IBF: An Intelligent Black-box Fuzzing Method for Distributed Responsive System

Distributed responsive system has the features of component separation based on IP network connectivity and real-time response based on message driven or event driven mechanism, which is widely used in the construction of large-scale serving systems. Due to confidentiality considerations, the access to their source code is usually unavailable. Therefore, when conducting security checks on the system externally, black-box fuzzing is generally used. However, the traditional black-box fuzzing is not capable for the guidance on the fuzzing data generation due to the lack of code coverage and other indicators, resulting in poor testing effectiveness. To address the problem, an Intelligent Black-box Fuzzing method for distributed responsive system (IBF) is proposed, which uses clustering algorithm (CA), Wasserstein generativ eadversarial network with gradient penalty (WGAN-GP), and genetic algorithm (GA) to assist the guidance on the fuzzing data generation. For the convenience of description, we use Evolved Packet Core (EPC) as the research target, which is a typical distributed responsive system in Long Term Evolution (LTE). The result of black-box fuzzing on open-source EPC platform shows that the method proposed has a better performance on triggering crashes of the target than traditional black-box fuzzing methods.