Research On Key Technology Of Security Protocols Vulnerability Analysis Based On Model Learning

With the continuous advancement of the information revolution,the Internet has become an indispensable part in people's life.Secure transmission of sensitive information on the Internet is related to the development of individuals,businesses and even the countries.Security protocols provide information protection services for users in the Internet based on cryptographic algorithms,which is an important part of information security.On the other hand,cyber security incidents frequently broke out and the number of cyber security vulnerabilities is increasing.Then the network information security is facing a huge challenge.Therefore,it is of great practical significance to evaluate the security of network security protocols.Based on the methods of model learning,recurrent neural networks and other related theories,this paper focuses on several key issues in the automated analysis of network security protocol vulnerabilities.The main work is summarized as follows:1.A vulnerability analysis method for network security protocols based on model learning technology is proposed.Based on the classical MAT(Minimally Adequate Teachers)framework,analysis of the target protocol implementation logic is implemented with black-box test by automatically deduced its protocol state machine.It can be used to provide a basis for vulnerability analysis and attack paths discovery;the time compression model and algorithm for the protocol implementation state machine are proposed to improve the vulnerability analysis and the efficiency of the attack paths discovery.2.A fuzzing method for security protocols based on LSTM neural network model is proposed.The fuzzing method with compiling-time instrumentation is applied to the vulnerability analysis of network security protocols.Then combine its advantage of directive test cases generation with deep neural network models to optimize the test case generation algorithm.Trains the LSTM model and improve the code coverage of the vulnerability analysis.3.The model-based learning method is applied to the vulnerability analysis of OpenVPN protocol to automatically infer the state machine of the server-side,and the special behavior paths and potential security risks outside the expected behavioral paths are discovered;the fuzzing method based on LSTM neural network is used for OpenSSH system analysis.Experiments show that the LSTM model's capability of paths discovering is improved by about 54% compared with the traditional method and about 17% higher than the random strategy.And the more training data,the longer training time and the better effect.In this paper,the black-box Inference of state machine and the intelligent generation of test cases are implemented.The vulnerability analysis method and optimization algorithm of the security protocols are proposed.Experiments show that the proposed methods are correct and effective.The research results of this paper provide theoretical and technical support for the vulnerability analysis of large-scale and practical security protocols.