Key establishment - security models, protocols and usage

Key establishment is the process whereby two or more parties derive a shared secret, typically used for subsequent confidential communication. However, identifying the exact security requirements for key establishment protocols is a nontrivial task. This thesis compares, extends and merges existing security definitions and models for key establishment protocols. The primary focus is on two-party key agreement schemes in the public-key setting. On one hand new protocols are proposed and analyzed in the existing Canetti-Krawzcyk model. On the other hand the thesis develops a security model and novel definition that capture the essential security attributes of the standardized Unified Model key agreement protocol. These analyses lead to the development of a new security model and related definitions that combine and extend the Canetti-Krawzcyk preand postspecified peer models in terms of provided security assurances. The thesis also provides a complete analysis of a one-pass key establishment scheme. There are security goals that no one-pass key establishment scheme can achieve, and hence the two-pass security models and definitions need to be adapted for one-pass protocols. The analysis provided here includes the description of the required modification to the underlying security model. Finally, a complete security argument meeting these altered conditions is presented as evidence supporting the security of the one-pass scheme. Lastly, validation and reusing short lived key pairs are related to efficiency, which is a major objective in practice. The thesis considers the formal implication of omitting validation steps and reusing short lived key pairs. The conclusions reached support the generally accepted cryptographic conventions that incoming messages should not be blindly trusted and extra care should be taken when key pairs are reused.