Crowd sensing truth value discovery method for privacy protection based on reputation value

Abstract

The invention discloses a crowd sensing truth value discovery method based on privacy protection of a reputation value, and aims to improve the accuracy and robustness of a result by combining the reputation value in data aggregation calculation under the condition of protecting the identity information privacy of workers and the privacy of data contents. The system network is composed of a credible authentication center, a data requester, a service platform and a worker. A worker registers a pseudonym in a credible authentication center in advance and initializes a reputation value. And when participating in the task, the service platform inquires whether the reputation value of the worker meets the task requirement from the credible authentication center, and selects the worker to participate in the task when the reputation value meets the condition. After a worker submits collected data, on the basis of a disturbance factor and an Elgamal encryption algorithm, on the premise that data privacy of the worker is protected, a more accurate truth value is obtained in combination with a reputation value, and the reputation value of the worker is dynamically updated according to the quality condition of the submitted data, so that the worker is guided to select and calculate the truth value to improve the data quality.

Description

Technical Field

[0001] The present invention belongs to the field of data collection for trusted computing and quality assurance in a crowd-sensing network, and particularly relates to a truth discovery method based on workers' reputation values while protecting the identity privacy and data privacy of workers in a crowd-sensing network. Background Art

[0002] Mobile crowd sensing (MCS) is a new data collection method that ingeniously combines the computing and sensing capabilities of mobile devices with a wide user group to achieve efficient and flexible information collection. In urban traffic planning, MCS technology can integrate real-time data from numerous vehicles and smartphones to help traffic management departments quickly identify congested sections, optimize signal control, and improve urban travel efficiency. In addition, MCS also plays an important role in many fields such as medical health and agricultural disaster warning, with the advantages of low cost and high coverage. However, the development of this technology faces two important problems: one is the imperfect privacy protection mechanism, and the other is the difficulty in guaranteeing data quality.

[0003] In a typical mobile crowd sensing (MCS) task process, data requesters (DRs) publish task requirements through a service platform (SP). Crowdsourcing workers can browse the task list and apply to participate. After screening, the platform assigns tasks to suitable workers. After workers complete data collection, they submit the data to the platform. A trust authority (TA) is responsible for pseudonym allocation and privacy protection in the process. In traditional mobile crowd sensing, the platform usually uses the CRH algorithm to integrate multi-source data and output results. The CRH algorithm is a method based on the weighted average. The first-round aggregation result can take the average value of the data of k workers. Then, the distance between the data submitted by each worker and the previous-round aggregation result is calculated respectively, and the sum of the distances from all data to the initial aggregation result is obtained. Next, the weight of the i-th worker is defined as the logarithm of the ratio of the distance from the data submitted by this worker to the previous-round aggregation result to the sum of the distances from all data to the previous-round aggregation result. In this way, the aggregation result of this round can be calculated. The algorithm will perform iterative calculations until the difference between the aggregation results of adjacent rounds is less than a specific threshold. At this time, the aggregation result of the last round is used as the final data aggregation result.

[0004] However, this solution has significant drawbacks. First, the solution fails to protect the privacy of workers' identities and data content. The data collected by workers is vulnerable to abuse by the platform or third parties, and there is a risk of worker privacy leakage. Therefore, workers may be reluctant to participate in crowd sensing tasks, hindering its development. Second, traditional data aggregation methods such as CRH are not robust enough in the face of worker collusion attacks. At most, more than half of the malicious workers can easily manipulate the results by submitting similar data, defrauding task rewards with low-cost false data. Third, the current worker evaluation mechanism based on reputation values is not yet mature. How to scientifically evaluate the reputation value of workers based on their task completion is still a problem that needs to be solved. These problems all to some extent restrict the efficient and sustainable development of crowd sensing tasks.

