Rate-Distortion-Authentication optimized streaming with Generalized Butterfly Graph authentication

摘要

This paper shows how Rate-Distortion-Authentication (R-D-A) optimized streaming may be performed with the Generalized Butterfly Graph (GBG) stream authentication method. R-D-A streaming is designed to compute an optimized transmission policy by accounting for both coding and authentication dependencies, and GBG is designed to protect the authenticity of a media stream. The GBG graph is better suited for streaming than the original Butterfly graph since it is highly flexible and can support an arbitrary number of packets and arbitrary overhead. However, the dependency chains within the GBG graph are much longer and are tangled with each other — making it very difficult to quantify the dependencies necessary to perform R-D-A streaming. We propose a method to estimate the authentication dependencies, which is then used by the R-D-A technique to compute the optimized transmission policy.