list to match All masks of are summed up separately to get . Unmask the task result to get the correct task result: .in is the result of data summation, is the number of workers involved. The remaining requesters operate with Same reason.

[0032] In step 7), the process of implementing multiple aggregation types is as follows:

[0033] 7.1) Mean: From step 7) requester Easy to get the mean: .

[0034] 7.2) Variance: The perception platform needs to recalculate a set of public prime numbers, let ,calculate ,in . After that, each worker submits additional multi-task data of the squared value of the perception data: ,in . Aggregated and extracted by the perception platform, the requester Available: . Finally, remove the mask to get the variance value: .

[0035] 7.3) Weighted sum and weighted average: The perception platform assigns a weight vector to each worker. The weight vector is , The last submitted data is: After aggregation and masking, the requester The weighted summation result can be obtained: . And the weighted average result can be further obtained: ,in .

[0036] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0037] 1. The present invention realizes multi-task data aggregation by using the Chinese remainder theorem, and can aggregate task data of multiple requesters in one round of communication. Compared with the existing multi-task data aggregation method using secret sharing, it has lower computational cost (less addition, multiplication and modular operations).

[0038] 2. The present invention ensures the privacy of each worker and requester, and is automatically executed by the smart contract deployed on the blockchain. The required mask can be generated through key negotiation in an offline and non-interactive manner without a trusted third party, thus achieving privacy protection. In addition, the present invention implements the integrity verification of the perception data submitted to the blockchain through signatures.

[0039] 3. The present invention realizes multiple types of data aggregation, including sum, average, variance, weighted sum and weighted average, which significantly expands the practical application scope of multi-task data aggregation. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a multi-task conceptual model diagram of the present invention.

[0041] Figure 2 Schematic diagram of the system model of the present invention.

[0042] Figure 3 It is the work flow chart of the present invention. Figure 4 It is a schematic diagram of the system operation of the present invention. DETAILED DESCRIPTION

[0043] The present invention will be further described below with reference to the accompanying drawings and examples.

[0044] The multi-task data aggregation method based on blockchain-assisted crowd intelligence perception with privacy protection provided in this example integrates multiple task data of users into one piece of data and participates in data aggregation through the Chinese remainder theorem, which not only improves the task concurrency of privacy-preserving data aggregation, but also reduces the computational overhead from privacy protection. In order to protect data privacy and adapt to multi-task data aggregation, key negotiation is used to generate a shared mask to mask data and protect the privacy of user data and requester aggregation. In addition, by deploying smart contracts on the blockchain to assist key negotiation and data verification, the reliability of aggregate computing is guaranteed without the need for a trusted third party.

[0045] like Figure 1 As shown, according to the number of requesters and tasks, existing data aggregation methods in MCS can be divided into three categories: single-task data aggregation (one task and one TR), multi-dimensional data aggregation (multiple tasks and one TR), and multi-task data aggregation (multiple tasks and multiple TRs). The present invention implements the third category, privacy-preserving multi-task data aggregation in MCS.

[0046] like Figure 2 As shown, the system model of the present invention includes the following entities: requester, perception platform and worker. Note that in the first step, each worker and requester will register and upload the elliptic curve public key on the blockchain for key negotiation. The following is an introduction to each entity:

[0047] Requester: It consists of multiple requesters, who are the initiators of multiple tasks. Due to the demand for sensor data in different areas or time periods, they cannot complete the tasks independently, which prompts them to initiate tasks to the sensor platform. In the present invention, each requester first sends their tasks to the requester. After that, a unique shared mask is established with each worker in an offline manner on the blockchain.

[0048] Perception Platform: The perception platform receives tasks from requesters and publishes them to workers. After receiving the workers’ sensor data, it is responsible for verifying and aggregating these multi-task data. Then it extracts the aggregated results of each covered task and sends them to the corresponding requester.

[0049] Workers: Responsible for completing multiple tasks issued by the perception platform. Collect and upload sensor data, and generate corresponding signatures for each sensor data.