[0005] Existing technologies have deficiencies in simultaneously meeting high data quality, worker identity, and reputation privacy protection. Therefore, there is an urgent need for an inventive method that can comprehensively solve the above problems. The present invention proposes an effective solution to the above problems. Summary of the Invention

[0006] The present invention discloses a truth discovery method for privacy protection based on reputation values in crowd sensing, aiming to improve the accuracy and robustness of the results by combining reputation values in data aggregation calculations while protecting the privacy of workers' identity information and data content. The entire system architecture includes a data requester (DR), a service platform (SP), a trusted authentication center (TA), and workers. Unless otherwise specified, the following entities are all represented by their corresponding symbols.

[0007] The present invention provides a truth discovery method for privacy protection based on reputation values, which is characterized by including the following parts:

[0008] 1. A truth discovery method for privacy protection based on reputation values, which is characterized by including the following steps:

[0009] Step 1: The trusted authentication center TA performs system initialization, selects a prime number Construct a cyclic group of order Generate a public-private key pair (pk , sk s ), the public key is s The private key sk is randomly selected from s ; TA assigns a unique pseudonym a to each worker, and generates it through the formula j , where n is the true identity of the worker, salt is a random value in j , stores it in TA, represents the exclusive OR operation, represents the exclusive OR operation, is a hash function; initially, the historical task success times α and failure times β of worker j when the cumulative task count is t are both 1, where j represents the j-th worker and t represents the t-th data collection. The estimated reputation value of worker j at the t-th data collection is j,y and j,t for worker j at the t-th data collection is Data requester: DR generates a public-private key pair (pk r , sk r ) and registers with TA.

[0010] Step 2: DR generates a sensing task requirement, and publishes the task budget and the required reputation value C for the workers in task i i,t to potential workers via the service platform: SP, where i represents the i-th task and t represents the t-th data collection during the task.

[0011] Step 3: After the worker receives the task and decides to participate, the worker sends a task application to SP using the pseudonym a j .

[0012] Step 4: After receiving the worker's application, SP asks TA whether the worker's reputation value meets the standard, that is, whether it is greater than or equal to the reputation threshold requirement C i,t . TA sorts the workers who meet the reputation value condition and returns the results according to the reputation value. Among the qualified workers, SP randomly selects 0.8×k workers before the seventy percentile and 0.2×k workers after the seventy percentile, a total of k workers to participate in the task, and sends the recruitment result to the workers If is 1, then the worker is selected, otherwise not. SP sends the randomly generated shared positive integer perturbation factor pair and and the public key pk of DR r to the worker j selected for task i at the t-th data collection.

[0013] Step 5: The selected worker collects data and encrypts it with the shared perturbation factor pair and according to the formula to obtain and , and then encrypts r and and using the Elgamal encryption algorithm with the public key pk of DR and transmits them to SP. The encryption process is

[0014] Step 6: After receiving the ciphertext, the SP transmits it to the DR for the data aggregation process based on the reputation value. Finally, the DR obtains the final result. The specific process is described in Part 2.

[0015] Step 7: The SP distributes rewards to the workers according to the weights of the workers in the last round.

[0016] Step 8: The SP compares the distance with the parameter γ, where is the data collected by the worker, is the task result calculated in the last round of iteration, and the parameter γ is the threshold for determining the validity of the collected data. The reputation value of the worker is updated. If the distance is less than or equal to γ, then (a j , 1) is sent to the TA, otherwise (a j , 0) is sent. The TA updates the worker's reputation accordingly. If (a j , 1) is received, then α j,t+1 = α j,t + 1, otherwise β j,t+1 = β j,t + 1.

[0017] 2. A truth discovery method for privacy protection based on reputation value according to claim 1, characterized in that: during the data aggregation process, by combining the perturbation factor, while protecting the privacy of the workers' data, the accuracy of the result and the robustness of the system against collusion attacks are improved by combining the reputation values of the workers.

[0018] Step 1: After receiving the ciphertext data sent by the SP, the DR decrypts it using the private key sk r to obtain and

[0019] Step 2: The DR calculates by subtracting the result of the previous round of iteration. After that, the DR sends the calculated to the SP, where represents the aggregation result of the nth round, and the initial value can be assigned as the average value of all workers' data.

