Project Description
Secure messaging systems are used by billions daily through applications like WhatsApp, Signal and iMessage. Secure messaging protocol design is complicated by the many security, functional and performance requirements that must be balanced. Due to their importance in protecting users from across society, it is important to understand exactly how these cryptographic protocols are designed, what security guarantees they provide, as well as what kind of guarantees are possible and at what cost.
Conversations in messaging can last for years, by contrast to traditional protocols like TLS, and users may periodically be offline, so protocols must support asynchronous communication and conversation initiation. Group messaging further complicates things, where users can be dynamically added or removed from groups, which can contain hundreds or even thousands of users. Further aspects of modern secure messaging include multi-device support, resilience to device compromise, and plausible deniability.
In this project, we propose to explore the theory and/or practice of secure messaging through a cryptographic lens. Below are a few salient research directions. We encourage the prospective applicant to talk to Daniel Collins to explore ideas and find a project that best suits their particular interests and skills.
The Security of Message Layer Security (MLS)
MLS has garnered significant attention from industry and academia, including through its adoption by Apple and Google for end-to-end encryption in Rich Communication Services (RCS), the successor to SMS, as well as by companies like Cisco and AWS in its design and deployment. Much of the work on MLS has focused on analysing and improving the security of the core cryptographic protocol specified in RFC 9420 [1], the IETF standard released in 2023. In this work, we will analyse the security of MLS, both in its (complex) cryptographic core as well as through various implementation considerations and use cases not captured by the core standard. The core of the project will be in cryptographic, provably secure modelling; see [2, 3] for how this has been done for MLS in the past.
Benchmarking in Messaging
Many academic and practical proposals exist for secure messaging in both two-party and group settings, but comparatively little work has evaluated their performance. Performance evaluation is particularly important for group messaging, where costs can be substantially higher and depend on the performance metrics, network conditions, and usage patterns considered. This project aims to evaluate the performance of existing messaging protocols, in the two-party and/or group messaging settings. Time permitting, it will entail a detailed comparison of secure group messaging approaches based on pairwise two-party channels, Sender Keys [4], and MLS under varying network conditions and usage scenarios, including differences in latency, bandwidth, and group size.
Leakage-Resilient Messaging
Since conversations are often long-lived and devices are particularly vulnerable to compromise, protocols like the Double/Triple Ratchet [5,6] and MLS [1] are resilient to cryptographic state compromise. Namely, they provide forward security (which protects past secrets upon compromise) as well as post-compromise security (automatically restoring security) through regular key rotation. Existing security models generally assume that the adversary compromises the entire state of a user. However, often the adversary may only learn partial information about a user’s state. In this case, [7] showed that it is possible to obtain additional security guarantees, following the tradition of leakage-resilient cryptography. However, [7] represents only an initial study of leakage-resilient messaging in a simplified communication model between two parties (namely, where only one party can send to the other), and so the goal of this project is to explore leakage-resilient messaging in more realistic models and settings and ideally bridge theory and practice.
References
- https://datatracker.ietf.org/doc/rfc9420/
- https://eprint.iacr.org/2019/1189
- https://eprint.iacr.org/2021/1083
- https://eprint.iacr.org/2023/1385
- https://signal.org/docs/specifications/doubleratchet/
- https://eprint.iacr.org/2025/078
- https://eprint.iacr.org/2025/332