[0050] like Figure 3 As shown, the above-mentioned multi-task data aggregation method based on privacy protection in blockchain-assisted crowd intelligence perception includes the following steps:

[0051] 1) System initialization: The perception platform generates and publishes public parameters and deploys smart contracts on the blockchain. Workers and requesters register the system and generate public and private keys. The detailed process of system initialization is as follows:

[0052] 1.1) Initialization of the perception platform: First, the perception platform specifies the security parameter λ and randomly generates a large prime number q that satisfies ||q|| = λ. From the finite field determined by q The generator g is randomly selected from the . Then the perception platform generates a set of prime numbers ,in is the upper limit of the number of crowd sensing tasks. .make ,calculate ,in Given three different secure hash functions And a secure elliptic curve Finally, the perception platform publishes public parameters , and deploy blockchain smart contract code.

[0053] 1.2) Worker and requester initialization: Each worker and requester registers a decentralized identity DID in the blockchain deployed by the perception platform. Assume that the DIDs of n workers and l requesters are , Each worker and requester follows the elliptic curve Generate their own public and private keys and upload their public keys to the blockchain. and requester The public and private keys are denoted as .

[0054] 2) Task release: The requester sends the perception task to the perception platform, which leads the worker selection and task allocation. The process of task release is as follows:

[0055] Assume that all requesters A task request is sent to the perception platform, with DID as The task request of the requester is recorded as ,in They are task ID, task details, aggregation type (such as sum, variance) and expiration time. After the perception platform receives tasks from l requesters, it will Broadcast to a collection of workers The platform allows workers to freely participate in perception tasks. The selection and assignment of workers in crowd-sensing belongs to another research area and is not covered by this patent.

[0056] 3) Mask calculation: The selected worker and the requester upload the public key to the blockchain, calculate the shared key between the two parties by querying the other party's public key, and further generate a shared mask. In step 3), the detailed process of mask calculation is as follows:

[0057] 3.1) Workers and Each holds a unique elliptic curve key pair First, the worker and the requester query the public keys of all entities of the other party through the smart contract. Perform key negotiation with all requesters in R to obtain a list of shared keys .in . Similar to the worker operation, the requester Also calculated .in The public key transfer of the above key negotiation is completed through the blockchain smart contract code.

[0058] 3.2) Finally, all workers and requesters further calculate the shared mask with their respective shared key lists: .for and The resulting mask lists are . Similarly, , different lists of masks can be calculated in turn: The above processes are all interactive and can be performed offline, and can be executed asynchronously with subsequent steps.

[0059] 4) Multi-task data submission: Each worker collects the perception data of multiple tasks and performs data masking and multi-dimensional data fusion, generates a verification sequence and uploads it to the blockchain, and sends the multi-task data to the perception platform. The detailed process of multi-task data submission is as follows:

[0060] 4.1) Workers From the task list published by the perception platform Select the appropriate task and generate a decision vector based on the selection result , where each element takes a Boolean value, 1 represents participation in the corresponding task, and 0 represents non-participation in the corresponding task. Here we assume Participated in all tasks, then The perception data of all tasks are collected and the perception data vector is obtained .

[0061] 4.2) Workers According to the first two shared mask lists obtained in step 3), and To cover up, we get and According to the Chinese remainder theorem and the public parameter, workers Multidimensional data with mask , Fusion into single-dimensional multi-task data .

[0062] 4.3) To ensure data integrity, Generate a verification sequence for a message: .at last Will Send to the perception platform Upload to the blockchain. The rest of the workers do the same.

[0063] 5) Multi-task data aggregation: The perception platform batch verifies and aggregates the multi-task data submitted by the workers. The aggregation results of multiple tasks with masks are extracted from the aggregation results and sent to the corresponding requesters. In step 5), the detailed process of multi-task data aggregation is as follows:

[0064] 5.1) Eliminate exited users: Due to user device or network issues, users may exit during the crowd intelligence perception process, resulting in the failure to submit data within the scheduled time. After the deadline is reached, the perception platform counts the users who have not sent messages and excludes them from the user set. The remaining users are set as the set .

[0065] 5.2) Data aggregation and verification: Assume that the perception platform receives user data from workers , sum up to get multi-task aggregate data: . Afterwards To verify, if the equation If the equation is not true, then the result is correct. Verify one by one: , remove all invalid multi-task data and then aggregate them.

