Pseudo Analog And Hybrid Digital Analog Wireless Video Transmission

Given the prevalence of mobile devices and the development of wireless tech-nologies,wireless video applications are more and more popular.However,as one of the key enabling technologies,wireless video transmission is facing many challenges,which lie mainly in the optimality problem and adaptation ability of wireless video communication systems.Video source is of huge data volume and is highly correlated,under constraint of limited channel bandwidth and transmission power budget,mini-mizing the distortion of received video is challenge for optimality.Due to the mobility,multipath effect and so on,noise level of wireless channel is time-varying;the shared nature of wireless frequency band among multiple devices will result in the fluctua-tion of channel bandwidth over time.Adaptation to variation of channel condition is also facing big challenge.Under fixed channel condition,optimality of conventional digital video transmission method can be theoretically guaranteed by Shannon's the-ory.But in wireless channel,digital system cannot adapt well to time-varying channel condition.Recently,pseudo-analog transmission system is proposed to well solve the channel noise adaptation problem.Inspired by this,hybrid digital analog solutions for video transmission over wireless network are proposed.In this paper,we firstly consider the adaptation ability of wireless video trans-mission and proposed a progressive pseudo-analog transmission scheme.Inspired by pseudo-analog transmission of video,this method focuses on solving the problem that pseudo-analog solution is fragile to channel bandwidth fluctuation.We drop fixed band-width assumption,and everytime it have opportunity to send a packet,the proposed progressive solution tries to maximize the distortion reduction of reconstructed video,by means of optimally scheduling DCT coefficients chunks of source video and per-forming power allocation for each packet.When the wireless channel bandwidth is un-known or hard to estimate,the proposed solution can well adapt to not only the channel noise variation but also the bandwidth variation.Via comprehensive expreiments,we verify that the proposed progressive pseudo-analog transmission solution shows well scalability over both bandwidth and noise variation and outperforms the state-of-the-art digital,pseudo-analog methods.Then,we challenge the optimality of existing solutions.Given the fact that human visual system is more sensitive to structure information,we argue that current solutions,which aim to minimizing the mean squared error of reconstructed video,are harmful for structure information when channel bandwidth is not sufficient.Therefore,we pro-pose structure-preserving hybrid digital analog transmission solution named SharpCast.SharpCast decomposes video into structure and content.For structure-preserving pur-pose,structure information is protected with a robust digital transmission scheme;in order for channel noise adaptation and transmission efficiency,a hybrid digital analog scheme is designed for the content part.Experiment results show that SharpCast out-performs the state-of-the-art digital,pseudo-analog and hybrid digital analog solutions.At last,current research efforts mainly focus on single view 2D video source,we consider the more challenging low delay wireless transmission of two-view stereo video.Taking the binocular suppression effect of human visual system on stereo visual percep-tion into consideration,in this paper,we propose a hybrid digital analog scheme,called Swift,for low delay transmission of stereo video.In Swift,a zigzag hybrid digital analog coding structure is invented,which can explore inter-view correlation and intra-view correlation for efficiency while remaining low encoding delay property.Besides,selectively referring to the side information is proposed to further improve decorrela-tion efficiency.Via objective visual quality measurement and subjective visual quality assessment,we validate superiority of the proposed solution.