Evaluation of Integrating Post-Quantum Cryptography for Signal Desktop
Abstract
Quantum computing poses a serious long-term risk to modern cryptographic systems, including the widely used Signal messaging protocol. This research work addresses that risk by integrating ML-KEM-768, a NIST-standardized post-quantum algorithm specified in FIPS 203, as an additional security layer within Signal Desktop. The proposed implementation applies a quantum-resistant wrapping layer to message content before encryption by Signal’s existing Double Ratchet protocol, thereby creating a hybrid cryptographic design that strengthens protection against both classical and quantum adversaries. The system uses ML-KEM-768 key encapsulation from the @noble/post-quantum library together with AES-256-GCM authenticated encryption from @noble/ciphers. This process produces 1088-byte KEM ciphertexts with 32-byte shared secrets and adds about 1104 bytes of overhead per message. Performance evaluation using two live Signal Desktop instances shows that the additional post-quantum layer introduces only modest latency. ML-KEM-768 encapsulation averages 0.67ms, decapsulation averages 2.27ms, and total message preparation overhead averages 13.0ms, which remains well below the 100ms threshold commonly associated with instantaneous user interaction. Additionally, a second set of experiments evaluated image attachment encryption (100 KB, 500 KB, and 2 MB), confirming that ML-KEM-768 encapsulation cost remains constant at approximately 0.4 ms regardless of payload size while AES-256-GCM bulk encryption scales linearly, making the post-quantum overhead negligible for large media. A third stress test evaluated robustness under load, achieving 913 messages per second in burst mode with sub-millisecond wrap latency, and confirming stable performance with no degradation across sustained operation at 10 messages per second over 60 s. Overall, this work demonstrates that post-quantum cryptography can be integrated into a real messaging system in a practical manner, helping address the harvest now, decrypt later threat while preserving usability and strengthening long-term communication security.