A malicious-resistant fast secure aggregation method and apparatus

Abstract

The application discloses an anti-malicious fast security aggregation method and device, and relates to an anti-malicious fast security aggregation method, which comprises the following steps of S1, each of a plurality of participants obtains a plurality of random numbers, wherein the number of the participants is the same as the number of the random numbers obtained by each participant; S2, each participant uses the plurality of random numbers to generate a corresponding number of pseudo-random numbers for a current aggregation round, and calculates the generated pseudo-random numbers to obtain pseudo-random number calculation results; S3, each participant processes to-be-aggregated data according to the pseudo-random number calculation results, and all the participants send the processing results to a server for aggregation calculation; and S4, each participant receives the aggregation calculation results returned by the server, processes the aggregation calculation results, and obtains the aggregated results.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, and particularly relates to a malicious-resistant fast and safe aggregation method and device. BACKGROUND

[0002] In the era of data explosion, massive data are generated and transmitted through various heterogeneous devices and systems. These data are usually scattered and stored in different institutions, forming the so-called "data island" phenomenon. Especially in the fields of medical treatment, finance, government and supply chain involving sensitive information, the data island problem is particularly prominent. The traditional centralized machine learning method relies on a large amount of high-quality data to train the model, but a single institution often has difficulty in independently owning sufficient data quantity and diversified distribution characteristics to support the construction of high-quality model. To solve this problem, a feasible method is to aggregate the data sets from various institutions.

[0003] Under the federated learning (FL) framework, each participant (i.e. client) retains its local data and only uploads the model parameters to the central server; the latter is responsible for summarizing all local updates and generating a global model, which is then distributed to each client for subsequent iterative optimization. However, research shows that even if only the model parameters are transmitted, it is still possible to cause a certain degree of information leakage, such as the member inference attack which can infer whether a particular sample is used for training process. In order to further strengthen the privacy protection of federated learning, researchers have proposed various schemes to construct privacy protection mechanisms using differential privacy, homomorphic encryption, function encryption, secure aggregation and other technologies.

[0004] Existing research mainly focuses on the following aspects:

[0005] Secure multi-party computation (SMPC) is a key technology for protecting data privacy. By allowing multiple parties to collaborate on computation without a single trusted third party, the risk of data leakage is avoided. Google introduced SMPC into federated learning in 2017 and proposed a federated learning framework based on SMPC. However, this scheme still requires all clients to broadcast communications frequently, resulting in high communication overhead.

[0006] Differential privacy (DP) protects privacy by introducing noise into the model update, so that the output model cannot accurately reflect the model of any single user. However, the introduction of noise will lead to a decline in the performance of the aggregated model, and this method needs to be further optimized.

[0007] Homomorphic Encryption (HE) includes data by encrypting model gradients, preventing third-party servers from obtaining gradient data, thereby avoiding the server from cracking the original data information of the participants from the gradient. However, this scheme has a large computational overhead due to the introduction of homomorphic computation, and each participant needs to use the same public-private key pair for data encryption and decryption, which cannot resist the collusion between the server and the participants.

[0008] In summary, although the prior art has made progress in cryptographic security or communication efficiency, there are still the following main deficiencies:

[0009] 1. Lack of effective protection against collusion attacks: Most existing schemes do not fully consider how to prevent potential collusion between malicious servers and some clients.

[0010] 2. High communication overhead: For example, SMPC-based techniques require frequent point-to-point communication, increasing the overall system burden.

[0011] 3. Precision loss: Certain strategies such as differential privacy blur individual contributions by introducing noise, which can improve privacy levels but also sacrifices model performance.

[0012] To address the above technical problems, the present application provides a fast and secure anti-malicious aggregation method and device. SUMMARY

[0013] The purpose of the present application is to address the deficiencies of the prior art, providing a fast and secure anti-malicious aggregation method and device, which not only solves the privacy leakage risk existing in the existing federated learning framework, but also greatly improves the communication efficiency and maintains the high precision of model training, providing strong support for large-scale data collaboration across institutions.

