An extended packetization-aware mapping algorithm for scalable video coding in finite-length fountain codes

We propose an extended packetization-aware mapping algorithm based on fountain codes to enhance video stream performance which is vulnerable to packet losses. By properly utilizing the proposed algorithm in finite-length cases, the edges connecting to a source symbol are scattered over multiple encoding packets so that the decoding probabilities are increased. Furthermore, an improved degree distribution is designed to obtain better decoding probabilities. Numerical results, it is confirmed that the proposed algorithm in finite-length cases can augment decoding probabilities and the improved degree distribution can increase peak signal-to-noise ratio (PSNR) of scalable video coding (SVC) in a hostile communication environment.