International Association for Cryptologic Research

International Association
for Cryptologic Research

IACR News item: 21 December 2023

Cas Cremers, Alexander Dax, Niklas Medinger
ePrint Report ePrint Report
Key Encapsulation Mechanisms (KEMs) are a critical building block for hybrid encryption and modern security protocols, notably in the post-quantum setting. Given the asymmetric public key of a recipient, the primitive establishes a shared secret key between sender and recipient. In recent years, a large number of abstract designs and concrete implementations of KEMs have been proposed, notably in the context of the NIST selection process for post-quantum primitives. The traditional security notion for KEMs has been the IND-CCA notion that was designed for public-key encryption (PKE). In recent work additional properties, such as robustness and anonymity, were lifted from the PKE setting to the KEMs setting.

In this work we introduce several stronger security notions for KEMs. Our new properties formalize in which sense outputs of the KEM uniquely determine, i.e., bind, other values. Our new notions are based on two orthogonal observations: First, unlike PKEs, KEMs establish a unique key, which leads to natural binding properties for the established keys. Our new binding properties can be used, e.g., to prove the absence of attacks that were not captured by prior security notions, such as re-encapsulation attacks. If we regard KEMs as one-pass key exchanges, our key-binding properties correspond to implicit key agreement properties. Second, to prove the absence of weak keys, we have to consider not only honestly generated key pairs but also adversarially-generated key pairs.

We define a hierarchy of security notions for KEMs based on our observations. We position properties from the literature within our hierarchy, provide separating examples, and give examples of real world KEMs in the context of our hierarchy.
Expand

Additional news items may be found on the IACR news page.