[0020] Step 3: The SP calculates the distance between each worker's data and the current aggregation result using the perturbation factor shared with the workers. The formula is

[0021] Step 4: The SP calculates the weight of each worker according to the formula and then sends to the TA.

[0022] Step 5: The TA uses the reputation value C of the worker​j,t and the data sent by the SP is calculated according to the formula and to calculate the adjusted weight and the sum of weights and send them to the DR.

[0023] Step6: After the DR receives the and sent by the TA, multiply the data by to obtain the weighted data set and normalize the data through the formula to aggregate the result of the current (n + 1)-th iteration.

[0024] Step7: Repeat the iteration until the difference between the results of two adjacent rounds is less than the preset threshold to obtain the final result.

[0025] Beneficial effects

[0026] The present invention proposes a privacy protection truth discovery method for crowdsensing based on reputation values. The experimental results of a large number of real data sets show that this method performs excellently in terms of data quality. Compared with the traditional CRH truth discovery method, the data quality is improved by 33.3%. When dealing with collusion attacks, the data quality can be improved by up to 90.5%. In addition, this system performs well in worker reputation recognition and can effectively evaluate the reputation of workers. In summary, the present invention comprehensively protects the worker identity information and data privacy, significantly improves the quality of aggregated data, enhances the ability of the system to resist collusion attacks, provides an efficient and reliable solution for the wide application of the crowdsensing network, and at the same time has certain innovation and practical application value, and has considerable promotion potential. Description of the drawings

[0027] The drawings forming a part of this application are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention.

[0028] Figure 1 is the system network composition diagram of the inventive method;

[0029] Figure 2 is the flowchart of the data aggregation process;

[0030] Figure 3 is the performance comparison diagram of this method and the traditional aggregation method; Detailed implementation manners

[0031] To facilitate the understanding of the present invention, the present invention will be described more comprehensively and meticulously below in conjunction with the accompanying drawings of the specification and preferred embodiments. However, the protection scope of the present invention is not limited to the following specific embodiments.

[0032] Unless otherwise defined, all the technical terms used hereinafter have the same meaning as commonly understood by those skilled in the art. The technical terms used herein are only for the purpose of describing specific embodiments and are not intended to limit the protection scope of the present invention.

[0033] Unless otherwise specifically stated, various raw materials, reagents, instruments, equipment, etc. used in the present invention can be obtained through market purchases or can be prepared by existing methods.

[0034] Embodiment:

[0035] In the crowd-sensing network, there is a situation where malicious workers collude to submit similar data to defraud rewards. When a subject needs to obtain higher real data such as temperature and humidity, air quality, etc. at a specific location and issues a task to the platform, and the platform recruits workers to collect data and calculate the results, the platform can use the method of the present invention to aggregate reliable results using the data sensed by the workers while protecting the privacy of the workers' identities and data, and has good robustness against malicious worker collusion attacks.

[0036] The experimental results of the inventive method are given below.

[0037] Figure 1 The system network composition diagram and working steps of the method of the present invention are given. The method of the present invention includes the following entities: a trusted authentication center, which is trusted; a service platform, the server is curious about the privacy of workers and data, but will strictly execute according to the data collection process; a data requester, the initiator of the task; and numerous workers, who can perform sensing tasks. The figure describes the working process of the system, and the specific content refers to the patent claims.

[0038] Figure 2 The flowchart of the second part of the patent claims is given. The data requester, the service platform, and the trusted authentication center can calculate a reliable true value under the condition that the data content of the workers is completely confidential according to the steps, preventing the abuse of workers' data.

[0039] Figure 3 The experimental results of the method of the present invention and other truth discovery methods in terms of data accuracy are given. There are two methods compared with the method of the present invention: one is the CRH method, and the other is the average value method (calculating the average value of the data submitted by all workers). From the experimental results Figure 3It can be seen that the method of the present invention has significant advantages compared with the other two methods. Compared with the traditional CRH method, the root mean square error is increased by 33.3%, and the data quality is also significantly improved. When facing collusive attacks, the method of the present invention has obvious superiority compared with the other two methods. As the proportion of malicious workers increases, the data quality of the CRH method drops severely. However, the method of the present invention has stronger robustness to it. In terms of data quality, the performance is increased by up to 90.5% compared with CRH.

