Network security protocol fuzz test method and system based on LSTM

Abstract

The invention provides a network security protocol fuzz test method and system based on an LSTM. The method comprises the steps of: S1, carrying out an initial fuzz test on a target network security protocol by adopting an initial test case set; S2, carrying out vulnerability analysis on the target network security protocol according to a new execution path generated in the fuzz test process, andusing the new execution path as an initial seed file set; S3, using the initial seed file set as training data, and carrying out training by utilizing a deep neural network learner to obtain an LSTM model; and S4, generating a new test case set by utilizing the LSTM model so as to carry out the fuzz test on the target network security protocol. The system comprises a first fuzz test module, a vulnerability analysis module, an LSTM model generation module and a second fuzz test module. According to the invention, by training the LSTM neural network model to optimize a generation algorithm of test cases, so that a code coverage rate of the test cases is improved.

Description

technical field [0001] The invention relates to the technical field of network information security, in particular to an LSTM-based network security protocol fuzz testing method and system. Background technique [0002] With the continuous advancement of the information revolution, the network has become an indispensable part of people's lives. The safe transmission of sensitive information on the network is related to the development of individuals, enterprises and even the country. Security protocols provide users with information protection services in the Internet based on cryptographic algorithms, which is an important content in the field of information security. But at the same time, network security incidents broke out frequently, the number of network security vulnerabilities showed an increasing trend, and network information security faced a huge challenge. Therefore, it is of great practical significance to evaluate the security of network security protocols. ...

AI Technical Summary

Problems solved by technology

When the target protocol is tested using the existing fuzzing methods, these verification algorithms or encryption algorithms will cause a large number of data packets in the target protocol to be discarded because they are invalid, resulting in the test coverage of the target protocol by the existing fuzzing tools Very narrow range and less efficient

Therefore, some fuzz testing methods have emerged to solve this problem by reproducing the verification algorithm or encryption algorithm of the target protocol, but the operation process is complicated and it is more difficult to reproduce complex verification algorithms or encryption algorithms

Images