Joint source-channel coding of scalable video on a Rayleigh fading channel

We present a joint source-channel coding scheme developed for transmitting scalable video on a Rayleigh fading channel. Our scheme consists of a motion compensated temporal subband decomposition applied on groups of frames, followed by a spatial wavelet transform of the temporal subbands and a vector quantization. The vector quantization method is based on a linear labelling by "maximum component diversity" lattice constellations which minimizes not only the channel distortion but at the same time the distortion of our video source. These lattice constellations, which are multidimensional rotations generated from totally real cyclotomic fields, increase the diversity order on a Rayleigh fading channel. In addition, a rate-compatible punctured convolutional code in combination with a perfect interleaving is applied, providing even higher diversity.