Cloud service providers (such as Tresorit) that offer end-to-end protection to client data rely on client-side encryption that ensures confidentiality of data in transit and data at rest even with respect to the service provider. The advantage of this model is that it minimizes the amount of trust that clients need to put in the cloud service provider. The downside, however, is that the service provider cannot offer services that require processing of the stored data, as they are encrypted and the service provider cannot decrypt them. A possible technology that can be used to address this issue is confidential computing, which offers protection for sensitive data during computation through hardware-enforced isolation and cryptographic verification of the execution environment. 

The task of the student is to design and implement a confidential computing architecture that allows clients to work with sensitive data stored in the cloud without actually downloading the data. The solution can be based on launching a confidential virtual machine (CVM) within the same cloud environment where the encrypted data is stored, loading the encrypted data into the CVM, and allowing clients to access the CVM remotely, and to decrypt and perform computation on the data within the CVM.

Tasks to be performed by the student include:

• Review the state-of-the-art of confidential computing and compare the available confidential computing technologies regarding trust assumptions, level of isolation provided, and usability aspects;
• Design a confidential computing architecture that relies on a CVM running in the cloud and allows for attestation of the CVM before using it, secure remote access to the CVM for clients, and decrypting of and working with sensitive data on the CVM;
• Implement the architecture in the Azure cloud environment and integrate it with Tresorit, such that a Tresorit client application can be executed in the CVM to decrypt encrypted data stored by Tresorit;
• Demonstrate the concept by integrating a collaborative document editing tool with the implemented architecture allowing multiple clients to join a collaborative session through the CVM and to work on a sensitive document;
• Evaluate the proposed solution qualitatively in terms of security, performance, and usability.