[0014] To achieve the above purpose, the present application adopts the following technical solutions:

[0015] A fast and secure anti-malicious aggregation method, comprising:

[0016] S1. Each of the plurality of participants obtains a plurality of random numbers, wherein the number of participants is the same as the number of random numbers obtained by each participant;

[0017] S2. Each participant uses a plurality of random numbers to generate a corresponding number of pseudo-random numbers for the current aggregation round, and calculates the generated pseudo-random numbers to obtain a pseudo-random number calculation result;

[0018] S3. Each participant processes the data to be aggregated according to the pseudo-random number calculation result, and all participants send the processed results to the server for aggregation calculation;

[0019] S4. Each participant receives the aggregated calculation result returned by the server and processes the aggregated calculation result to obtain the aggregated result.

[0020] Further, the step S1 is specifically:

[0021] S11. Each participant generates a public key and a private key based on an asymmetric encryption algorithm, and broadcasts the public key to other participants;

[0022] S12. The first participant generates a plurality of random numbers, and performs splicing and encryption processing on the generated random numbers, and sends the processing result to other participants;

[0023] S13. Other participants generate a plurality of random numbers, and perform encryption processing on the generated random numbers, and send the processing result to other participants;

[0024] S14. Other participants decrypt the encryption results of the first participant and other participants to obtain the final plurality of random numbers.

[0025] Further, the step S3 in which each participant processes the to-be-aggregated data according to the pseudo-random number calculation result comprises:

[0026] The first participant processes the to-be-aggregated data, denoted as:

[0027] E(D1) = D1 + f1(k) + F(r 11 ,k)

[0028] Wherein, E(D1) represents the result of the first participant processing the to-be-aggregated data; D1 represents the to-be-aggregated data of the first participant; f1(k) represents the pseudo-random number calculation result of the first participant; F(r 11 ,k) represents the pseudo-random number of the first participant.

[0029] Further, the step S3 in which each participant processes the to-be-aggregated data according to the pseudo-random number calculation result comprises:

[0030] Other participants perform encryption processing on the to-be-aggregated data, denoted as:

[0031] E(D i ) = D i +f i (k)

[0032] Wherein, E(D i ) represents the result of the other participants processing the to-be-aggregated data; D i represents the to-be-aggregated data of the other participants; f i(k) represents the calculation result of the pseudo-random number of other participants; i∈2, 3, …, N; N represents the number of participants.

[0033] Further, the server performs the aggregation calculation in the step S3, which is represented as:

[0034] E(D) = E(D1) + E(D2) + … + E(DN) N )

[0035] Wherein, E(D) represents the calculation result of the server; E(D2) represents the result of the second participant processing the data to be aggregated; E(DN) represents the result of the Nth participant processing the data to be aggregated. N

[0036] Further, the server performs the aggregation calculation in the step S3, which is represented as:

[0037] D = E(D) - F(r 11 ,k)

[0038] Wherein, D represents the result of each participant processing the aggregation calculation result.

[0039] Correspondingly, an anti-malicious fast and secure aggregation device is also provided for executing the anti-malicious fast and secure aggregation method, and the aggregation device comprises a processor, and the processor is provided with:

[0040] A generation module for generating N random numbers for each participant in a plurality of participants, wherein the N random numbers generated by each participant include self-generated and / or other participant-generated;

[0041] A generation module for each participant to generate N pseudo-random numbers for the current aggregation round using the N random numbers, and to calculate f i (k) according to the N pseudo-random numbers;

[0042] An aggregation module for each participant to perform encryption processing on the data to be aggregated according to f i (k), and all participants send the encryption processing result to the server for aggregation calculation;

[0043] A decryption module for each participant to receive the aggregation calculation result returned by the server and to perform decryption processing on the aggregation calculation result to obtain the aggregated result.

[0044] Further, the processor is connected with a memory

[0045] Further, the processor is connected with a transmission device for data transmission with external devices.

[0046] ​Further, the processor is connected with an input and output device for inputting or outputting information.

[0047] Compared with the prior art, the present application has the following beneficial effects:

[0048] 1. Anti-collusion attack: Through the carefully designed random number blind mechanism, the data information of any participant cannot be obtained unless all other participants (i.e. N-1 participants) collude with the server. This design greatly improves the security of the system and effectively prevents malicious behavior from inside or outside.

[0049] 2. High efficiency and low communication overhead: The gradient is encrypted by using the random number blind technology, which avoids the frequent interaction process in the traditional method, thereby significantly reducing the communication cost. In addition, this method has higher computing performance and can realize faster data processing speed under the premise of maintaining security.

