GOP-based unequal error protection for scalable video over packet erasure channel

Scalable video coding has been desired for many years to meet the requirement of heterogeneous networks and different end-users. It intends to encode the signal only once, but enables decoding from partial streams depending on the specific rate and resolution. The joint video team has recently developed a new scalable video coding standard, known as SVC, which is the scalable extension of AVC. A full scalability is provided by SVC including temporal scalability, spatial scalability and SNR scalability. In each scalability dimension, the compressed bitstream can be segmented into layers with different contributions to the overall video quality. This paper addresses the problem of unequal error protection for the scalable video over packet-erasure channel considering both the channel condition and the video characteristics. In addition, a local search algorithm is proposed to quickly allocate unequal amount of channel rate to protect different layers to obtain acceptable quality video with smooth degradation under different transmission error conditions. The advantage of the proposed algorithms can be demonstrated by the theoretical analysis and experimental results.