Automatic discovery of software attacks via backward reasoning

Security risk management and mitigation are two of the most important items on several companies’ agendas. In this scenario, software attacks pose a major threat to the reliable execution of services, thus bringing negative effects on businesses. This paper presents a formal model that allows the identification of all the attacks against the assets embedded in a software application. Our approach can be used to perform the identification of the threats that loom over the assets and help to determine the potential countermeasures, that is the protections to deploy for mitigating the risks. The proposed model uses a Knowledge Base to represent the software assets, the steps that can be executed to mount an attack and their relationships. Inference rules permit the automatic discovery of attack step combinations towards the compromised assets that are discovered using a backward programming methodology. This approach is very usable as the attack discovery is fully automatic, once the Knowledge Base is populated with the information regarding the application to protect. In addition, it has been proven highly efficient and exhaustive.