Model-based delay-distortion optimization for video streaming using packet interleaving

Bursty channel losses have been shown generally to produce larger total mean-square error distortion in streaming video than an equivalent number of isolated losses. The paper proposes a simple packet interleaving scheme to combat the effect of bursty losses by dispersing them. The optimal interleaver for minimizing the expected total distortion of the decoded video, subject to a delay constraint, is determined using a model that accurately predicts the expected distortion for different packet loss patterns. Compared to other forms of error-resilience, packet interleaving has the advantages of (1) simplicity and (2) not requiring any extra bitrate. For a simple burst loss channel, where each loss event has 100 ms duration, packet interleaving can provide between 0.24 dB and 0.87 dB gain over conventional non-interleaved streaming, at the expense of a delay increase of 400 ms.