Wireless Video Broadcast On Soft Wavelet Domain

Video multicast is becoming more and more popular in wireless multimediaapplications, in whichonemajor challenge is to offer heterogeneous users with agraceful degradation against varying packet loss ratios and channel noise.Considering that the traditional wireless video multicast cannot overcome cliffeffect and offer a graceful degradation to users, two new multicast schemes beproposed in this paper.Firstly, a multi-scale compressed sensingbased wireless video multicastscheme is proposed, abbreviated as MS-cast. The encoder of MS-castdecomposes each video framethrougha discrete wavelet transform (DWT) andexplores an optimizedCS-rate to sample/measure each DWT level. The CSmeasurements are then packed in sucha way that all packets are made as equallyimportant as possible, while each packet includes different percentages ofdifferent DWT levels. We demonstrate that, because of larger percentages ofmore important DWT levels in each packet, packet-loss producesa reducedinfluence on the reconstruction quality. Experimental resultsshow that ourMS-cast preserves the property of graceful degradation for heterogeneous usersand outperforms the state-of-the-art SoftCast by1-5dB in PSNR at high packetloss ratios (over the same noisy channel).Secondly, a wavelet-based QuadTree multicast schemeis proposed. Weeliminate the intra-frame correlation througha discrete wavelet transform (DWT),and reorganize the DWT coefficients by using QuadTree.This scheme includesthree important components: DWT, QuadTree, power allocation andwhitening.Experimental results also showthat the reconstruction quality of ourwavelet-based QuadTree multicast scheme outperforms state-of-the-art SoftCastby1dB in PSNR at all packet loss ratios.