Abstract—Exchanging cryptographic keys to encrypt the media stream in the Session Initiation Protocol (SIP) has proven difficult. The challenge is to effectively exchange keys while preserving the features of the protocol (e.g., forking, re-targeting, request recursion, etc.), minimizing key exposure to unintended parties, eliminating voice clipping, maintaining end-to-end key privacy, interfacing with PSTN, etc. In this paper, we survey three key management protocols — SDES, ZRTP and DTLS-SRTP — that have been proposed for media keying, and evaluate them for use with SIP. To aid in the evaluation, we first extract (and justify) a core feature set from SIP. We then survey each key management protocol in detail and proceed to analyze the cores of the three protocols against this feature set to annotate their weaknesses and strengths.
Vijay K. Gurbani, Vladimir Kolesnikov