[0066] 5.3) Data extraction: The perception platform Extract the aggregation results corresponding to each requester in turn: Finally, the perception platform will Sent to the corresponding requester .

[0067] 6) Aggregation result verification and demasking: Each requester verifies the correctness of the masked aggregation result received from the perception platform, removes the mask and obtains the task result. The detailed process of aggregation result verification and demasking is as follows:

[0068] 6.1) Data Verification: Requester Received Afterwards, according to Confirm that the user has not exited and call the smart contract to query the worker verification sequence set stored in the blockchain Then for the masked task results Verify separately, if the equation If it is established, it means the result is correct and you can proceed to the next step. Otherwise, it means the integrity of the task result is destroyed.

[0069] 6.2) Data masking: Query the first two masks in your list to match All masks of are summed up separately to get . Unmask the task result to get the correct task result: .in is the result of data summation, is the number of workers involved. The remaining requesters operate with Same reason.

[0070] 7) Implementation of multiple aggregation types: According to the aggregation requirements of the requester, the perception platform can adjust the calculations of the above steps accordingly to obtain multiple task results such as sum, mean, variance, weighted sum, weighted mean, etc. The detailed process of implementing multiple aggregation types is as follows:

[0071] 7.1) Mean: From step 7) requester Easy to get the mean: .

[0072] 7.2) Variance: The perception platform needs to recalculate a set of public prime numbers, let ,calculate ,in . After that, each worker submits additional multi-task data of the squared value of the perception data: ,in . Aggregated and extracted by the perception platform, the requester Available: . Finally, remove the mask to get the variance value: .

[0073] 7.3) Weighted sum and weighted average: The perception platform assigns a weight vector to each worker. The weight vector is , The last submitted data is: After aggregation and masking, the requester The weighted summation result can be obtained: . And the weighted average result can be further obtained: ,in .

Claims

1. A privacy-preserving multi-task data aggregation method based on crowd intelligence assisted by blockchain, characterized in that: The following steps are included: 1) System initialization: The perception platform generates and publishes public parameters, and deploys smart contracts on the blockchain. Workers and requesters register the system and generate public and private keys. 2) Task release: The requester sends the perception task to the perception platform, and the platform leads the worker selection and task allocation. 3) Mask calculation: The selected worker and requester upload the public key to the blockchain, calculate the shared key between the two parties by querying the other party's public key, and further generate a shared mask. 4) Multi-task data submission: Each worker collects the perception data of multiple tasks and performs data masking and multi-dimensional data fusion, generates a verification sequence and uploads it to the blockchain, and sends the multi-task data to the perception platform. 5) Multi-task data aggregation: The perception platform performs batch verification and data aggregation on the multi-task data submitted by the workers. The aggregation results of multiple tasks with masks are extracted from the aggregation results and sent to the corresponding requesters. 6) Aggregation result verification and demasking: Each requester verifies the correctness of the masked aggregation results received from the perception platform, removes the mask to obtain the task results. 7) Implementation of multiple aggregation types: According to the aggregation requirements of the requester, the perception platform can adjust the calculations of the above steps accordingly to obtain multiple task results such as sum, mean, variance, weighted sum, weighted mean, etc.

2. The method for multi-task data aggregation based on blockchain-assisted crowd intelligence perception and privacy protection according to claim 1 is characterized in that: In step 1), the detailed process of system initialization is as follows: a) Perception platform initialization: First, the perception platform specifies the security parameter λ and randomly generates a large prime number q that satisfies ||q|| = λ. From the finite field determined by q The generator g is randomly selected from the . Then the perception platform generates a set of prime numbers ,in is the upper limit of the number of crowd sensing tasks. .make ,calculate ,in Given three different secure hash functions And a secure elliptic curve Finally, the perception platform publishes public parameters , and deploy blockchain smart contract code. b) Worker and requester initialization: Each worker and requester registers a decentralized identity DID in the blockchain deployed by the perception platform. Assume workers and The DIDs of the requesters are , Each worker and requester follows the elliptic curve Generate their own public and private keys and upload their public keys to the blockchain. and requester The public and private keys are denoted as .

