Search

KR-20260065367-A - METHOD AND APPARATUS FOR COMPUTING ENCRYPTED DATA USING MULTI-HOMOMORPHIC ENCRYPTION SYSTEM

KR20260065367AKR 20260065367 AKR20260065367 AKR 20260065367AKR-20260065367-A

Abstract

A video source software identification device according to one embodiment can perform the operation of acquiring a data set including a video file transmitted by a predetermined software; the operation of extracting a value of metadata including a container file format, internal data of the container file, and encoding parameters of each video file; and the operation of creating a classification model based on a deep learning algorithm that learns a parameter reflecting the correlation between the value of metadata of each video and the software to which each video file was transmitted.

Inventors

  • 백윤흥
  • 남기빈
  • 주유연

Assignees

  • 서울대학교산학협력단

Dates

Publication Date
20260508
Application Date
20241101

Claims (20)

  1. In a method performed by an encryption data operation device operated by a processor, An operation to obtain source code including operation rules for input data and first format data of an RLWE encryption scheme to be used as input data for said source code; An operation to classify a first code that performs arithmetic operations and a second code that performs comparison operations among the above source codes; The operation of performing homomorphic encryption operations according to the above source code, wherein when operating on a portion of the first code, arithmetic operations are performed using the system of the first format data, and when operating on a portion of the second code, the system of the first format data is converted into the second format data of the LWE encryption system and comparison operations are performed; and A process including an operation of performing calculations according to the above source code and returning the final output result value to the system of the above first format data. method.
  2. In paragraph 1, The above encryption data processing device It performs the role of a cloud server that processes requests from user terminals, and The above source code and the above first format data Obtained from the above user terminal, method.
  3. In paragraph 2, The above first format data Includes encrypted data of the RLWE system, which is input data to be processed through the above source code encrypted with the secret key of the user terminal, method.
  4. In paragraph 3, The above-mentioned returning operation is A method including the operation of transmitting the result of the operation of the first format data, which is finally output by performing calculations according to the above source code, to the user terminal. method.
  5. In paragraph 4, The above user terminal The operation result of the first format data, which is finally output by performing operations according to the above source code, is decrypted with the above secret key to obtain the actual operation result, method.
  6. In paragraph 1, The above-mentioned classification operation is The operation of classifying code among the above source codes that includes operations for addition, subtraction, multiplication, and division into a first code; and A process including the operation of classifying code among the above source codes that includes operations for greater than/less than comparison and equality comparison into a second code. method.
  7. In paragraph 6, The above-mentioned classification operation is The operation of calculating the total computation load based on the ratio of the amount of computation included in the first code and the second code among the source codes above. method.
  8. In Paragraph 7, The above-mentioned classification operation is The operation of setting the number of parameters to perform the operation of the first code and the number of parameters to perform the operation of the second code based on the total operation load above. method.
  9. In paragraph 8, The above-mentioned classification operation is Based on the above total computational load, the operation of setting the sum of the number of parameters to perform operations of the first code and the number of parameters to perform operations of the second code to the minimum number of parameters capable of handling the total computational load, method.
  10. In paragraph 8, The above-mentioned classification operation is The operation of setting the number of parameters to perform the operation of the first code and the number of parameters to perform the operation of the second code differently based on the total operation load above, method.
  11. In paragraph 8, The above-mentioned classification operation is Based on a target value of target operation time or target operation accuracy, the number of parameters to perform an operation of the first code and the number of parameters to perform an operation of the second code are pre-set, and the operation of setting the number of parameters to the pre-set value according to the target value of target operation time or target operation accuracy is included. method.
  12. In paragraph 1, The above-mentioned operation is When performing calculations on the portion of the second code above, the operation of converting first format data in a slot encoding state into second format data in a coefficient encoding state is included. method.
  13. In Paragraph 12, The operation of converting to the above-mentioned second format data is The operation of converting the first-form data into a polynomial form based on the SlottoCoeff technique; and A method comprising the operation of generating second-form data of the LWE encryption scheme by extracting coefficients of the above polynomial based on the LWE-extract technique. method.
  14. In Paragraph 13, The operation of converting the first-form data into a polynomial form based on the above SlottoCoeff technique is First type data in vector form From polynomial data based on the SlottoCoeff method including an operation to convert to, method.
  15. In Paragraph 13, The operation of generating second-form data of the LWE encryption system by extracting the coefficients of the polynomial based on the above LWE-extract technique is Data in the form of the above polynomial From the coefficient values of the polynomial color A process comprising extracting each of the above coefficient values and generating second-format data of an LWE encryption scheme in which the coefficient values are plaintext, method.
  16. In paragraph 1, The above-mentioned operation is When performing calculations on a portion of the first code above, the operation of converting the second format data into a polynomial form through the RLWE-pack technique; and A method comprising the operation of recovering the ciphertext modulus in the form of the polynomial through the Modrecover technique and converting it into the first format data of the RLWE encryption scheme. method.
  17. In Paragraph 16, The operation of converting the second type data into a polynomial form through the above RLWE-pack technique is coefficients of a polynomial Data in the form of a polynomial from the second-form data of the LWE encryption scheme using as plaintext including an operation to restore, method.
  18. In Paragraph 16, The operation of converting the above polynomial into the RLWE encryption scheme through the above Modrecover technique is polynomial data First format data in vector form from including an operation to restore, method.
  19. Memory containing instructions; and It includes a processor that performs a predetermined operation based on the above instructions, and The operation of the above processor is, An operation to obtain source code including operation rules for input data and first format data of an RLWE encryption scheme to be used as input data for said source code; An operation to classify a first code that performs arithmetic operations and a second code that performs comparison operations among the above source codes; The operation of performing homomorphic encryption operations according to the above source code, wherein when operating on a portion of the first code, arithmetic operations are performed using the system of the first format data, and when operating on a portion of the second code, the system of the first format data is converted into the second format data of the LWE encryption system and comparison operations are performed; and A process including an operation of performing calculations according to the above source code and returning the final output result value to the system of the above first format data. Encryption data processing unit.
  20. When executed on at least one processor, The above processor An operation to obtain source code including operation rules for input data and first format data of an RLWE encryption scheme to be used as input data for said source code; An operation to classify a first code that performs arithmetic operations and a second code that performs comparison operations among the above source codes; The operation of performing homomorphic encryption operations according to the above source code, wherein when operating on a portion of the first code, arithmetic operations are performed using the system of the first format data, and when operating on a portion of the second code, the system of the first format data is converted into the second format data of the LWE encryption system and comparison operations are performed; and A process including an operation of performing calculations according to the above source code and returning the final output result value to the system of the above first format data. A computer program stored on a computer-readable recording medium.

