AuDiNoMiC: um gerenciador autonômico para auditorias em segurança ofensiva

Abstract

Information security (IS) is of significant importance today because all computer systems need some sort of security layer to maintain their reliability and knowledge base. For a system to have this type of security, it is necessary to employ invasion tests capable of verifying the existence of vulnerabilities that allow its preventive correction against real digital attacks. Such tests are specialized audits and require specific knowledge and considerable investment, both for the hiring of a specialized third party or for necessary training of human resources already available in an institution. In this way, automated computational tools can facilitate the process of the invasion tests, providing diagnoses that can still be interpreted by the expert during the process of these tests. This dissertation proposes an autonomic computational tool called AuDiNoMiC to deal with audit problems in information security. The purpose of this tool is to automate the execution of the intrusion tests to provide diagnoses similar to those of an information security specialist. To validate the tool, a field experiment is carried out, carrying out invasion tests in real domains on the Internet. The results of this experiment indicate that the tool is able to perform the invasion tests successfully, providing information equivalent to those of a human expert. Therefore, the proposed tool is a viable alternative to reduce the operational investments of the tests and losses with incidents related to information security.