[0050] 3. No precision loss: Unlike the method of introducing noise to protect privacy, this scheme does not cause the precision of model training to decrease in the calculation process. The accuracy of the final aggregation result is completely consistent with that of direct aggregation using plaintext data, ensuring the construction of high-quality models. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 is a flow chart of an anti-malicious fast and secure aggregation method provided by embodiment one;

[0052] Figure 2 is a structure diagram of fast and secure data aggregation provided by embodiment one. DETAILED DESCRIPTION

[0053] The embodiments of the present application will be described in detail below with specific specific examples. Those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in the specification. The present application can also be implemented or applied by different specific embodiments, and the details in the specification can be modified or changed based on different views and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0054] The purpose of the present application is to overcome the defects of the prior art, and provide a data anonymization transmission method, system and device based on data hosting.

[0055] Embodiment one

[0056] The present embodiment provides an anti-malicious fast and secure aggregation method, as shown in Figures 1-2 , which comprises:

[0057] S1. Each of the plurality of participants obtains a plurality of random numbers, wherein the number of participants is the same as the number of random numbers obtained by each participant;

[0058] S2. Each participant uses the plurality of random numbers to generate a corresponding number of pseudo-random numbers for the current aggregation round, and calculates the generated pseudo-random numbers to obtain a pseudo-random number calculation result;

[0059] S3. Each participant processes the data to be aggregated according to the pseudo-random number calculation result, and all participants send the processed results to the server for aggregation calculation;

[0060] S4. Each participant receives the aggregation calculation result returned by the server and processes the aggregation calculation result to obtain the aggregated result.

[0061] In this embodiment, N participants (P1, P2, … P N ), a third-party server S.

[0062] Each participant has a piece of data to be aggregated D i , and each participant obtains the aggregated result D = D1 + D2 + … + DN. N .

[0063] In step S1, each of the plurality of participants obtains a plurality of random numbers, wherein the number of participants is the same as the number of random numbers obtained by each participant, specifically:

[0064] S11. Each participant generates a public key and a private key based on an asymmetric encryption algorithm, and broadcasts the public key to other participants;

[0065] Each of the N participants will generate a pair of public and private keys (sk i , pk i ) based on an asymmetric encryption algorithm, and each participant broadcasts the public key so that other participants obtain the public key pk i , wherein pk i represents the public key, sk i represents the private key; i represents the ith participant, i∈1, 2, …, N; N represents the number of participants.

[0066] S12. The first participant generates a plurality of random numbers and performs splicing and encryption processing on the generated random numbers, and sends the processing result to other participants;

[0067] The participant P1 generates M random numbers r 11 ,r 12 ,..., r 1M , and sends r 11 and r 1xSplice into a number R x = r 11 || r 1x , using pk i Encrypt R x , get the encryption result E(R x ), send the encryption result E(R x ) to the i-th participant. Splicing is to splice two strings into one, such as 123 and 456, into 123||456.

[0068] Where P1 represents the first participant; M represents the number of random numbers, and the value of M is the same as the value of N; x represents the x-th random number, x∈2, …, M; R x represents the spliced value; E(R x ) represents the encryption result. In this embodiment, each participant knows in advance which participant the random number of itself is to be sent to, so when encrypting the spliced value using the public key, the public key of the participant after sending is used, for example: P1 wants to send the generated random number to P2, P3, and P4, so when encrypting the spliced value in P1, the public keys of P2, P3, and P4 are used, and then P2, P3, and P4 can use their own private keys to decrypt in the subsequent decryption. It should be noted that the random number is sent backward, that is, P2, P3, and P4 can obtain the random number sent by P1, but P2 cannot obtain the random number sent by P3, and P3 cannot obtain the random number sent by P4.

[0069] S13. Other participants generate multiple random numbers and encrypt the generated random numbers, and send the processing result to other participants;

[0070] The participant P i generates M-i(M-k) random numbers r iy (i i y), and encrypts r iy using the public key pk iy , gets the encryption result E(r iy ), and sends the encryption result E(r j ) to other participants P .

[0071] In step S13, i∈2, 3, …, N; y represents the y-th random number, y∈i+1, …, M.

