Available Thesis Topics

When applying for a thesis topic, follow the procedure described here and CC the advisor(s) in your email.

Click on a topic for more details.


Bachelor Topics

Master Topics

Memory tiering technologies such as persistent memory (PMEM) and CXL are leading to setups where several hosts share the same memory pool. Operating systems, however, still assume that they exclusively own the metadata describing physical memory.

In Linux, each physical page is described by a struct page. For PMEM devices, this array of descriptors (the memmap) can either be kept in each host’s DRAM (--map=mem), or stored on the device itself (--map=dev) to save DRAM. With --map=dev, however, kernels sharing the same device overwrite each other’s struct page metadata.

Goal of this thesis: Using struct page as a case study, you will explore whether simple kernel data structures can be shared across multiple kernels. In particular, you will:

  • Set up multiple KVM virtual machines attached to a single shared, emulated PMEM device
  • Identify which fields of struct page must be shared across kernels and which must be kept private
  • Design and prototype a kernel mechanism that makes --map=dev safe across hosts
  • Evaluate your solution in terms of correctness, DRAM savings and performance overhead

Target: M.Sc. students

Prerequisites:

Advisor: Jérôme Coquisart

Confidential computing aims at protecting data in use, even from a compromised or malicious cloud provider. To this end, all major CPU vendors now provide hardware support for confidential virtual machines (CVMs), whose memory is encrypted and isolated from the hypervisor: Intel TDX, AMD SEV-SNP, Arm CCA, and the upcoming RISC-V CoVE. These technologies are already deployed by major cloud providers.

By design, each CVM’s memory is private. Today, CVMs that need to exchange data must either go through unencrypted memory shared with the hypervisor, or communicate over the network, in both cases paying the cost of encrypting and copying the data in software. However, many use cases, such as co-located services or shared caches, would benefit from an encrypted memory region shared between multiple, mutually trusting CVMs. Such a mechanism has recently been proposed for Arm CCA (CAEC), but the question remains open for other architectures, in particular x86.

Goals of the thesis: In this context, you will:

  • Study the design of secure VM technologies from Intel, AMD, Arm and RISC-V, and compare their security guarantees and memory protection mechanisms
  • Survey existing research work and real-world deployments that build on these technologies
  • Explore whether, and how, these technologies could support sharing an encrypted memory region between multiple CVMs, with a focus on x86 (AMD SEV-SNP and Intel TDX)
  • Design and prototype such a sharing mechanism on an x86 platform (real hardware or emulated)
  • Evaluate your prototype and compare it with CAEC, in terms of security guarantees, performance and required hardware/software changes

Target: M.Sc. students

Prerequisites:

  • Programming language: C
  • Good knowledge of operating systems and computer architecture (virtual memory, virtualization)
  • Linux Kernel Programming is appreciated

To get more familiar with the topic, you can start with having a look at the following papers:

  • M. Misono, D. Stavrakakis, N. Santos, and P. Bhatotia, “Confidential VMs Explained: An Empirical Analysis of AMD SEV-SNP and Intel TDX,” in ACM SIGMETRICS, 2025.
  • X. Li, X. Li, C. Dall, R. Gu, J. Nieh, Y. Sait, and G. Stockwell, “Design and Verification of the Arm Confidential Compute Architecture,” in USENIX OSDI, 2022.
  • R. Sahita et al., “CoVE: Towards Confidential Computing on RISC-V Platforms,” arXiv:2304.06167, 2023.
  • J. Z. Yu, S. Shinde, T. E. Carlson, and P. Saxena, “Elasticlave: An Efficient Memory Model for Enclaves,” in USENIX Security, 2022.
  • S. Abdollahi, A. Al Sadi, D. Kotz, M. Kogias, and H. Haddadi, “CAEC: Confidential, Attestable, and Efficient Inter-CVM Communication with Arm CCA,” arXiv:2512.01594, 2025.

Advisor: Redha Gouicem