A Design Method of Indistinguishable Obfuscator Based on Adversarial Graph Neural Network Structure

Abstract

The invention discloses a design method of an indistinguishable obfuscator based on an adversarial graph neural network structure. Replacement of circuits; random coverage of the positions of equivalent circuits by sampling by subcircuit order; evolutionary obfuscation of input circuits by subcircuit-based differential evolution; dual adversarial graph neural network structure for differential The evolution results are differentiated and fed back to achieve statistical indistinguishability. The method of the present invention automatically constructs an obfuscated circuit with the same circuit function for a computing circuit of any structure through the automatic evolution technology of the graph and the confrontation structure, and satisfies the statistical suboptimal indistinguishability, versatility and function retention.

Description

technical field [0001] The invention belongs to the technical field of computer program security protection, and particularly relates to a design method of an indistinguishable obfuscator based on an adversarial graph neural network structure. Background technique [0002] Indistinguishable obfuscator (iO) is an important concept in cryptographic design in recent years. It was originally developed from the realization of the copyright protection mechanism of program code to achieve the purpose of preserving the function of the code without revealing the execution process. For a similar purpose, iO may encrypt the function and execution process of the cryptographic algorithm to protect the execution of the cryptographic algorithm. In a broad sense, the use of secure and efficient iO can easily implement asymmetric encryption algorithms, identity-based encryption algorithms and various key encapsulation algorithms, so that data encryption is no longer limited by the selection ...

AI Technical Summary

Problems solved by technology

The iO constructed according to this idea is used to further construct a higher-level cryptographic protocol, which will make it difficult to effectively implement the new cryptographic protocol in terms of complexity and operational efficiency.

[0006] In summary, the current design of indistinguishable obfuscators is facing a contradiction between versatility and circuit function maintenance, and a new design method for indistinguishable obfuscators based on an adversarial graph neural network structure is urgently needed

Images