Joint Source-Channel Adaptation for Perceptually Optimized Scalable Video Transmission

Providing perceptual quality guarantees for video transmission over wireless channels is important for the end-user experience. The paper proposes an algorithm for perceptual quality optimization of scalable H.264 video with temporal and quality scalability. The algorithm supports adaptive unequal error protection so that different video layers are protected according to their relevance to video quality. For each video layer, a target packet error rate (PER) is selected such that a perceptual quality guarantee, measured using the multi-scale structural similarity (MS-SSIM) index, is satisfied. Given the target PER per video layer, the algorithm selects the modulation and coding scheme and the number of temporal and quality layers to transmit adaptively based on the channel state information (CSI) and source rates per video layer. Results show that the algorithm provides immunity against short term channel fluctuations, balances reliability and perceptual quality, reduces playback buffer starvation probability, and provides a convenient buffer management policy.