Study of self-healing mechanisms for embedded systems

While topics around protection mechanisms or threat detection are currently well covered, post-attack repair is still not very present in the scientific literature.

Objectives: The objective of this thesis is to identify solutions to restore critical functions or tasks of an embedded device after the detection of an unwanted event (due to a malicious attack or a software or hardware failure). For the last twenty years, many research works have focused on the definition of restoration solutions in the field of safety, the application of these techniques to the field of security is more recent, and still few works address the applicability of these methods to this field.

If the restoration principles can be identical in the safety and security domains, in the case of malicious action, it is necessary to make the restoration system robust against corruption attempts by the attacker. Indeed, the attacker can evolve his attack in order to lure the restoration system or even corrupt the restoration itself. One of the starting hypotheses taken in this thesis is that the system already has one or more modules for detecting abnormal events. From this abnormal state, the solution must allow the restoration of the software after having verified the integrity of the failed software block (rollback) thanks to a context backup (checkpoint) at regular intervals. Within the framework of this thesis, the study will only focus on embedded targets such as processors integrating an embedded Linux but should be portable to other environments

Facebook

Twitter