[0072] In this embodiment, each participant knows in advance which participant the random number of itself is to be sent to, and therefore when the random number is encrypted using a public key, the public key of the participant to which the random number is to be sent is used, for example, P2 is to send the generated random number to P3 and P4, and therefore when the random number in P2 is encrypted, the public keys of P3 and P4 are used, and then P3 and P4 can use their own private keys to decrypt in the subsequent decryption. It should be noted that the random number is sent backward, that is, P3 and P4 can obtain the random number sent by P2, but P3 cannot obtain the random number sent by P4.

[0073] S14. Other participants decrypt the encrypted results of the first participant and other participants to obtain the final random numbers.

[0074] Participant P j uses its own private key sk i to decrypt the E(R x ) sent by participant P1 in step S12 to obtain r 11 ||r 1x , and then splits r 11 ||r 1x to obtain r i and r iy . Then, the E(r iy ) from other participants P j in step S13 is decrypted to obtain r i (i<y).

[0075] In step S14, the value of j in participant P x is the same as the value of i in the private key sk 1x , the value of x in E(R iy ), and the value of y in E(r iy ). The value of j in P j is greater than the value of i in P i .

[0076] As shown in Table 1 below, an example is given to illustrate the random numbers obtained by four participants after steps S11-S14:

[0077] Table 1

[0078] Participants Random number [P1] r 11 , r 12 , r 13 , r 14 ]]> [P2] r 11 , r 12 , r 23 , r 24 ]]> [P3] r 11 , r 13 , r 23 , r 34 ]]> [P4] r 11 , r 14 , r 24 , r 34 ]]>

[0079] In Table 1, r 11 , r 12 , r 13 , and r 14 in participant P1 are generated by P1 itself, r 23 and r 24 in participant P2 are generated by P2 itself; r 34P3 is generated by P3 itself.

[0080] In Table 1, r 11 in P2 is obtained from P1, r 12 in P3 is obtained from P1, r 11 in P3 is obtained from P2, r 13 in P4 is obtained from P1, r 23 in P4 is obtained from P2, and r 11 in P4 is obtained from P3. 14 24 34

[0081] After the method according to steps S11-S14 is executed, each participant P i ∈(P1,...,P N ) has M random numbers.

[0082] In step S2, each participant uses the random numbers to generate a corresponding number of pseudo-random numbers for the current aggregation round, and calculates the generated pseudo-random numbers to obtain a pseudo-random number calculation result.

[0083] The participant P i ∈(P1,...,P N ) uses the M random numbers to perform a zero secret sharing on the current aggregation round k to generate M pseudo-random numbers F(r iy , k), where y ∈ i+1,..., M, i ∈ 1, 2, 3,..., N; and calculates the generated pseudo-random numbers to obtain a pseudo-random number calculation result f i (k), which is represented as f i (k) = [-F(r 1i , k)] +... + [F(r iM , k)], where the positive and negative signs before F depend on whether r is generated by itself or obtained from other participants. If it is generated by itself, it is positive, and if it is obtained from other participants, it is negative.

[0084] As shown in Table 2 below, an example of pseudo-random numbers for four participants is shown:

[0085] Table 2

[0086] Participants f i (k)]]> [P1] F(r 12 ,k)+F(r 13 ,k)+F(r 14 ,k)]]> [P2] -F(r 12 ,k)+F(r 23 ,k)+F(r 24 ,k) <!-- 5 -->]]> [P3] -F(r 13 ,k)-F(r 23 ,k)+F(r 34 ,k)]]> [P4] -F(r 14 ,k)-F(r 24 ,k)-F(r 34 ,k)]]>

[0087] In step S3, each participant processes the data to be aggregated according to the pseudo-random number calculation result, and all participants send the processed results to the server for aggregation calculation.

[0088] ​​​The first participant P1 performs blinding processing on the aggregated data D1, represented as E(D1) = D1 + f1(k) + F(r 11 ,k).

[0089] The other participants P i ∈(P2,...,P N ) perform processing on the aggregated data D i , represented as E(D i ) = D i +f i (k).

[0090] All participants send the processing results to the third-party server S for aggregation calculation.

[0091] The third-party server S performs aggregation calculation, represented as E(D) = E(D1) + E(D2) +... + E(D N ), and returns the result to the other participants.

