Keynotes

Trent Jaeger, Pennsylvania State University (USA)

How Much Control Should Customers Demand over Their Applications When Deployed in the Cloud?

Abstract: Cloud computing promises its customers simplified access to managed resources, but to obtain this benefit customers must relinquish visibility and control over their computing infrastructure. As a result, customers who aim to transfer applications from their data centers to the cloud lose the ability to monitor the security of their applications accurately. For example, customers cannot determine whether their applications are deployed as expected nor monitor their runtime behavior. This problem is exacerbated because cloud environments also introduce a variety of software infrastructure and services that applications must depend upon to execute in the cloud. In this talk, we will survey several challenges and proposed solutions to the problem of restoring customer control their computing infrastructure in IaaS clouds. First, we will discuss the problem of monitoring applications in clouds administered by others. Second, we will examine the problem of providing customers with the flexibility to perform the kinds of monitoring that they desire. Third, we will investigate the problem of reducing the cloud's trusted computing base. Thoughout this talk, we will account for the cloud vendors' desire to simplify application deployment and maintain the security of their cloud environment for all customers. Within those limitations, however, we argue that customers can demand greater control and cloud vendors can provide such methods without compromising their ability to manage their cloud environments.

Rosario Gennaro, City University, New York (USA)

A Survey of Verifiable Delegation of Computations

Abstract: In this talk I will give an overview of past and recent research on the area of Verifiable Delegation of Computation. The goal is to enable a computationally weak client to "outsource" the computation of a function F on various inputs x_1,...,x_k to one or more powerful servers. The server must return the result of the function evaluation, e.g., y_i=F(x_i), as well as a proof that the computation of F was carried out correctly on the given value x_i. A crucial requirement is that the verification of the proof should require substantially less computational effort than computing F(x_i) from scratch. For the "general purpose" case (protocols that work for any function F) I will discuss the different ways this problem has been approached theoretically, particularly the line of research that links Interactive Proofs, to Probabilistic Checkable Proofs, to Succinct Non-Interactive Arguments. I will also survey recent exciting experimental results that show how these techniques are on the verge of becoming practical. I will also talk about "ad hoc" protocols that aim to verify specific computations of particular importance in practice.

Stefan Katzenbeisser, Technische Universität Darmstadt (Germany)

Towards Practical Secure Computation

Abstract: The increasing use of cloud services brings new challenges regarding the protection of privacy sensitive data. While in the past privacy was mainly assured through regulatory approaches, access control and audits, these mechanisms tend to be inappropriate for largely distributed systems. New cryptographic protection mechanisms come to rescue: they allow to make sensitive data available for various applications, while protecting them from misuse. This talk gives an overview of recent approaches in the construction of cryptographic protocols that are based on homomorphic encryption and Secure Multiparty Computation, which allow to privately process sensitive data and prevent data leakage to insiders.