Profile-Based Scalable Video Adaptation Employing MD-FEC Interleaving over Loss-Burst Channels

Loss-burst channels impose a grand challenge on delivery of scalable video streams with assured perceptual quality. Previous works seek for distortion optimization through joint adaptation of video layers and unequal error protection(UEP) at the application layer. However, such optimization usually incurs high computational cost and is hard to accurately estimate perceptual quality. In this paper, we propose an application-layer profile-based adaptation framework. We first propose a profile-based distortion model, which specifies a spectrum of target perceptual qualities along with their protection requirements. During transmission, we then employ a simple interleaved MD-FEC adaptation algorithm which finds the best target quality of which required protection can be supported subject to the current available bandwidth and loss condition. Through simulation, we confirm that our scheme can agilely perform the adaptation for scalable videos with profile specific perceptual quality guarantee even in highly loss-burst channels, compared with the ones applying fixed-UEP/interleaving or no protection.