[0092] In step S4, each participant receives the aggregation calculation result returned by the server and processes the aggregation calculation result to obtain the aggregated result.

[0093] The participants P i ∈(P1,...,P N ) perform de-blinding processing on the aggregated result E(D), represented as D = E(D) - F(r 11 ,k).

[0094] Compared with the prior art, the present application has the following beneficial effects:

[0095] 1. Anti-collusion attack: Through the carefully designed random number blinding mechanism, the data information of any participant cannot be obtained unless all other participants (i.e. N-1 participants) collude with the server. This design greatly improves the security of the system and effectively prevents malicious behavior from inside or outside.

[0096] 2. High efficiency and low communication overhead: The gradient is encrypted using the random number blinding technology, avoiding the frequent interaction process in traditional methods, thereby significantly reducing the communication cost. In addition, this method has higher computing performance and can achieve faster data processing speed while maintaining security.

[0097] 3. No precision loss: Unlike methods that introduce noise to protect privacy, this scheme does not cause the precision of model training to decrease during the calculation process. The accuracy of the final aggregation result is exactly the same as that when using plaintext data for aggregation, ensuring the construction of high-quality models.

[0098] Embodiment Two

[0099] The embodiment provides an anti-malicious fast security aggregation device for executing an anti-malicious fast security aggregation method of the embodiment one, and the aggregation device comprises a processor, and the processor is internally provided with:

[0100] A generating module is configured to generate N random numbers for each of a plurality of participants, wherein the N random numbers generated by each participant include self-generated and / or other participant-generated random numbers.

[0101] A generating module is configured to generate N random numbers for each of a plurality of participants, wherein the N random numbers generated by each participant include self-generated and / or other participant-generated random numbers. i (k) for each participant using the N random numbers, and calculate f

[0102] An aggregating module is configured to perform encryption processing on the data to be aggregated according to f i (k) for each participant, and all participants send the encryption processing results to a server for aggregation calculation.

[0103] A decrypting module is configured to receive the aggregation calculation result returned by the server by each participant, and perform decryption processing on the aggregation calculation result to obtain the aggregated result.

[0104] The processor can comprise a central processing unit (CPU), or a specific integrated circuit (Application Specific Integrated Circuit, referred to as ASIC), or can be configured to implement one or more integrated circuits of the embodiment of the application.

[0105] The processor of the present embodiment is connected with a memory for storing data or instructions. The memory can include a mass storage for data or instructions. By way of example and not limitation, the memory can include a hard disk drive (HDD), a floppy disk drive, a solid state drive (SSD), a flash drive, a compact disc (CD), a magneto-optical disc, magnetic tape, or a Universal Serial Bus (USB) drive or a combination of two or more of these. The memory can include removable or non-removable (or fixed) media, where appropriate. The memory can be internal or external to the data processing device, where appropriate. In particular embodiments, the memory is a nonvolatile memory. In particular embodiments, the memory includes a Read-Only Memory (ROM) and a Random Access Memory (RAM). Where appropriate, this ROM can be mask- programmed ROM, Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Electrically Alterable Read-Only Memory (EAROM), or flash memory (FLASH), or a combination of two or more of these. Where appropriate, this RAM can be Static Random-Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), which can be Fast Page Mode Dynamic Random Access Memory (FPMDRAM), Extended Data Output Dynamic Random Access Memory (EDODRAM), Extended Data Output Dynamic Dynamic Random Access Memory (EDODRAM), or a combination of two or more of these.

[0106] The memory can be used to store or cache various data files that need to be processed and / or communicated, and possible computer program instructions executed by the processor. The processor reads and executes the computer program instructions stored in the memory.

[0107] The processor is connected with a transmission device for data transmission with external devices. The transmission device can be used to receive or send data via a network. The above-mentioned network examples can include wired or wireless networks provided by the communication provider of the electronic device. In one example, the transmission device includes a network adapter (NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device can be a radio frequency (RF) module, which is used to communicate with the Internet in a wireless manner.

[0108] The processor is connected with an input / output device for inputting or outputting information. The input / output device is used for inputting or outputting information. In the embodiment, the input information is the data to be aggregated by each participant, and the output information is the aggregated result.