3. The method for multi-task data aggregation based on blockchain-assisted crowd intelligence perception and privacy protection according to claim 1 is characterized in that: In step 2), the task publishing process is as follows: Assume that all requesters A task request is sent to the perception platform, with DID as The task request of the requester is recorded as ,in They are task ID, task details, aggregation type (such as sum, variance), and expiration time. After the requester's task, all tasks Broadcast to a collection of workers .

4. The method for multi-task data aggregation based on blockchain-assisted crowd intelligence perception and privacy protection according to claim 1 is characterized in that: In step 3), the detailed process of mask calculation is as follows: a) Worker and Each holds a unique elliptic curve key pair First, the worker and the requester query the public keys of all entities of the other party through the smart contract. Perform key negotiation with all requesters in R to obtain a list of shared keys .in . Similar to the worker operation, the requester Also calculated .in The public key transmission of the above key negotiation is completed through the blockchain smart contract code. b) Finally, all workers and requesters further calculate the shared mask with their respective shared key lists: .for and The resulting mask lists are . Similarly, , different lists of masks can be calculated in turn: The above processes are all interactive and can be performed offline, and can be executed asynchronously with subsequent steps.

5. The method for multi-task data aggregation based on blockchain-assisted crowd intelligence perception and privacy protection according to claim 1 is characterized in that: In step 4), the process of multi-task data submission is as follows: a) Workers From the task list published by the perception platform Select the appropriate task and generate a decision vector based on the selection result , where each element takes a Boolean value, 1 represents participation in the corresponding task, and 0 represents non-participation in the corresponding task. Here we assume Participated in all tasks, then The perception data of all tasks are collected and the perception data vector is obtained b) Workers According to the first two shared mask lists obtained in step 3), and To cover up, we get and . According to the Chinese remainder theorem and the public parameter, workers Multidimensional data with mask , Fusion into single-dimensional multi-task data c) To ensure data integrity, Generate a verification sequence for a message: .at last Will Send to the perception platform Upload to the blockchain. The rest of the workers do the same.

6. The method for multi-task data aggregation based on blockchain-assisted crowd intelligence perception and privacy protection according to claim 1 is characterized in that: In step 5), the process of multi-task data aggregation is as follows: a) Eliminate exited users: Due to user device or network reasons, users may exit during the crowd intelligence perception process, resulting in the failure to submit data within the scheduled time. After the deadline is reached, the perception platform counts the users who have not sent messages and excludes them from the user set, and the remaining users are set as the set b) Data aggregation and verification: Assume that the perception platform receives user data from workers , sum up to get multi-task aggregate data: . Afterwards To verify, if the equation If the equation is not true, then the result is correct. Verify one by one: , remove all invalid multi-task data and then aggregate them. c) Data extraction: The perception platform extracts Extract the aggregation results corresponding to each requester in turn: Finally, the perception platform will Sent to the corresponding requester .

7. The method for multi-task data aggregation based on blockchain-assisted crowd intelligence perception and privacy protection according to claim 1 is characterized in that: In step 6), the detailed process of aggregation result verification and masking is as follows: a) Data verification: Requester Received Afterwards, according to Confirm that the user has not exited and call the smart contract to query the worker verification sequence set stored in the blockchain Then for the masked task results Verify separately, if the equation If it is established, it means the result is correct and you can proceed to the next step. Otherwise, it means the integrity of the task result is destroyed. b) Data masking: Query the first two masks in your list to match All masks of are summed up separately to get . Unmask the task result to get the correct task result: .in is the result of data summation, is the number of workers involved. The remaining requesters operate with Same reason.

8. The method for multi-task data aggregation based on blockchain-assisted crowd intelligence perception and privacy protection according to claim 1 is characterized in that: In step 7), the process of implementing various aggregation types is as follows: a) Mean: by step 7) requester Easy to get the mean: b) Variance: The perception platform needs to recalculate a set of public prime numbers, ,calculate ,in . After that, each worker submits additional multi-task data of the squared value of the perception data: ,in . Aggregated and extracted by the perception platform, the requester Available: . Finally, remove the mask to get the variance value: c) Weighted summation and weighted average: The perception platform assigns a weight vector to each worker. The weight vector is , The last submitted data is: After aggregation and masking, the requester The weighted summation result can be obtained: . And the weighted average result can be further obtained: ,in

Images