Research On Parallel Detection Method For Software Vulnerability

With the rise of technology in cloud computing,Internet plus and Internet finance and other traditional industries,massive data and business are migrating to the Internet,the attendant security issues have to arouse our deep concern,to timely find and fix software vulnerabilities before the attack is the core to solve security problems.In order to adapt to the scene of processing massive data and business,parallelization is a key method to improve the speed of response to security incidents.After reasearching the key technology of software vulnerability detection method in parallel optimization,and the advantages and disadvantages of existing methods of work in parallel multi pattern string matching,this paper proposes a parallel multiple pattern string matching algorithm based on multi-core,through the heuristic algorithm optimization in pattern set,to improve the execution speed of parallel string matching algorithm.At the same time,this paper proposes a sensitive information based parallel dynamic symbolic execution path optimization method through the sensitivity function judgment and path pruning optimization in order to optimize the high computational cost problem generated by the path explosion from parallel dynamic symbolic execution.In the study of the PHP interpreter implementation process,our work put forward a method of real-time application self-protection based on image code execution,for the relief of problems of affecting business performance in the former adaptive security architecture of RASP system and other scanning methods that deployed in the production environment.Finally,this paper will put forward the methods and results of application in practice,using Python and MPI technology to realize a distributed real-time application image code execution system based on self-protection,then Web vulnerability testing platform deployment on the system,in order to test for theprotective performance in known and unknown vulnerabilities.The test results of false positive and false negative rates were 11.1% and 29.6%;the acceleration ratio of the response time with various number of nodes is 1.28;in unknown vulnerability mining,the effect of detection with the combination of active dynamic symbolic execution test and passive testing compared to only passive testing has an increase of about 10.4%.