[0109] It should be noted that the above are only preferred embodiments of the present application and the technical principles applied. Those skilled in the art will understand that the present application is not limited to the specific embodiments described herein, and those skilled in the art can make various obvious changes, readjustments and substitutions without departing from the scope of the present application. Therefore, although the present application has been described in detail through the above embodiments, the present application is not limited to the above embodiments, and can include more other equivalent embodiments without departing from the concept of the present application, and the scope of the present application is determined by the scope of the appended claims.

Claims

1. A fast secure aggregation method against malice, characterized in that, The method comprises the following steps: S1. Each of the plurality of participants obtains a plurality of random numbers, wherein the number of participants is the same as the number of random numbers obtained by each participant; S2. Each participant uses the plurality of random numbers to generate a corresponding number of pseudo-random numbers for the current aggregation round, and calculates the generated pseudo-random numbers to obtain a pseudo-random number calculation result; S3. Each participant processes the data to be aggregated according to the pseudo-random number calculation result, and all participants send the processed results to the server for aggregation calculation; S4. Each participant receives the aggregation calculation result returned by the server and processes the aggregation calculation result to obtain the aggregated result. The step S1 is specifically: S11. Each participant generates a public key and a private key based on an asymmetric encryption algorithm, and broadcasts the public key to other participants; S12. The first participant generates a plurality of random numbers, and performs splicing and encryption processing on the generated random numbers, and sends the processing result to other participants; S13. Other participants generate a plurality of random numbers, and perform encryption processing on the generated random numbers, and send the processing result to other participants; S14. Other participants decrypt the encryption results of the first participant and other participants to obtain the final plurality of random numbers; In the step S3, each participant processes the data to be aggregated according to the pseudo-random number calculation result, which comprises: The first participant processes the data to be aggregated, which is represented as: E(D1) = D1 + f1(k) + F(r 11 ,k) Wherein, E(D1) represents the result of the first participant processing the data to be aggregated; D1 represents the data to be aggregated of the first participant; f1(k) represents the calculation result of the pseudo-random number of the first participant; F(r 11 ,k) represents the pseudo-random number of the first participant; r 11 represents the random number generated by the first participant; k represents the aggregation round; In the step S3, each participant processes the data to be aggregated according to the pseudo-random number calculation result, which comprises: Other participants perform encryption processing on the data to be aggregated, which is represented as: E(D i )=D i +f i (k) wherein E(D i ) represents the result of processing the aggregated data by the other participants; D i represents the data to be aggregated of the other participants; f i (k) represents the result of pseudo-random number calculation of the other participants; i∈2, 3, …, N; N represents the number of participants. In the step S4, the aggregation calculation result is processed, which is represented as: D = E(D) - F(r 11 , k) Wherein, D represents the result of processing the aggregation calculation result by each participant.

2. The method of claim 1, wherein the method is further characterized by, In the step S3, the server performs aggregation calculation, which is represented as: E(D) = E(D1) + E(D2) +... + E(D N ) Wherein, E(D) represents the result of the server performing aggregation calculation; E(D2) represents the result of the second participant processing the data to be aggregated; E(D N ) represents the result of the Nth participant processing the data to be aggregated.

3. An anti-malicious fast secure aggregation apparatus for performing the anti-malicious fast secure aggregation method of any one of claims 1-2, the aggregation apparatus comprising a processor, wherein, The processor is provided with: A generation module for generating N random numbers for each of the plurality of participants, wherein the N random numbers generated by each participant include self-generated and / or other participant-generated; a generating module, configured to generate N pseudo-random numbers for each participant using N random numbers in a current aggregation round, and calculate f i (k); A aggregation module is used for each participant to aggregate the data according to f i (k) The data to be aggregated is encrypted, and all participants send the encrypted results to the server for aggregation calculation. A decryption module for each participant to receive the aggregation calculation result returned by the server and perform decryption processing on the aggregation calculation result to obtain the aggregated result.

4. The anti-malicious fast security aggregation device of claim 3, wherein, The processor is connected with a memory for storing data or instructions.

5. The anti-malicious fast secure aggregation device of claim 3, wherein, The processor is connected with a transmission device for data transmission with external devices.

6. The anti-malicious fast secure aggregation device of claim 3, wherein, The processor is connected with an input and output device for inputting or outputting information.

Images