Description

Method and apparatus for computing encrypted data using a multi-homomomorphic encryption system The present invention relates to homomorphic encryption technology, and more particularly to a method and apparatus for performing encryption data operations that utilize a multi-type homomorphic encryption system to optimize performance while maintaining the accuracy of operations in an encrypted state. Homomorphic encryption is a technology that enables computations while remaining encrypted, protecting data privacy. It plays a crucial role in maximizing information security while allowing computations to be performed. Homomorphic encryption is broadly divided into two types: Ring Learning With Error (RLWE) based homomorphic encryption and Learning With Error (LWE) based homomorphic encryption. These two methods have their respective strengths and weaknesses and are utilized to suit different computational requirements. RLWE-based homomorphic encryption is specialized for arithmetic operations such as addition and multiplication, allowing for very fast processing. However, the RLWE method does not directly support non-arithmetic operations, and processing them by replacing them with approximation formulas results in a decrease in accuracy. In other words, while RLWE-based encryption offers excellent performance, it has the limitation of being difficult to guarantee the accuracy of operations. On the other hand, LWE-based homomorphic encryption can support all forms of operations, but it has the disadvantage of slow processing speed. The LWE method guarantees accurate calculations based on inner product operations, but it is inefficient compared to the RLWE method in terms of performance. In particular, performance can be severely degraded in applications requiring artificial neural networks or complex operations. Existing homomorphic encryption methods performed operations by selecting one of these two encryption methods. In particular, since systematic conversion between RLWE and LWE was not supported, it was impossible to utilize the advantages of both methods simultaneously. As a result of having to choose only one method, existing methods found it difficult to strike a balance between computational performance and accuracy, and in many cases, acted as a factor that reduced computational efficiency. Therefore, generally, RLWE-based methods are used when performance is prioritized, while LWE-based methods are used when accuracy is emphasized. However, in this case, users face a situation where they must sacrifice either performance or accuracy depending on the requirements of a specific application. In particular, in fields such as artificial neural network computation or Private Set Intersection processing sensitive data, accuracy is critical, but practical application has been difficult due to the problem of having to excessively sacrifice performance. To overcome these limitations, it is necessary to introduce conversion technology between RLWE and LWE to utilize a method that takes advantage of both methods. FIG. 1 is a configuration diagram of an encryption data operation system according to one embodiment. FIG. 2 is a configuration diagram of a user terminal and an encryption data operation device according to one embodiment. FIG. 3 is a flowchart showing the steps of an operation performed by an encryption data operation device according to one embodiment. FIG. 4 is a conceptual diagram briefly illustrating the flow of data processed according to the operation performed by an encryption data operation device according to one embodiment. FIG. 5 is an example diagram showing the process of converting the encryption system of data while performing operations on the first code and the second code included in the source code according to one embodiment. FIG. 6 is an example diagram illustrating the mutual conversion process between the first format data of the RLWE system and the second format data of the LWE system performed by an encryption data operation device according to one embodiment. Figure 7 is a comparison table comparing the computation time and accuracy of the existing encryption methods RLWE and LWE with the method of an embodiment of the present invention. Detailed information regarding the purpose, technical configuration, and resulting effects of the present invention will be more clearly understood through the following detailed description based on the drawings attached to the specification of the present invention. An embodiment according to the present invention will be described in detail with reference to the attached drawings. The embodiments disclosed herein should not be interpreted or used to limit the scope of the invention. It is obvious to those skilled in the art that the description including the embodiments herein has various applications. Accordingly, any embodiments described in the detailed description of the invention are illustrative for better explaining the invention