My research in 90 seconds: Outsourcing computational tasks. How can you trust the result?

09.02.18

University of LuxembourgDiesen Artikel drucken

“Of course, just outsource it to a cloud!” – A common practice for a lot of firms. But is that reliable? Can you trust the result?

Explaining its research in less than 90 seconds: 7 young researchers from Luxembourg took up the challenge in the new video series "My research in 90 seconds". In this episode, dive into the world of cloud computing with Balázs Pejó! More about this in the video - and further details in his article.

Author text: Balázs Pejó 
Video:
Jean-Paul Bertemes (FNR) & Moast Creative Studios

Outsourcing to the Cloud

If you pay someone to do something instead of doing it yourself is called outsourcing. It is one of the big buzzwords since the turn of the millennium, and its popularity is growing ever since. It helps firms to mitigate a shortage of skill or expertise and offers greater budget flexibility and control. A special form of outsourcing is cloud computing; a type of internet-based computing that provides computer processing resources on demand. Despite the numerous advantages, cloud computing poses significant risks such as privacy and data corruption. And at the end remains the question: Can I trust the results? How can I know that the calculations have really been done correctly?

Cryptography vs. Game Theory

During the last few decades, various mathematical solutions emerged to tackle these problems, however, using them on very large datasets is still inefficient and also unnecessary: Cryptographic protocols protect against malicious players. However in real life adversaries are not malicious, but rational: they will maximize their profit instead of attacking in any case. This leads to use Game Theory instead of Cryptography, which is the study of conflict and cooperation between rational decision-makers.

Protection Against Rational Servers

Scientists from Luxembourg Institute of Science (LIST) and University of Luxembourg (UNILU) developed a framework which can determine the minimal amount someone should pay to a cloud server to ensure that the calculation will be done correctly. They therefore used not cryptography, but the game-theory. The game-theoretic model formulated by Qiang Tang - an ex-researcher at the Interdisciplinary Centre of Security, Reliability and Trust (SnT) - and Balázs Pejó - a PhD candidate at the Applied Security and Information Assurance Group (ApSIA) - captures the behavior of a rational cloud service provider. In the game design, a client declares an offer to the cloud server who either rejects this offer or accepts it and executes the requested computation either correctly or with some level of manipulation.

Making Cheating Irrational

The client's goal is to declare such an offer in which case the server's best interest is to do the calculations correctly. For this reason, the payment should be sufficiently high. On the other hand, the client wants to pay as little as possible. The model helps the client to determine this sweet-spot: the payment just enough to make the rational server to do the computation honestly.

Auch in dieser Rubrik

FNR Attract Fellows: Cesar Pascual Garcia entwickelt die nächste Generation medizinischer Sensoren

19.01.18 Sensoren sind auch in der Medizin verbreitet: Der Nanotechnologe Cesar Pascual Garcia will sie noch besser machen. > Ganzen Artikel lesen

Wie im echten Leben: Math.en.Jeans macht aus Mathematik ein Erlebnis

10.01.18 Mathe kann rocken. Wenn reale Themen behandelt werden, macht das „Spiel“ mit den Zahlen Spaß – wie das Projekt MATh.en.JEANS beweist. > Ganzen Artikel lesen

Raumfahrttechnologie mit weniger Gewicht: Naoufal Bahlawane entwickelt eine „Super-Schwarz-Beschichtung“

19.12.17 Naoufal Bahlawane entwickelt am LIST eine so genannte „Super-Schwarz-Beschichtung“ für die Europäische Weltraumorganisation > Ganzen Artikel lesen

Mit „Kniwwelino“ lernen schon Grundschulkinder programmieren

06.12.17 Die Forscherin Valérie Maquil vom LIST ermöglicht eine außerschulische »Informatikausbildung«. > Ganzen Artikel lesen

Autonomes Fahren: Intel und SnT kooperieren, um selbstfahrende Autos sicherer zu machen

30.11.17 Je komplexer autonome Fahrzeuge sind, desto größer die Gefahr von Hackerangriffen. Forscher arbeiten an einem System, das Angriffe automatisch neutralisiert. > Ganzen Artikel lesen