Claims

1. A truth discovery method for privacy protection based on reputation value, characterized in that Including the following steps: Step 1: The Trusted Authentication Center TA performs system initialization, selects a prime number to construct a cyclic group of order and generates a public-private key pair (pk , sk s , sk s ). The public key is and the private key sk s is randomly selected from . TA assigns a unique pseudonym a j to each worker, which is generated through the formula , where is the real identity of the worker, salt is a random value from , stored in TA, represents the exclusive OR operation, is a hash function; initially, for worker j at the cumulative task count of t, the historical task success count α j,t and failure count β j,t are both 1, where j represents the j-th worker and t represents the t-th data collection. The estimated reputation value of worker j at the t-th data collection is Data requester: DR generates a public-private key pair (pk r , sk r ) and registers with TA. Step 2: The DR generates a sensing task requirement and sends the task budget and the required credit value C for the worker in task i i,t to the potential workers through the service platform: SP, where i represents the i-th task and t represents the t-th data collection during the task execution. Step 3: After the worker receives the task and decides to participate, use the pseudonym a j Send a task application to the SP. Step 4: After receiving the worker's application, SP asks TA whether the credit value of this worker meets the standard (i.e., whether it is greater than or equal to the required credit threshold C i,t ). TA sorts the workers who meet the credit value condition and returns the results according to the credit value from high to low. Among the qualified workers, SP selects k workers to participate in the task according to the stratification (randomly selects 0.8×k workers before the seventy percentile and 0.2×k workers after the seventy percentile), and sends the recruitment result to the workers If is 1, then it is selected; otherwise, it is not selected. SP sends the randomly generated shared positive integer perturbation factor pair and and the public key pk of DR r . Step 5: The selected worker collects data and combines it with the shared perturbation factor pair and According to the formula to obtain the encrypted and and then uses the DR public key pk r to encrypt using the Elgamal encryption algorithm and and transmits them to the SP. The encryption process is Step 6: After the SP receives the ciphertext, it is transmitted to the DR for the data aggregation process based on the reputation value. Finally, the DR obtains the final result. The specific process is shown in Part 2. Step 7: The SP distributes rewards to the workers according to the weights of the workers in the last round Step 8. SP comparison distance and parameter γ, where is the data collected by the worker, is the task result calculated in the last round of iteration. The threshold γ for determining the validity of the collected data is set to update the worker's reputation value. If the distance is less than or equal to γ, then send (a j , 1) to TA, otherwise send (a j , 0). TA updates the worker's reputation accordingly. If it receives (a j , 1), then α j,t+1 = α j,t + 1, otherwise β j,t+1 = β j,t + 1.

2. The truth discovery method for privacy protection based on reputation value according to claim 1, characterized in that: During the data aggregation process, by combining the perturbation factor, while protecting the privacy of workers' data, the accuracy of the result and the robustness of the system against collusion attacks are improved by combining the reputation values of workers. Step1: After the DR receives the ciphertext data sent by the SP, it uses the private key sk r to decrypt it and obtain and Step 2: DR calculates by subtracting the result of the previous iteration After that, DR will send the calculated to SP, where represents the aggregation result of the nth round, and the initial value can be assigned as the average of all worker data. Step 3: The SP uses the perturbation factor shared with the workers to calculate the distance between each worker's data and the current aggregation result. The formula is Step4: SP calculates the weight of each worker according to the formula and then sends it to TA. ​ Step5: TA utilizes the worker's credibility value C j,t and the data sent by SP according to the formulas and to calculate the adjusted weight and the sum of weights and send them to DR. Step6: After the DR receives the and , it multiplies the data by to obtain the weighted data set and normalizes the data through the formula to aggregate the result of the current (n + 1)-th iteration. Step 7: Repeat the iterative process until the difference between the results of two adjacent rounds is less than the preset threshold Obtain the final result.

Images