Optimized protection of streaming media authenticity

Abstract

This thesis studies the problem of providing quality-optimized authentication for multimedia content delivered over public and lossy network, which has two aspects: authenticity and quality. The former is to assure that any unauthorized alteration should be detected by a receiver, while the latter is to optimize the quality of the authenticated media. Towards these goals, we first propose a Butterfly Authentication, a graph-based authentication method achieving high robustness against packet loss. We then extend it to Generalized Butterfly Graph (GBG) Authentication to allow a flexible balance between overhead and robustness. In addition, considering that media packets are of unequal importance, we propose a Content-aware Optimized media stream authentication method to compute the best authentication graph achieving optimized media quality. Finally, we propose a Rate-Distortion-Authentication (R-D-A) Optimized streaming method, to compute the packet transmission schedule achieving optimized media quality, based on both packet importance and the authentication dependency.