Security of Homomorphic Encryption
Cyber Security
Client company:Fontys Research Group
Lukas Jansen & Erik van de Laar
Project description
The main research question of the project is: "What are the security implications of using Homomorphic Encryption?"
During this project, we have set up our own test environment in C++, using the Microsoft SEAL library, in order to conduct empirical tests on Homomorphic Encryption.
Context
A use case for Homomorphic Encryption could be to encrypt data between a medical institution and a cloud provider. Medical organizations have lots of patient-related data, and would like to perform complex computations on that data. These computations oftentimes require a lot of compute power, and would preferably be outsources. However, you then have a privacy issue. How can the hospital guarantee the confidentiality of the data once it has been given to the cloud provider?
Using Homomorphic Encryption, the hospital could homomorphically encrypt the data, such that the cloud provider could still compute over it. The computations happen on encrypted data, so the cloud provider will not know the contents.
Results
In our research, we dove into the topic of Homomorphic Encryption. On a practical level we explored how it works and how it could be implemented.
We investigated ways to deduce information from the encrypted data (cryptanalysis), and explored various research papers that have investigated similar security questions.
Our research was not just theoretical. We also applied the knowledge empirically, and used the SEAL library to create our own 'test environment' in the C++ programming language. In this environment, we conduced several test cases to find out more about HE's possible security implications.
About the project group
Erik and I (Lukas) both come from a software engineering background. For the both of us, this is the second cyber security semester. In this semester, Erik focused on Red Teaming, while I focused on Security Development.
At the start of the semester, we were the only two students that were interested in the Homomorphic Encryption topic. Under normal circumstances, a group project would consist of about 5 to 7 students, so we are lucky that we were allowed to continue with the project.