No abstract

No abstract

No abstract

Control systems can show robustness to many events, like disturbances and model inaccuracies. It is natural to speculate that they are also robust to alterations of the control signal pattern, due to sporadic late completions (called deadline misses) when implemented as a digital task on an embedded platform. Recent research analysed stability when imposing constraints on the maximum number of consecutive deadlines that can be missed. This is only one type of characterization, and results in a pessimistic analysis when applied to more general cases. To overcome this limitation, this paper proposes a comprehensive stability analysis for control systems subject to a set of generic constraints, describing the possible sequences of correct completions and deadline misses. The proposed analysis brings the assessment of control systems robustness to computational problems one step closer to the controller implementation.

Predictable and repeatable execution is the key to ensuring functional correctness for real-time systems. Scheduling algorithms are designed to generate schedules that repeat after a certain amount of time has passed. However, this repeatability is also a vulnerability when side-channel attacks are considered.Side-channel attacks are attacks based on information gained from the implementation of a system, rather than on weaknesses in the algorithm. Side-channel attacks have exploited the predictability of real-time systems to disrupt their correct behavior.Schedule Randomization has been proposed as a way to mitigate this problem. Online, the scheduler selects a schedule among a set of available ones, trying to achieve an execution trace that is as different as possible from previous ones, therefore minimizing the amount of information that the attacker can gather.This thesis investigates fundamental limitations of schedule randomization for a generic taskset. We then propose an algorithm to construct a set of schedules that achieves a differentation level as high as possible, using the fewest number of schedules, for tasksets with implicit deadlines. The approach is validated with synthetically generated tasksets and the taskset of an industrial case study, showing promising results. Abstract Keywords Classification system and/or index terms (if any) Supplementary bibliographical information ISSN and key title ISBN Language Number of pages Recipient's notes Security classification