|Team||Distributed Systems, Information Retrieval and Mobility|
|Institution||Institut National des Sciences Appliquées de Lyon|
|Location||Blaise Pascal (INSA)|
|mohamed.maouche at liris.cnrs.fr|
|Subject||User-Centric Dependability and Privacy in Distributed Systems: Towards Multi-Criteria Guarantees|
|Abstract||Distributed systems have always been the scene of various software and hardware failures. These failures can have diverse sources such as the crash of machines, bugs, misconfigurations, the use of untmsted data sources,as well as malicious attacks and users that deliberately tamper with their software to gain some beneﬁt. These failures are especially difficult to deal with when the distributed system directly relies on end-users (e.g., peer-to-pecr systems) or spans over multiple administrative domains, also referred to as MAD distributed systems (e.g., multi-cloud based systems).
A number of solutions have been proposed in the last decade to deal with software and hardware failures indistributed systems. The most effective solutions in this domain often rely on the secure logging of the nodes interactions and on the periodic inspection of these logs in order to deter anomalies in the behaviour of nodes).
However, most of these solutions are highly intrusive on the one hand (as logs may reveal sensitive information about participants) and poorly efﬁcient on the other hand. Both issues have a direct impact on the user perceived experience, whether on the observed performance degradation or on the potential privacy leakage imposed by the used log inspection mechanisms.
Contrary to existing solutions that trade robustness against privacy and performance, this PhD project aims at developing a multi—criteria approach for designing distributed systems thus conciliating dependability, privacy and performance objectives. Speciﬁcally, this PhD project aims at investigating novel languages, protocols and tools for modelling and providing multi—eriteria guarantees for building resilient, privacy-preserving and user-centric distributed systems. This objective is particularly challenging for highly distributed MAD distributed systems, where the user experience is crucial in the evaluation of the quality delivered by the overall system.
|Advisor||Sonia Ben Mokhtar|
Last update : 2016-09-22 11:12:23