Reconciling Package Inconsistencies in Satellite Operations: A Data Mining Approach

When trying to penetrate a system, one advantage an attacker can have is the time a system administrator might take to update the packages or software running on the systems. The packages may be subject to a number of vulnerabilities. A company in the satellite industry uses ground control systems to fly and to maneuver orbiting satellites. While these systems pursue a common and identical goal, their configurations show package inconsistencies. The purpose of this thesis is the reconciliation of those inconsistencies. This task can be difficult if systems are numerous, critical and complex. When they have accumulated these inconsistencies for some time, it is difficult to return to a healthy state. In the first place, a study of the machine configurations is proposed, then, three approaches are proposed to reconcile the packages. The first approach selects outlier packages based on the configuration of a centroid machine. The second approach selects outlier packages based on a healthy machine. The third approach computes the intersection of configurations to detect outlier packages. After removal of the selected packages, systems are more secure and standardized, allowing better administration for the company and those that encounter the same kind of problem.